APOLOGIES TO ALL: PIEZO TWEETERS DO SOUND LIKE SHIT!!!! [Archive]

Chuck[_12_]

February 15th 17, 02:40 PM

On Tue, 14 Feb 2017 22:31:54 -0700, Paul > wrote:

>
> GO AHEAD AND TAKE TURNS SAYING: "I TOLD YOU SO, DUMB MOTHER-****ER!"
>
> **** YOU ALL FOR BEING RIGHT! HAHA!
>
> :)
>
> IN MY DEFENSE, I'M NOT STUPID....JUST CHEAP, FRUGAL, AND STUBBORN
>AS HELL! THERE IS A DIFFERENCE!
>
> HAHAHAAA!!! :)
>
> This was apparent after comparing A/B NON-CROSSOVERED piezos with
>one of my Yamaha HS80Ms.
>
> The incriminating test song was Pink Floyd's "Time"! As soon as
>the vocals come in, it's SCREECH CITY with the piezo! BOY THAT SOUNDS
>BAD! I mean, it depends on where you are listening wrt the horn, but
>if you are right in front of it....UUUUGGGHH!!
>
> In contrast, you can listen anywhere in front of the Yammies, and
>although they may sound bright, and have tons of "presence", they never
>get HARSH like the piezos!
>
> So then I experimented with adding a crossover, and settled on
>series 2uF non-polarized cap, and shunt 22 Ohm resistor before the
>piezos in parallel (a series 0.1uF after the shunt resistor, for -6dB,
>was too much, and almost killed off the signal completely).
>
> The crossover makes it more tolerable, but I can't say it sounds
>good delivering high fidelity music. For some reason, just singing
>through it with an SM58, makes the harshness less noticeable, so
>it could still be useful for that usage.
>
> I'll probably sell this for a few bucks to a poor garage band,
>telling them exactly what I did.
>
> I DIDN'T WASTE MY TIME AND ENERGY WITH THIS EXPERIMENT: I LEARNED
>QUITE A BIT, AND THAT MAKES TINKERING WORTH DOING!
>
> Thanks for the feedback, everyone....
>
> :)
>
>
>
Motorola sent a sample of these tweeters to a speaker company I worked
for in the 1970s. We tested it then smashed it with a hammer and
threw it in the trash. Dahlquist used one in a very good sounding
speaker but they crossed it over above 10 khz.

---
This email has been checked for viruses by Avast antivirus software.
https://www.avast.com/antivirus

Scott Dorsey

February 15th 17, 04:13 PM

In article >, Paul > wrote:
>
> In contrast, you can listen anywhere in front of the Yammies, and
>although they may sound bright, and have tons of "presence", they never
>get HARSH like the piezos!

Okay, you saw that 4kc peak on the piezos. Now, assuming you're using a
12dB/octave filter, if you crossed over at 8kc then the peak would only be
12dB down. You'd probably want to cross over at 16kc in order to really
control the problem, or use a sharper filter. Which kind of makes it useless.

You'd think maybe you could use a Zobel network like you would with an
electrodynamic tweeter to control the peak, but really nobody has ever been
able to make that work well. It might not be minimum phase.

It MIGHT be possible to use an acoustic network in order to deal with the
problem, but it's hard to do that and not screw up the pattern.

In the 1980s some grad student built a PA speaker system using a horizontal
array of piezo tweeters and phase-shift networks, which somehow wound up in
the EE department auditorium at gatech, probably because nobody else wanted
them. They had oversized Motorola drivers with the worst of the ugliness
around 1kc, and they were crossed over around 5kc using a conventional
midrange driver. Even though the plot wasn't so horrible, there was still
severe harshness due to the nonlinearities.

Now... Jon Dhalquist with the DQ-10 actually did use a piezo tweeter and
did actually get some benefit from it. But he was crossing them over at
18 KHz or so and using them only as a supertweeter in order to add a little
more air.
--scott

--
"C'est un Nagra. C'est suisse, et tres, tres precis."

Richard Kuschel

February 15th 17, 04:19 PM

On Wednesday, February 15, 2017 at 9:13:13 AM UTC-7, Scott Dorsey wrote:
Snip

> Now... Jon Dhalquist with the DQ-10 actually did use a piezo tweeter and
> did actually get some benefit from it. But he was crossing them over at
> 18 KHz or so and using them only as a supertweeter in order to add a little
> more air.
> --scott
>
> --
> "C'est un Nagra. C'est suisse, et tres, tres precis."

A friend on mine had a pair of those Dahlquists and they were a good sounding system.
I knew that they had piezo tweeters, but wasn't awate that they were crossed that high.
A piezo will respond to a 40kHz signal but all that is going to do is annoy the dog.

Paul[_13_]

February 15th 17, 05:36 PM

On 2/15/2017 3:16 AM, geoff wrote:
> On 15/02/2017 6:31 PM, Paul wrote:
>
>>
>> I DIDN'T WASTE MY TIME AND ENERGY WITH THIS EXPERIMENT: I LEARNED
>> QUITE A BIT, AND THAT MAKES TINKERING WORTH DOING!
>>
>> Thanks for the feedback, everyone....
>>
>> :)
>
> So a positive outcome after all that !
>
> ;-)
>

Oh, completely!

I didn't just learn new things: Many electrical engineering
concepts are the same, no matter what the frequency range. So this
was also a review for potential Radio Frequency job interviews that
might be coming up.

:)

Also, this stuff is fascinating, and FUN!

And you are never wasting your time if you're having fun!

BWAAAHHAHHAAA!!

Paul[_13_]

February 15th 17, 05:57 PM

On 2/15/2017 9:13 AM, Scott Dorsey wrote:
> In article >, Paul > wrote:
>>
>> In contrast, you can listen anywhere in front of the Yammies, and
>> although they may sound bright, and have tons of "presence", they never
>> get HARSH like the piezos!
>
> Okay, you saw that 4kc peak on the piezos. Now, assuming you're using a
> 12dB/octave filter, if you crossed over at 8kc then the peak would only be
> 12dB down. You'd probably want to cross over at 16kc in order to really
> control the problem, or use a sharper filter. Which kind of makes it useless.
>
> You'd think maybe you could use a Zobel network like you would with an
> electrodynamic tweeter to control the peak, but really nobody has ever been
> able to make that work well. It might not be minimum phase.

http://www.wavecor.com/html/zobel_networks.html

So when the cap becomes low impedance at higher frequencies,
it essentially adds a shunt resistor to a dynamic speaker's voice coil
inductance, attenuating the higher frequencies I assume.

But maybe this wouldn't work on piezos since they look capacitive?

>
> It MIGHT be possible to use an acoustic network in order to deal with the
> problem, but it's hard to do that and not screw up the pattern.
>
> In the 1980s some grad student built a PA speaker system using a horizontal
> array of piezo tweeters and phase-shift networks, which somehow wound up in
> the EE department auditorium at gatech, probably because nobody else wanted
> them. They had oversized Motorola drivers with the worst of the ugliness
> around 1kc, and they were crossed over around 5kc using a conventional
> midrange driver. Even though the plot wasn't so horrible, there was still
> severe harshness due to the nonlinearities.
>
> Now... Jon Dhalquist with the DQ-10 actually did use a piezo tweeter and
> did actually get some benefit from it. But he was crossing them over at
> 18 KHz or so and using them only as a supertweeter in order to add a little
> more air.
> --scott
>

So it appears piezos are only acceptable in the high fidelity
world from 10kHz and above?

Non-linearities implies clipping and odd order harmonic generation.
What is it about the physics of the piezos that would cause such
distortion below 10kHz?

It appears piezo tweeters are everywhere! Obviously because they
are cheaper to manufacture. MP3s and earbuds are also widely used
by the public at large, but MP3s never made the high end unlistenable!

PStamler

February 15th 17, 07:38 PM

Capacitors, unfortunately, don't become low impedance at high frequencies; since they're made of rolled-up layers of foil and dielectric, they have an inductive component, and at high frequencies this becomes the most important; real-world capacitors show an impedance that falls with increasing frequency up to a point, then bottoms out, and rises as the frequency increases..

Peace,
Paul

PStamler

February 15th 17, 07:42 PM

Paul[_13_]

February 15th 17, 11:13 PM

On 2/15/2017 12:42 PM, PStamler wrote:
> Capacitors, unfortunately, don't become low impedance at high frequencies; since they're made of rolled-up layers of foil and dielectric, they have an inductive component, and at high frequencies this becomes the most important; real-world capacitors show an impedance that falls with increasing frequency up to a point, then bottoms out, and rises as the frequency increases.
>

Well, of course I'm aware that all capacitors have parasitic
inductances (and Equivalent Series Resistance, or ESR), which is the
reason why we keep leads lengths short in the radio frequency world,
when not using chip capacitors:

http://sound.whsites.net/articles/capacitors.htm#s30

But in the audio frequency world, where we are only concerned about
up to 22kHz or so, I we shouldn't see the impedance start rising again.

Let's assume from the above link that we have about 5nH of
parasitic inductance mainly due to lead length:

XL=2*pi*f*L

And using f=22kHz, and L=5nH, we still only get about 69 milli-Ohms!

> Glad you had fun. Watch out, though -- speaker building is addictive.
>

No worries on that....as you can see, I'm shopping for
an already built PA speaker now.

Sometimes it's best to leave things to the professionals!

:)

> Peace,
> The Other Paul
>

Phil Allison[_4_]

February 16th 17, 02:58 AM

PStamler wrote:
>
> Capacitors, unfortunately, don't become low impedance at high
> frequencies;
>

** Yes they do.

> since they're made of rolled-up layers of foil and dielectric,
> they have an inductive component,

** That is not true.

The foil is connected along the whole length of one side, which cancels inductance completely.

> real-world capacitors show an impedance that falls with increasing
> frequency up to a point, then bottoms out, and rises as the
> frequency increases.

** Real world capacitors have no more inductance than a piece of wire the same length. This is so small, it never matters in audio circuits or even SMPS and the like.

FYI, a typical value is 15nH for leaded types and about 2nH for SMD types.

You have just trotted out a stupid myth that I really hoped had gone away.

..... Phil

PStamler

February 16th 17, 04:57 AM

At the risk of attracting one of Mr. Allison's personal attacks, I have measured the rise in capacitors' impedance at high frequencies -- in some cases they switch from being capacitative to inductive well within the audio band.

Peace,
Paul

PStamler

February 16th 17, 05:13 AM

At the risk of incurring a personal attack, I wish to report that I have tested common commercially-available capacitors, looking for the resonant frequency (frequency at which the device's impedance bottoms out; the cap stops behaving like a capacitor above this frequency and starts behaving like an inductor). The lowest resonant frequency I found was 5.3kHz, well within the audio range.

Peace,
The Other Paul

Paul[_13_]

February 16th 17, 07:07 AM

On 2/15/2017 10:13 PM, PStamler wrote:
> At the risk of incurring a personal attack, I wish to report that I have tested common commercially-available capacitors, looking for the resonant frequency (frequency at which the device's impedance bottoms out; the cap stops behaving like a capacitor above this frequency and starts behaving like an inductor). The lowest resonant frequency I found was 5.3kHz, well within the audio range.
>

What was the capacitor value for that measurement?

geoff

February 16th 17, 10:04 AM

On 16/02/2017 8:07 PM, Paul wrote:
> On 2/15/2017 10:13 PM, PStamler wrote:
>> At the risk of incurring a personal attack, I wish to report that I
>> have tested common commercially-available capacitors, looking for the
>> resonant frequency (frequency at which the device's impedance bottoms
>> out; the cap stops behaving like a capacitor above this frequency and
>> starts behaving like an inductor). The lowest resonant frequency I
>> found was 5.3kHz, well within the audio range.
>>
>
> What was the capacitor value for that measurement?
>
>

..... and what type of cap ?

geoff

Scott Dorsey

February 16th 17, 03:20 PM

Phil Allison > wrote:
>PStamler wrote:
>>
>> Capacitors, unfortunately, don't become low impedance at high
>> frequencies;
>>
>
>** Yes they do.

This depends a lot of what kind of capacitor it is. If you look at an
electrolytic, they tend to become inductors down in the 100s of KHz range.

On the other hand if you look at stacked film capacitors, they tend not
to become inductors at all, just resistors.

I have come to love stacked films.
--scott
--
"C'est un Nagra. C'est suisse, et tres, tres precis."

PStamler

February 16th 17, 07:21 PM

On Thursday, February 16, 2017 at 1:07:31 AM UTC-6, Paul wrote:
> On 2/15/2017 10:13 PM, PStamler wrote:
> > At the risk of incurring a personal attack, I wish to report that I have tested common commercially-available capacitors, looking for the resonant frequency (frequency at which the device's impedance bottoms out; the cap stops behaving like a capacitor above this frequency and starts behaving like an inductor). The lowest resonant frequency I found was 5.3kHz, well within the audio range.
> >
>
> What was the capacitor value for that measurement?

It was a 3,300µF/50V Panasonic Series NHG electrolytic.

geoff

February 16th 17, 07:27 PM

On 17/02/2017 8:21 AM, PStamler wrote:
> On Thursday, February 16, 2017 at 1:07:31 AM UTC-6, Paul wrote:
>> On 2/15/2017 10:13 PM, PStamler wrote:
>>> At the risk of incurring a personal attack, I wish to report that I have tested common commercially-available capacitors, looking for the resonant frequency (frequency at which the device's impedance bottoms out; the cap stops behaving like a capacitor above this frequency and starts behaving like an inductor). The lowest resonant frequency I found was 5.3kHz, well within the audio range.
>>>
>>
>> What was the capacitor value for that measurement?
>
> It was a 3,300µF/50V Panasonic Series NHG electrolytic.
>

Maybe a little 'sweeping' to attribute the characteristics of that
capacitor to all capacitors . Especially given the typical application
of that type and value of cap.

geoff

PStamler

February 16th 17, 07:59 PM

This was the worst case, but several electrolytic caps had resonances within the audible range.

Peace,
The Other Paul

Don Pearce[_3_]

February 16th 17, 08:06 PM

On Thu, 16 Feb 2017 11:59:47 -0800 (PST), PStamler
> wrote:

>This was the worst case, but several electrolytic caps had resonances within the audible range.
>
>Peace,
>The Other Paul

And the problem here is that the better the cap the sharper the
resonance. Get a nice rubbish cap with a huge ESR and the effect is
much diminished.

d

---
This email has been checked for viruses by Avast antivirus software.
https://www.avast.com/antivirus

[email protected]

February 16th 17, 08:24 PM

so if your 3,300 uF cap had a series resonance at 5.3 kHz, that implies an
inductance of about 0.25 uH which has a Z of about 9 mOhms.

The ESR will easily swamp that and Z is so low so as not to have any practical impact at any audiofreq.

m

Paul[_13_]

February 16th 17, 08:43 PM

On 2/16/2017 12:21 PM, PStamler wrote:
> On Thursday, February 16, 2017 at 1:07:31 AM UTC-6, Paul wrote:
>> On 2/15/2017 10:13 PM, PStamler wrote:
>>> At the risk of incurring a personal attack, I wish to report that I have tested common commercially-available capacitors, looking for the resonant frequency (frequency at which the device's impedance bottoms out; the cap stops behaving like a capacitor above this frequency and starts behaving like an inductor). The lowest resonant frequency I found was 5.3kHz, well within the audio range.
>>>
>>
>> What was the capacitor value for that measurement?
>
> It was a 3,300µF/50V Panasonic Series NHG electrolytic.
>

http://www.murata.com/~/media/webrenewal/products/emc/emifil/knowhow/12to14.ashx

So resonant freq f=1/(2*Pi*(L*C)**0.5)

So L=273nH.

So you had 273nH of parasitic/lead inductance? BULL****.

Also, where in the **** would you need such a large cap
in a speaker crossover?

IF DON'T WANT PERSONAL ATTACKS, DON'T MAKE STUPID **** UP!!!

John Williamson

February 16th 17, 09:11 PM

On 16/02/2017 21:43, Paul wrote:

> Also, where in the **** would you need such a large cap
> in a speaker crossover?
>
In the PSU for an active speaker? He didn't say he used it in a
crossover, just that he tested it and found the resonant frequency he
mentioned.

Incidentally, the tan δ for that capacitor is given in the maker's
datasheets as as 0.16 at 120 Hz, so the ESR isn't fanstasic. There is
no mention of self inductance values, but as it's an aluminium film type
capacitor, the self inductance figures won't be all that low.
--
Tciao for Now!

John.

JackA

February 16th 17, 09:17 PM

On Thursday, February 16, 2017 at 4:11:19 PM UTC-5, John Williamson wrote:
> On 16/02/2017 21:43, Paul wrote:
>
> > Also, where in the **** would you need such a large cap
> > in a speaker crossover?
> >
> In the PSU for an active speaker? He didn't say he used it in a
> crossover, just that he tested it and found the resonant frequency he
> mentioned.
>
> Incidentally, the tan δ for that capacitor is given in the maker's
> datasheets as as 0.16 at 120 Hz, so the ESR isn't fanstasic. There is
> no mention of self inductance values, but as it's an aluminium film type
> capacitor, the self inductance figures won't be all that low.
> --
> Tciao for Now!
>
> John.

Wonder if they use anodized aluminum to rid of any insulating film?

Jack

Scott Dorsey

February 16th 17, 09:17 PM

In article >,
> wrote:
>so if your 3,300 uF cap had a series resonance at 5.3 kHz, that implies an
>inductance of about 0.25 uH which has a Z of about 9 mOhms.
>
>The ESR will easily swamp that and Z is so low so as not to have any practical impact at any audiofreq.

Right. Much more of a worry is nonlinearity when the ripple voltages get
higher.

However, the series resonances in the ultrasonic can be much more dramatic
and those can lead to stability issues if you don't do the math right.
--scott

--
"C'est un Nagra. C'est suisse, et tres, tres precis."

Scott Dorsey

February 16th 17, 11:48 PM

John Williamson > wrote:
>
>Incidentally, the tan δ for that capacitor is given in the maker's
>datasheets as as 0.16 at 120 Hz, so the ESR isn't fanstasic. There is
>no mention of self inductance values, but as it's an aluminium film type
>capacitor, the self inductance figures won't be all that low.

270 nF does not seem out of the question to me, but it's kind of hard to
look at this as a lumped-sum problem since you're dealing with both
distributed capacitance and inductance across all the winds.

I haven't personally seen resonances that low in capacitors, but I have seen
some only a couple octaves higher. Thankfully not huge ones.
--scott

--
"C'est un Nagra. C'est suisse, et tres, tres precis."

Trevor

February 17th 17, 12:38 AM

On 16/02/2017 6:38 AM, PStamler wrote:
> Capacitors, unfortunately, don't become low impedance at high
> frequencies; since they're made of rolled-up layers of foil and
> dielectric,

Not all are made like that.

> they have an inductive component, and at high frequencies
> this becomes the most important; real-world capacitors show an
> impedance that falls with increasing frequency up to a point, then
> bottoms out, and rises as the frequency increases.

Which is rarely a problem at audio frequencies though.

Trevor.

Trevor

February 17th 17, 12:43 AM

On 16/02/2017 3:57 PM, PStamler wrote:
> At the risk of attracting one of Mr. Allison's personal attacks,I
> have measured the rise in capacitors' impedance at high frequencies
> -- in some cases they switch from being capacitative to inductive
> well within the audio band.

If you often use power supply filter caps etc. for audio coupling
purposes, that's definitely something you'd would want to take into
account I guess. :-) There is a reason why one selects components for
purpose of course.

Trevor.

Trevor

February 17th 17, 12:49 AM

On 17/02/2017 6:21 AM, PStamler wrote:
> On Thursday, February 16, 2017 at 1:07:31 AM UTC-6, Paul wrote:
>> On 2/15/2017 10:13 PM, PStamler wrote:
>>> At the risk of incurring a personal attack, I wish to report that
>>> I have tested common commercially-available capacitors, looking
>>> for the resonant frequency (frequency at which the device's
>>> impedance bottoms out; the cap stops behaving like a capacitor
>>> above this frequency and starts behaving like an inductor). The
>>> lowest resonant frequency I found was 5.3kHz, well within the
>>> audio range.
>>>
>>
>> What was the capacitor value for that measurement?
>
> It was a 3,300µF/50V Panasonic Series NHG electrolytic.

Most people use those for power supply filtering etc. rather than audio.
What was their inductance at 50-120Hz?
Try using components for their intended purpose perhaps?

Trevor.

JackA

February 17th 17, 12:51 AM

On Thursday, February 16, 2017 at 3:43:15 PM UTC-5, Paul wrote:
> On 2/16/2017 12:21 PM, PStamler wrote:
> > On Thursday, February 16, 2017 at 1:07:31 AM UTC-6, Paul wrote:
> >> On 2/15/2017 10:13 PM, PStamler wrote:
> >>> At the risk of incurring a personal attack, I wish to report that I have tested common commercially-available capacitors, looking for the resonant frequency (frequency at which the device's impedance bottoms out; the cap stops behaving like a capacitor above this frequency and starts behaving like an inductor). The lowest resonant frequency I found was 5.3kHz, well within the audio range.
> >>>
> >>
> >> What was the capacitor value for that measurement?
> >
> > It was a 3,300µF/50V Panasonic Series NHG electrolytic.
> >
>
>
>
> http://www.murata.com/~/media/webrenewal/products/emc/emifil/knowhow/12to14.ashx
>
> So resonant freq f=1/(2*Pi*(L*C)**0.5)
>
> So L=273nH.
>
> So you had 273nH of parasitic/lead inductance? BULL****.
>
> Also, where in the **** would you need such a large cap
> in a speaker crossover?
>
> IF DON'T WANT PERSONAL ATTACKS, DON'T MAKE STUPID **** UP!!!

You know you stuff! Welcome to see someone with your knowledge here!

Jack

david gourley[_2_]

February 17th 17, 01:05 AM

Trevor > :

> On 17/02/2017 6:21 AM, PStamler wrote:
>> On Thursday, February 16, 2017 at 1:07:31 AM UTC-6, Paul wrote:
>>> On 2/15/2017 10:13 PM, PStamler wrote:
>>>> At the risk of incurring a personal attack, I wish to report that
>>>> I have tested common commercially-available capacitors, looking
>>>> for the resonant frequency (frequency at which the device's
>>>> impedance bottoms out; the cap stops behaving like a capacitor
>>>> above this frequency and starts behaving like an inductor). The
>>>> lowest resonant frequency I found was 5.3kHz, well within the
>>>> audio range.
>>>>
>>>
>>> What was the capacitor value for that measurement?
>>
>> It was a 3,300µF/50V Panasonic Series NHG electrolytic.
>
> Most people use those for power supply filtering etc. rather than audio.
> What was their inductance at 50-120Hz?
> Try using components for their intended purpose perhaps?
>
> Trevor.
>
>

My Dynaco ST120 channels use that value for output coupling to the speaker.

david

---
This email has been checked for viruses by Avast antivirus software.
https://www.avast.com/antivirus

Scott Dorsey

February 17th 17, 01:41 AM

In article >, Trevor > wrote:
>On 16/02/2017 3:57 PM, PStamler wrote:
>> At the risk of attracting one of Mr. Allison's personal attacks,I
>> have measured the rise in capacitors' impedance at high frequencies
>> -- in some cases they switch from being capacitative to inductive
>> well within the audio band.
>
>If you often use power supply filter caps etc. for audio coupling
>purposes, that's definitely something you'd would want to take into
>account I guess. :-) There is a reason why one selects components for
>purpose of course.

Sadly, that was the technology of the 1970s. People were designing with
transistors but they were still thinking about tubes in their heads, so
everything was capacitively coupled and electrolytics were needed in order
to deal with the high values required due to the low impedances.

I was at a mastering facility a few years back with some audiophile label
guys who were looking at having some LPs cut. They asked the mastering
engineer if there were any electrolytic capacitors in the signal path of
the Neumann lathe amplifier and he about spit himself. "Millions of them!"
he said. "Millions!"

And so, because we live with a lot of older equipment designed in this
regime, we have to deal with it and we have to find capacitors appropriate
for the application.
--scott
--
"C'est un Nagra. C'est suisse, et tres, tres precis."

Scott Dorsey

February 17th 17, 01:42 AM

david gourley > wrote:
>
>My Dynaco ST120 channels use that value for output coupling to the speaker.

Is mentioning an ST120 like mentioning Hitler? Is this thread closed now?
--scott

--
"C'est un Nagra. C'est suisse, et tres, tres precis."

Paul[_13_]

February 17th 17, 01:44 AM

On 2/16/2017 5:49 PM, Trevor wrote:
> On 17/02/2017 6:21 AM, PStamler wrote:
>> On Thursday, February 16, 2017 at 1:07:31 AM UTC-6, Paul wrote:
>>> On 2/15/2017 10:13 PM, PStamler wrote:
>>>> At the risk of incurring a personal attack, I wish to report that
>>>> I have tested common commercially-available capacitors, looking
>>>> for the resonant frequency (frequency at which the device's
>>>> impedance bottoms out; the cap stops behaving like a capacitor
>>>> above this frequency and starts behaving like an inductor). The
>>>> lowest resonant frequency I found was 5.3kHz, well within the
>>>> audio range.
>>>>
>>>
>>> What was the capacitor value for that measurement?
>>
>> It was a 3,300µF/50V Panasonic Series NHG electrolytic.
>
> Most people use those for power supply filtering etc. rather than audio.
> What was their inductance at 50-120Hz?
> Try using components for their intended purpose perhaps?
>
> Trevor.
>

+1

geoff

February 17th 17, 01:46 AM

On 17/02/2017 2:05 p.m., david gourley wrote:
> My Dynaco ST120 channels use that value for output coupling to the speaker.
>
> david
>
> ---
> This email has been checked for viruses by Avast antivirus software.
> https://www.avast.com/antivirus
>
Output coupling caps on a power amp ? I haven't seen those for decades !

geoff

Paul[_13_]

February 17th 17, 01:53 AM

On 2/16/2017 4:48 PM, Scott Dorsey wrote:
> John Williamson > wrote:
>>
>> Incidentally, the tan δ for that capacitor is given in the maker's
>> datasheets as as 0.16 at 120 Hz, so the ESR isn't fanstasic. There is
>> no mention of self inductance values, but as it's an aluminium film type
>> capacitor, the self inductance figures won't be all that low.
>
> 270 nF does not seem out of the question to me,

So let's assume the leaded capacitor has about 15nH of self
inductance.

270-15= 255 nH.

And let's guesstimate 6nH of inductance per cm of lead length.

42.5 cm of added lead length???

THAT WOULD BE SLOPPY ENGINEERING!!!!

:/

http://sound.whsites.net/articles/capacitors.htm

> I haven't personally seen resonances that low in capacitors, but I have seen
> some only a couple octaves higher. Thankfully not huge ones.
> --scott
>

PStamler

February 17th 17, 02:23 AM

On Thursday, February 16, 2017 at 2:43:15 PM UTC-6, Paul wrote:
> On 2/16/2017 12:21 PM, PStamler wrote:
> > On Thursday, February 16, 2017 at 1:07:31 AM UTC-6, Paul wrote:
> >> On 2/15/2017 10:13 PM, PStamler wrote:
> >>> At the risk of incurring a personal attack, I wish to report that I have tested common commercially-available capacitors, looking for the resonant frequency (frequency at which the device's impedance bottoms out; the cap stops behaving like a capacitor above this frequency and starts behaving like an inductor). The lowest resonant frequency I found was 5.3kHz, well within the audio range.
> >>>
> >>
> >> What was the capacitor value for that measurement?
> >
> > It was a 3,300µF/50V Panasonic Series NHG electrolytic.
> >
>
>
>
> http://www.murata.com/~/media/webrenewal/products/emc/emifil/knowhow/12to14.ashx
>
> So resonant freq f=1/(2*Pi*(L*C)**0.5)
>
> So L=273nH.
>
> So you had 273nH of parasitic/lead inductance? BULL****.
>
> Also, where in the **** would you need such a large cap
> in a speaker crossover?

I never said you would. A 3,300µF cap would more likely be found in a power supply, or perhaps in series with Rin in a noninverting opamp circuit..

>> IF DON'T WANT PERSONAL ATTACKS, DON'T MAKE STUPID **** UP!!!

I am making nothing up; I'm simply reporting the result of aome tests I ran, no counter the assertion that inductance isn't an issue with electrolytic capacitors.

Peace,
The Other Paul

PStamler

February 17th 17, 02:27 AM

On Tuesday, February 14, 2017 at 11:32:09 PM UTC-6, Paul wrote:
> GO AHEAD AND TAKE TURNS SAYING: "I TOLD YOU SO, DUMB MOTHER-****ER!"
>
> **** YOU ALL FOR BEING RIGHT! HAHA!
>
> :)
>
> IN MY DEFENSE, I'M NOT STUPID....JUST CHEAP, FRUGAL, AND STUBBORN
> AS HELL! THERE IS A DIFFERENCE!
>
> HAHAHAAA!!! :)
>
> This was apparent after comparing A/B NON-CROSSOVERED piezos with
> one of my Yamaha HS80Ms.
>
> The incriminating test song was Pink Floyd's "Time"! As soon as
> the vocals come in, it's SCREECH CITY with the piezo! BOY THAT SOUNDS
> BAD! I mean, it depends on where you are listening wrt the horn, but
> if you are right in front of it....UUUUGGGHH!!
>
> In contrast, you can listen anywhere in front of the Yammies, and
> although they may sound bright, and have tons of "presence", they never
> get HARSH like the piezos!
>
> So then I experimented with adding a crossover, and settled on
> series 2uF non-polarized cap, and shunt 22 Ohm resistor before the
> piezos in parallel (a series 0.1uF after the shunt resistor, for -6dB,
> was too much, and almost killed off the signal completely).
>
> The crossover makes it more tolerable, but I can't say it sounds
> good delivering high fidelity music. For some reason, just singing
> through it with an SM58, makes the harshness less noticeable, so
> it could still be useful for that usage.
>
> I'll probably sell this for a few bucks to a poor garage band,
> telling them exactly what I did.
>
> I DIDN'T WASTE MY TIME AND ENERGY WITH THIS EXPERIMENT: I LEARNED
> QUITE A BIT, AND THAT MAKES TINKERING WORTH DOING!
>
> Thanks for the feedback, everyone....
>
> :)

david gourley[_2_]

February 17th 17, 02:33 AM

(Scott Dorsey) :

> david gourley > wrote:
>>
>>My Dynaco ST120 channels use that value for output coupling to the speaker.
>
> Is mentioning an ST120 like mentioning Hitler? Is this thread closed now?
> --scott
>

Wow, sorry didn't know it was THAT bad. I've used it for a guitar and bass
amp.

david

---
This email has been checked for viruses by Avast antivirus software.
https://www.avast.com/antivirus

david gourley[_2_]

February 17th 17, 02:34 AM

geoff > said...news:Q8Gdnbn5apB0yDvFnZ2dnUU7-
:

> On 17/02/2017 2:05 p.m., david gourley wrote:
>> My Dynaco ST120 channels use that value for output coupling to the
speaker.
>>
>> david
>>
>> ---
>> This email has been checked for viruses by Avast antivirus software.
>> https://www.avast.com/antivirus
>>
> Output coupling caps on a power amp ? I haven't seen those for decades
!
>
>
> geoff
>
>

It's decades old, too.

david

---
This email has been checked for viruses by Avast antivirus software.
https://www.avast.com/antivirus

Paul[_13_]

February 17th 17, 02:39 AM

On 2/16/2017 7:23 PM, PStamler wrote:
> On Thursday, February 16, 2017 at 2:43:15 PM UTC-6, Paul wrote:
>> On 2/16/2017 12:21 PM, PStamler wrote:
>>> On Thursday, February 16, 2017 at 1:07:31 AM UTC-6, Paul wrote:
>>>> On 2/15/2017 10:13 PM, PStamler wrote:
>>>>> At the risk of incurring a personal attack, I wish to report that I have tested common commercially-available capacitors, looking for the resonant frequency (frequency at which the device's impedance bottoms out; the cap stops behaving like a capacitor above this frequency and starts behaving like an inductor). The lowest resonant frequency I found was 5.3kHz, well within the audio range.
>>>>>
>>>>
>>>> What was the capacitor value for that measurement?
>>>
>>> It was a 3,300µF/50V Panasonic Series NHG electrolytic.
>>>
>>
>>
>>
>> http://www.murata.com/~/media/webrenewal/products/emc/emifil/knowhow/12to14.ashx
>>
>> So resonant freq f=1/(2*Pi*(L*C)**0.5)
>>
>> So L=273nH.
>>
>> So you had 273nH of parasitic/lead inductance? BULL****.
>>
>> Also, where in the **** would you need such a large cap
>> in a speaker crossover?
>
> I never said you would. A 3,300µF cap would more likely be found in a power supply, or perhaps in series with Rin in a noninverting opamp circuit.
>
>>> IF DON'T WANT PERSONAL ATTACKS, DON'T MAKE STUPID **** UP!!!
>
> I am making nothing up; I'm simply reporting the result of aome tests I ran, no counter the assertion that inductance isn't an issue with electrolytic capacitors.
>

But you still haven't addressed why you had such a high
parasitic/lead inductance.

Typical values are around 15nH for a leaded component....

????

PStamler

February 17th 17, 03:12 AM

On Thursday, February 16, 2017 at 8:39:44 PM UTC-6, Paul wrote:
> On 2/16/2017 7:23 PM, PStamler wrote:
> > On Thursday, February 16, 2017 at 2:43:15 PM UTC-6, Paul wrote:
> >> On 2/16/2017 12:21 PM, PStamler wrote:
> >>> On Thursday, February 16, 2017 at 1:07:31 AM UTC-6, Paul wrote:
> >>>> On 2/15/2017 10:13 PM, PStamler wrote:
> >>>>> At the risk of incurring a personal attack, I wish to report that I have tested common commercially-available capacitors, looking for the resonant frequency (frequency at which the device's impedance bottoms out; the cap stops behaving like a capacitor above this frequency and starts behaving like an inductor). The lowest resonant frequency I found was 5.3kHz, well within the audio range.
> >>>>>
> >>>>
> >>>> What was the capacitor value for that measurement?
> >>>
> >>> It was a 3,300µF/50V Panasonic Series NHG electrolytic.
> >>>
> >>
> >>
> >>
> >> http://www.murata.com/~/media/webrenewal/products/emc/emifil/knowhow/12to14.ashx
> >>
> >> So resonant freq f=1/(2*Pi*(L*C)**0.5)
> >>
> >> So L=273nH.
> >>
> >> So you had 273nH of parasitic/lead inductance? BULL****.
> >>
> >> Also, where in the **** would you need such a large cap
> >> in a speaker crossover?
> >
> > I never said you would. A 3,300µF cap would more likely be found in a power supply, or perhaps in series with Rin in a noninverting opamp circuit.
> >
> >>> IF DON'T WANT PERSONAL ATTACKS, DON'T MAKE STUPID **** UP!!!
> >
> > I am making nothing up; I'm simply reporting the result of aome tests I ran, no counter the assertion that inductance isn't an issue with electrolytic capacitors.
> >
>
> But you still haven't addressed why you had such a high
> parasitic/lead inductance.
>
> Typical values are around 15nH for a leaded component....
>
> ????

I can only report what I measured; I don't know why the caps measured that way. Incidentally, I also found, when I measured other values of cap:

100µF - 27.9kHz - 34kHz resonance
330µF - 16.2kHz - 17.8kHz "
1,000µF - 9.3kHz - 12.8kHz "
3,300µF - 5.3kHz - 8.6kHz "

So there's a clear correlation -- resonance frequency goes down as capacitance value goes up, broadly speaking. Within a given capacitance value, however, there seemed to be no correlation between physical size and resonance frequency (I confess that I expected to find one, but didn't).

This experiment was done as part of a power supply design project, so most of the capacitors I tested were ones you'd expect to find in power supplies..

Peace,
The Other Paul

Paul[_13_]

February 17th 17, 03:28 AM

Phil Allison[_4_]

February 17th 17, 05:31 AM

PStamler wrote:

>
> I can only report what I measured;
>

** But how no idea of how to measure.

>
> 100µF - 27.9kHz - 34kHz resonance
> 330µF - 16.2kHz - 17.8kHz "
> 1,000µF - 9.3kHz - 12.8kHz "
> 3,300µF - 5.3kHz - 8.6kHz "
>

** All the resonance values are wrong because you included connection lead inductance in the tests.

There is a good reason most electro cap ESR meters work at 100kHz - it is the frequency where electro impedance is at its very lowest over a wide range of values.

The true readings for your 3,300Uf cap are Fo at 19kHz with impedance rising above the ESR value only beyond 250kHz.

The inductance is the same as* 1 inch of wire * and you did not take that fact into account.

The ESR is higher in value until the frequency is in the long wave RF range..

....... Phil

Paul[_13_]

February 17th 17, 05:43 AM

On 2/16/2017 8:12 PM, PStamler wrote:
> On Thursday, February 16, 2017 at 8:39:44 PM UTC-6, Paul wrote:
>> On 2/16/2017 7:23 PM, PStamler wrote:
>>> On Thursday, February 16, 2017 at 2:43:15 PM UTC-6, Paul wrote:
>>>> On 2/16/2017 12:21 PM, PStamler wrote:
>>>>> On Thursday, February 16, 2017 at 1:07:31 AM UTC-6, Paul wrote:
>>>>>> On 2/15/2017 10:13 PM, PStamler wrote:
>>>>>>> At the risk of incurring a personal attack, I wish to report that I have tested common commercially-available capacitors, looking for the resonant frequency (frequency at which the device's impedance bottoms out; the cap stops behaving like a capacitor above this frequency and starts behaving like an inductor). The lowest resonant frequency I found was 5.3kHz, well within the audio range.
>>>>>>>
>>>>>>
>>>>>> What was the capacitor value for that measurement?
>>>>>
>>>>> It was a 3,300µF/50V Panasonic Series NHG electrolytic.
>>>>>
>>>>
>>>>
>>>>
>>>> http://www.murata.com/~/media/webrenewal/products/emc/emifil/knowhow/12to14.ashx
>>>>
>>>> So resonant freq f=1/(2*Pi*(L*C)**0.5)
>>>>
>>>> So L=273nH.
>>>>
>>>> So you had 273nH of parasitic/lead inductance? BULL****.
>>>>
>>>> Also, where in the **** would you need such a large cap
>>>> in a speaker crossover?
>>>
>>> I never said you would. A 3,300µF cap would more likely be found in a power supply, or perhaps in series with Rin in a noninverting opamp circuit.
>>>
>>>>> IF DON'T WANT PERSONAL ATTACKS, DON'T MAKE STUPID **** UP!!!
>>>
>>> I am making nothing up; I'm simply reporting the result of aome tests I ran, no counter the assertion that inductance isn't an issue with electrolytic capacitors.
>>>
>>
>> But you still haven't addressed why you had such a high
>> parasitic/lead inductance.
>>
>> Typical values are around 15nH for a leaded component....
>>
>> ????
>
> I can only report what I measured; I don't know why the caps measured that way. Incidentally, I also found, when I measured other values of cap:
>
> 100µF - 27.9kHz - 34kHz resonance
> 330µF - 16.2kHz - 17.8kHz "
> 1,000µF - 9.3kHz - 12.8kHz "
> 3,300µF - 5.3kHz - 8.6kHz "
>
> So there's a clear correlation -- resonance frequency goes down as capacitance value goes up, broadly speaking. Within a given capacitance value, however, there seemed to be no correlation between physical size and resonance frequency (I confess that I expected to find one, but didn't).
>
> This experiment was done as part of a power supply design project, so most of the capacitors I tested were ones you'd expect to find in power supplies.
>

Actually, you have to watch your ESR and ESL on switching power
supplies even MORE! They will typically use a switching frequency
of 100kHz to several MHz:

http://cds.linear.com/docs/en/application-note/AN140fa.pdf

And from here:

http://www.ti.com/lit/an/slta055/slta055.pdf

For input caps: "Ceramic capacitors placed right at the input of the
regulator reduce ripple voltage amplitude. Only
ceramics have the extremely low ESR that is needed to reduce the ripple
voltage amplitude. These
capacitors must be placed close to the regulator input pins to be
effective. Even a few nanohenries of
stray inductance in the capacitor current path raises the impedance at
the switching frequency to levels
that negate their effectiveness.
Large bulk capacitors do not reduce ripple voltage. The ESR of aluminum
electrolytics and most tantalums
are too high to allow for effective ripple reduction. Large input ripple
voltage can cause large amounts of
ripple current to flow in the bulk capacitors, causing excessive power
dissipation in the ESR parasitic."

For output caps: "The self resonant frequency is considered to be the
maximum usable frequency for a capacitor. Above
this frequency the impedance of the capacitor begins to rise as the ESL
of the capacitor begins to
dominate. Note that each capacitor type has a specific frequency band
over which it is most effective.
Therefore, a capacitor network of multiple capacitor types is more
effective in reducing impedance than
just one type."

Figure 5 and 7 and VERY interesting, and show how you have to design
around the resonant frequency of the caps.

COOL ****!

:)

PStamler

February 17th 17, 06:47 AM

On Thursday, February 16, 2017 at 11:31:35 PM UTC-6, Phil Allison wrote:

> ** All the resonance values are wrong because you included connection lead inductance in the tests.

> The inductance is the same as* 1 inch of wire * and you did not take that fact into account.

I offer the following results in rebuttal. All were obtained with the same test setup, including the samw test leads. From the resonaance figures I calculated the inductaance. These figures are for 3,300µF capacitors.

Incidentally, the inductance of 1" of wire, if it's 22 AWG, is 0.022µH; if it's 12AWG the inductance is 0.0162µH.

Capacitor Inductance
1 0.273µH
2 0.213µH
3 0.213µH
4 0.206µH
5 0.200µH
6 0.161µH
7 0.157µH
8 0.148µH
9 0.148µH
10 0.133µH
11 0.133µH
12 0.104µH

I submit that a ratio of 2.05:1 in inductances suggests that I'm measuring something besides the test leads' inductances.

Peace,
The Other Paul

Trevor

February 17th 17, 07:09 AM

On 17/02/2017 12:41 PM, Scott Dorsey wrote:
> In article >, Trevor > wrote:
>> On 16/02/2017 3:57 PM, PStamler wrote:
>>> At the risk of attracting one of Mr. Allison's personal attacks,I
>>> have measured the rise in capacitors' impedance at high frequencies
>>> -- in some cases they switch from being capacitative to inductive
>>> well within the audio band.
>>
>> If you often use power supply filter caps etc. for audio coupling
>> purposes, that's definitely something you'd would want to take into
>> account I guess. :-) There is a reason why one selects components for
>> purpose of course.
>
> Sadly, that was the technology of the 1970s. People were designing with
> transistors but they were still thinking about tubes in their heads, so
> everything was capacitively coupled and electrolytics were needed in order
> to deal with the high values required due to the low impedances.

Name ONE item that used a 3,300uF cap (as mentioned elsewhere) for
coupling? I never saw one.

> I was at a mastering facility a few years back with some audiophile label
> guys who were looking at having some LPs cut. They asked the mastering
> engineer if there were any electrolytic capacitors in the signal path of
> the Neumann lathe amplifier and he about spit himself. "Millions of them!"
> he said. "Millions!"

Well hundreds anyway, and perfectly adequate for the purpose. Least of
your worries with vinyl!

>
> And so, because we live with a lot of older equipment designed in this
> regime, we have to deal with it and we have to find capacitors appropriate
> for the application.

Exactly. Not 3,300uF filter caps for coupling purposes.

Trevor.

Trevor

February 17th 17, 07:19 AM

On 17/02/2017 2:12 PM, PStamler wrote:
> On Thursday, February 16, 2017 at 8:39:44 PM UTC-6, Paul wrote:
>> On 2/16/2017 7:23 PM, PStamler wrote:
>>> On Thursday, February 16, 2017 at 2:43:15 PM UTC-6, Paul wrote:
>>>> On 2/16/2017 12:21 PM, PStamler wrote:
>>>>> On Thursday, February 16, 2017 at 1:07:31 AM UTC-6, Paul
>>>>> wrote:
>>>>>> On 2/15/2017 10:13 PM, PStamler wrote:
>>>>>>> At the risk of incurring a personal attack, I wish to
>>>>>>> report that I have tested common commercially-available
>>>>>>> capacitors, looking for the resonant frequency (frequency
>>>>>>> at which the device's impedance bottoms out; the cap
>>>>>>> stops behaving like a capacitor above this frequency and
>>>>>>> starts behaving like an inductor). The lowest resonant
>>>>>>> frequency I found was 5.3kHz, well within the audio
>>>>>>> range.
>>>>>>>
>>>>>>
>>>>>> What was the capacitor value for that measurement?
>>>>>
>>>>> It was a 3,300µF/50V Panasonic Series NHG electrolytic.
>>>>>
>>>>
>>>>
>>>>
>>>> http://www.murata.com/~/media/webrenewal/products/emc/emifil/knowhow/12to14.ashx
>>>>
>>>>
>>>>
So resonant freq f=1/(2*Pi*(L*C)**0.5)
>>>>
>>>> So L=273nH.
>>>>
>>>> So you had 273nH of parasitic/lead inductance? BULL****.
>>>>
>>>> Also, where in the **** would you need such a large cap in a
>>>> speaker crossover?
>>>
>>> I never said you would. A 3,300µF cap would more likely be found
>>> in a power supply, or perhaps in series with Rin in a
>>> noninverting opamp circuit.
>>>
>>>>> IF DON'T WANT PERSONAL ATTACKS, DON'T MAKE STUPID **** UP!!!
>>>
>>> I am making nothing up; I'm simply reporting the result of aome
>>> tests I ran, no counter the assertion that inductance isn't an
>>> issue with electrolytic capacitors.
>>>
>>
>> But you still haven't addressed why you had such a high
>> parasitic/lead inductance.
>>
>> Typical values are around 15nH for a leaded component....
>>
>> ????
>
> I can only report what I measured; I don't know why the caps measured
> that way. Incidentally, I also found, when I measured other values of
> cap:
>
> 100µF - 27.9kHz - 34kHz resonance 330µF - 16.2kHz - 17.8kHz "
> 1,000µF - 9.3kHz - 12.8kHz " 3,300µF - 5.3kHz - 8.6kHz "
>
> So there's a clear correlation -- resonance frequency goes down as
> capacitance value goes up, broadly speaking. Within a given
> capacitance value, however, there seemed to be no correlation between
> physical size and resonance frequency (I confess that I expected to
> find one, but didn't).
>
> This experiment was done as part of a power supply design project, so
> most of the capacitors I tested were ones you'd expect to find in
> power supplies.

Exactly, and those resonances are hardly a problem at 50-120Hz. People
do often worry about which caps they choose in the 100uF range for
coupling purposes however.

Trevor.

Trevor

February 17th 17, 07:27 AM

On 17/02/2017 5:47 PM, PStamler wrote:
> I offer the following results in rebuttal. All were obtained with the
> same test setup, including the samw test leads. From the resonaance
> figures I calculated the inductaance. These figures are for 3,300µF
> capacitors.
>
> Incidentally, the inductance of 1" of wire, if it's 22 AWG, is
> 0.022µH; if it's 12AWG the inductance is 0.0162µH.
>
> Capacitor Inductance 1 0.273µH 2 0.213µH 3
> 0.213µH 4 0.206µH 5 0.200µH 6 0.161µH 7
> 0.157µH 8 0.148µH 9 0.148µH 10 0.133µH 11
> 0.133µH 12 0.104µH
>
> I submit that a ratio of 2.05:1 in inductances suggests that I'm
> measuring something besides the test leads' inductances.

According to your own figures, a foot of test lead (*2), will be in the
same ball park as your readings, or more!

Trevor.

PStamler

February 17th 17, 08:18 AM

On Friday, February 17, 2017 at 1:28:00 AM UTC-6, Trevor wrote:
> On 17/02/2017 5:47 PM, PStamler wrote:
> > I offer the following results in rebuttal. All were obtained with the
> > same test setup, including the samw test leads. From the resonaance
> > figures I calculated the inductaance. These figures are for 3,300µF
> > capacitors.
> >
> > Incidentally, the inductance of 1" of wire, if it's 22 AWG, is
> > 0.022µH; if it's 12AWG the inductance is 0.0162µH.
> >
> > Capacitor Inductance 1 0.273µH 2 0.213µH 3
> > 0.213µH 4 0.206µH 5 0.200µH 6 0.161µH 7
> > 0.157µH 8 0.148µH 9 0.148µH 10 0.133µH 11
> > 0.133µH 12 0.104µH
> >
> > I submit that a ratio of 2.05:1 in inductances suggests that I'm
> > measuring something besides the test leads' inductances.
>
> According to your own figures, a foot of test lead (*2), will be in the
> same ball park as your readings, or more!

Actually just 1 foot; my clip leads were 6" long.

But you make a good point.

Peace,
The Other Paul

Paul[_13_]

February 17th 17, 10:25 AM

On 2/17/2017 1:18 AM, PStamler wrote:
> On Friday, February 17, 2017 at 1:28:00 AM UTC-6, Trevor wrote:
>> On 17/02/2017 5:47 PM, PStamler wrote:
>>> I offer the following results in rebuttal. All were obtained with the
>>> same test setup, including the samw test leads. From the resonaance
>>> figures I calculated the inductaance. These figures are for 3,300µF
>>> capacitors.
>>>
>>> Incidentally, the inductance of 1" of wire, if it's 22 AWG, is
>>> 0.022µH; if it's 12AWG the inductance is 0.0162µH.
>>>
>>> Capacitor Inductance 1 0.273µH 2 0.213µH 3
>>> 0.213µH 4 0.206µH 5 0.200µH 6 0.161µH 7
>>> 0.157µH 8 0.148µH 9 0.148µH 10 0.133µH 11
>>> 0.133µH 12 0.104µH
>>>
>>> I submit that a ratio of 2.05:1 in inductances suggests that I'm
>>> measuring something besides the test leads' inductances.
>>
>> According to your own figures, a foot of test lead (*2), will be in the
>> same ball park as your readings, or more!
>
> Actually just 1 foot; my clip leads were 6" long.
>
> But you make a good point.
>

Redo the test, without the test leads.

The resonant frequency should go up...

Don Pearce[_3_]

February 17th 17, 10:45 AM

On Fri, 17 Feb 2017 03:25:18 -0700, Paul > wrote:

>On 2/17/2017 1:18 AM, PStamler wrote:
>> On Friday, February 17, 2017 at 1:28:00 AM UTC-6, Trevor wrote:
>>> On 17/02/2017 5:47 PM, PStamler wrote:
>>>> I offer the following results in rebuttal. All were obtained with the
>>>> same test setup, including the samw test leads. From the resonaance
>>>> figures I calculated the inductaance. These figures are for 3,300µF
>>>> capacitors.
>>>>
>>>> Incidentally, the inductance of 1" of wire, if it's 22 AWG, is
>>>> 0.022µH; if it's 12AWG the inductance is 0.0162µH.
>>>>
>>>> Capacitor Inductance 1 0.273µH 2 0.213µH 3
>>>> 0.213µH 4 0.206µH 5 0.200µH 6 0.161µH 7
>>>> 0.157µH 8 0.148µH 9 0.148µH 10 0.133µH 11
>>>> 0.133µH 12 0.104µH
>>>>
>>>> I submit that a ratio of 2.05:1 in inductances suggests that I'm
>>>> measuring something besides the test leads' inductances.
>>>
>>> According to your own figures, a foot of test lead (*2), will be in the
>>> same ball park as your readings, or more!
>>
>> Actually just 1 foot; my clip leads were 6" long.
>>
>> But you make a good point.
>>
>
> Redo the test, without the test leads.
>
> The resonant frequency should go up...

Easier way. Just calibrate by connecting the test leads together.
Measure the inductance and subtract it from whatever is later
measured.

d

---
This email has been checked for viruses by Avast antivirus software.
https://www.avast.com/antivirus

geoff

February 17th 17, 11:39 AM

On 17/02/2017 7:47 PM, PStamler wrote:
> On Thursday, February 16, 2017 at 11:31:35 PM UTC-6, Phil Allison wrote:
>
>> ** All the resonance values are wrong because you included connection lead inductance in the tests.
>
>
>> The inductance is the same as* 1 inch of wire * and you did not take that fact into account.
>
> I offer the following results in rebuttal. All were obtained with the same test setup, including the samw test leads. From the resonaance figures I calculated the inductaance. These figures are for 3,300µF capacitors.
>
> Incidentally, the inductance of 1" of wire, if it's 22 AWG, is 0.022µH; if it's 12AWG the inductance is 0.0162µH.
>
> Capacitor Inductance
> 1 0.273µH
> 2 0.213µH
> 3 0.213µH
> 4 0.206µH
> 5 0.200µH
> 6 0.161µH
> 7 0.157µH
> 8 0.148µH
> 9 0.148µH
> 10 0.133µH
> 11 0.133µH
> 12 0.104µH
>
> I submit that a ratio of 2.05:1 in inductances suggests that I'm measuring something besides the test leads' inductances.
>
> Peace,
> The Other Paul
>

Why not actually test capacitors that are likely to be found in series
in a signal path, rather than used in their designed application as
reservoir capacitors in a power supply (and maybe in one model
audiophool valve/tube power amp) ?

And on input to my old KEF R105 crossovers (but I'm sure that was a
designed-in factor)

geoff

Don Pearce[_3_]

February 17th 17, 12:34 PM

On Sat, 18 Feb 2017 00:39:04 +1300, geoff >
wrote:

>On 17/02/2017 7:47 PM, PStamler wrote:
>> On Thursday, February 16, 2017 at 11:31:35 PM UTC-6, Phil Allison wrote:
>>
>>> ** All the resonance values are wrong because you included connection lead inductance in the tests.
>>
>>
>>> The inductance is the same as* 1 inch of wire * and you did not take that fact into account.
>>
>> I offer the following results in rebuttal. All were obtained with the same test setup, including the samw test leads. From the resonaance figures I calculated the inductaance. These figures are for 3,300µF capacitors.
>>
>> Incidentally, the inductance of 1" of wire, if it's 22 AWG, is 0.022µH; if it's 12AWG the inductance is 0.0162µH.
>>
>> Capacitor Inductance
>> 1 0.273µH
>> 2 0.213µH
>> 3 0.213µH
>> 4 0.206µH
>> 5 0.200µH
>> 6 0.161µH
>> 7 0.157µH
>> 8 0.148µH
>> 9 0.148µH
>> 10 0.133µH
>> 11 0.133µH
>> 12 0.104µH
>>
>> I submit that a ratio of 2.05:1 in inductances suggests that I'm measuring something besides the test leads' inductances.
>>
>> Peace,
>> The Other Paul
>>
>
>Why not actually test capacitors that are likely to be found in series
>in a signal path, rather than used in their designed application as
>reservoir capacitors in a power supply (and maybe in one model
>audiophool valve/tube power amp) ?
>
>And on input to my old KEF R105 crossovers (but I'm sure that was a
>designed-in factor)
>

Hang on there. Power supply capacitors absolutely are in the signal
path. The alternating speaker current flows through them as a series
element.

d

---
This email has been checked for viruses by Avast antivirus software.
https://www.avast.com/antivirus

Scott Dorsey

February 17th 17, 02:24 PM

david gourley > wrote:
(Scott Dorsey) :
>
>> david gourley > wrote:
>>>
>>>My Dynaco ST120 channels use that value for output coupling to the speaker.
>>
>> Is mentioning an ST120 like mentioning Hitler? Is this thread closed now?
>
>Wow, sorry didn't know it was THAT bad. I've used it for a guitar and bass
>amp.

It has some interesting issues. It's slew-limited, which was typical of
amps back then, but in part because of single supply rail and the
non-complementary output stage, the slew limiting is asymmetric.

On top of all that it has that huge blocking capacitor on the output which
isn't helping anything... and then they wind the output choke around the
blocking capacitor which is not exactly a linear core... that becomes a
really remarkable source of distortion.

Probably make a perfectly fine bass amp, though. Because of all that DC
blocking and that huge choke on the output to move all the poles to the
left, the thing is much more stable than many of the amps from that era.

But, when I think about why solid state electronics had such a bad reputation
for sound quality in the seventies, the ST120 is one of the first things I
think about.
--scott

--
"C'est un Nagra. C'est suisse, et tres, tres precis."

Scott Dorsey

February 17th 17, 02:26 PM

In article >, Paul > wrote:
>
> But you still haven't addressed why you had such a high
>parasitic/lead inductance.

He hasn't, but that doesn't mean it's not there. And it's not the lead
inductance at all, it's the inductance of the foil wrap. We're talking
some hundreds of turns in many capacitors.

If all this bothers you, use a stacked film cap and don't worry.
--scott
--
"C'est un Nagra. C'est suisse, et tres, tres precis."

[email protected]

February 17th 17, 03:17 PM

On Friday, February 17, 2017 at 9:26:38 AM UTC-5, Scott Dorsey wrote:
> In article >, Paul > wrote:
> >
> > But you still haven't addressed why you had such a high
> >parasitic/lead inductance.
>
> He hasn't, but that doesn't mean it's not there. And it's not the lead
> inductance at all, it's the inductance of the foil wrap. We're talking
> some hundreds of turns in many capacitors.
>
> If all this bothers you, use a stacked film cap and don't worry.
> --scott
> --
> "C'est un Nagra. C'est suisse, et tres, tres precis."

yes.... and regardless of what the cause of that inductance that the OP is seeing, the value of that inductance and the value of the Z at audio is sooo looow that it is of no concern.

m

david gourley[_2_]

February 17th 17, 05:44 PM

(Scott Dorsey) :

> david gourley > wrote:
(Scott Dorsey) said...news:o85ked$gqe$1
@panix2.panix.com:
>>
>>> david gourley > wrote:
>>>>
>>>>My Dynaco ST120 channels use that value for output coupling to the
speaker.
>>>
>>> Is mentioning an ST120 like mentioning Hitler? Is this thread closed
now?
>>
>>Wow, sorry didn't know it was THAT bad. I've used it for a guitar and
bass
>>amp.
>
> It has some interesting issues. It's slew-limited, which was typical of
> amps back then, but in part because of single supply rail and the
> non-complementary output stage, the slew limiting is asymmetric.
>
> On top of all that it has that huge blocking capacitor on the output
which
> isn't helping anything... and then they wind the output choke around the
> blocking capacitor which is not exactly a linear core... that becomes a
> really remarkable source of distortion.
>
> Probably make a perfectly fine bass amp, though. Because of all that DC
> blocking and that huge choke on the output to move all the poles to the
> left, the thing is much more stable than many of the amps from that era.
>
> But, when I think about why solid state electronics had such a bad
reputation
> for sound quality in the seventies, the ST120 is one of the first things
I
> think about.
> --scott
>

Surely those Southwest Tiger amps fit in some place near the top ? <g>

david

---
This email has been checked for viruses by Avast antivirus software.
https://www.avast.com/antivirus

Mat Nieuwenhoven

February 17th 17, 05:58 PM

On 17 Feb 2017 09:26:36 -0500, Scott Dorsey wrote:

>In article >, Paul > wrote:
>>
>> But you still haven't addressed why you had such a high
>>parasitic/lead inductance.
>
>He hasn't, but that doesn't mean it's not there. And it's not the lead
>inductance at all, it's the inductance of the foil wrap. We're talking
>some hundreds of turns in many capacitors.
>
>If all this bothers you, use a stacked film cap and don't worry.

The capacity of some capacitors (especially multi layer ceramic) is
dependent on the voltage across them; in some cases the value gets
halved! You don't want such in an audio path if the audio voltage is
a significant part of the blocking voltage (if any). See
http://www.eetimes.com/author.asp?section_id=30&doc_id=1330877& .

Also
http://www.intersil.com/content/dam/Intersil/documents/an13/an1325.pdf
has a nice table of different capacitor types and their trade-offs.

Mat Nieuwenhoven

Scott Dorsey

February 17th 17, 06:28 PM

david gourley > wrote:
>
>Surely those Southwest Tiger amps fit in some place near the top ? <g>

They have some stability issues, to say the least. But as far as exploding
into flames go, they are no match for the Phase Linears.
--scott

--
"C'est un Nagra. C'est suisse, et tres, tres precis."

Paul[_13_]

February 17th 17, 07:02 PM

On 2/17/2017 7:26 AM, Scott Dorsey wrote:
> In article >, Paul > wrote:
>>
>> But you still haven't addressed why you had such a high
>> parasitic/lead inductance.
>
> He hasn't, but that doesn't mean it's not there. And it's not the lead
> inductance at all, it's the inductance of the foil wrap. We're talking
> some hundreds of turns in many capacitors.
>
> If all this bothers you, use a stacked film cap and don't worry.
> --scott
>

So let's assume the leaded capacitor has about 15nH of self
inductance.

270-15= 255 nH.

And let's guesstimate 6nH of inductance per cm of lead length.

42.5 cm of added lead length???

THAT WOULD BE SLOPPY ENGINEERING!!!!

:/

http://sound.whsites.net/articles/capacitors.htm

[email protected]

February 17th 17, 09:11 PM

..
>
> The capacity of some capacitors (especially multi layer ceramic) is
> dependent on the voltage across them; in some cases the value gets
> halved! You don't want such in an audio path if the audio voltage is
> a significant part of the blocking voltage (if any). See
> http://www.eetimes.com/author.asp?section_id=30&doc_id=1330877& .
>
> Also
> http://www.intersil.com/content/dam/Intersil/documents/an13/an1325.pdf
> has a nice table of different capacitor types and their trade-offs.
>
> Mat Nieuwenhoven

no it's not that bad

if the cap "in the audio path" is a coupling cap
then it's job is to have as little as possible audio voltage drop
across it... and it will be sized accordingly.

So then only at very high amplitude and very low frequency bass, i.e. below the - 3 dB point, will there be any significant voltage drop ACROSS the cap.

In that special case, there may be added distortion.

The other case is when the cap is used as part of a filter and there is
large audio voltage ACROSS (not through) the cap.

In most ordinary coupling cap applications, it is not a problem.

m

Chuck[_12_]

February 17th 17, 10:59 PM

On 17 Feb 2017 13:28:50 -0500, (Scott Dorsey) wrote:

>david gourley > wrote:
>>
>>Surely those Southwest Tiger amps fit in some place near the top ? <g>
>
>They have some stability issues, to say the least. But as far as exploding
>into flames go, they are no match for the Phase Linears.
>--scott
I came up with a mod for these that made them extremely stable.

---
This email has been checked for viruses by Avast antivirus software.
https://www.avast.com/antivirus

John Hardy

February 17th 17, 11:24 PM

On 2/17/17 11:58 AM, Mat Nieuwenhoven wrote:
> On 17 Feb 2017 09:26:36 -0500, Scott Dorsey wrote:

>
> The capacity of some capacitors (especially multi layer ceramic) is
> dependent on the voltage across them; in some cases the value gets
> halved! You don't want such in an audio path if the audio voltage is
> a significant part of the blocking voltage (if any). See
> http://www.eetimes.com/author.asp?section_id=30&doc_id=1330877& .
>
> Also
> http://www.intersil.com/content/dam/Intersil/documents/an13/an1325.pdf
> has a nice table of different capacitor types and their trade-offs.
>
> Mat Nieuwenhoven
>

Both of those references seem to discuss the shortcomings of the
crappier ceramic capacitors, with just a passing reference to the
premium ceramic capacitors known as the "COG" or "NP0" types. The
COG/NP0 type deserves special consideration. If anyone is interested,
page 8 of my 990 data package describes some of the differences between
the three most common types of ceramic capacitors, the COG/NP0, X7R and Z5U.

http://www.johnhardyco.com/pdf/990.pdf

Thank you.

John Hardy
The John Hardy Co.

Scott Dorsey

February 18th 17, 01:11 AM

Mat Nieuwenhoven > wrote:
>The capacity of some capacitors (especially multi layer ceramic) is
>dependent on the voltage across them; in some cases the value gets
>halved! You don't want such in an audio path if the audio voltage is
>a significant part of the blocking voltage (if any). See
>http://www.eetimes.com/author.asp?section_id=30&doc_id=1330877& .

Hey! I bet I could use that as a compressor!
--scott

--
"C'est un Nagra. C'est suisse, et tres, tres precis."

Scott Dorsey

February 18th 17, 01:13 AM

In article >, Paul > wrote:
> So let's assume the leaded capacitor has about 15nH of self
>inductance.
>
> 270-15= 255 nH.
>
> And let's guesstimate 6nH of inductance per cm of lead length.
>
> 42.5 cm of added lead length???

Sounds about right, since you have more than a meter of foil wrapped up
inside that thing, and you have mutual coupling between winds.

> THAT WOULD BE SLOPPY ENGINEERING!!!!

Don't like it? Use a stacked film type!
--scott
--
"C'est un Nagra. C'est suisse, et tres, tres precis."

Phil Allison[_4_]

February 18th 17, 01:44 AM

PStamler wrote:
>
> Phil Allison wrote:
>
> > ** All the resonance values are wrong because you included connection lead inductance in the tests.
>
>
> > The inductance is the same as* 1 inch of wire * and you did not take that fact into account.
>
>
> I offer the following results in rebuttal.
>

>
> Incidentally, the inductance of 1" of wire, if it's 22 AWG, is 0.022µH;
> if it's 12AWG the inductance is 0.0162µH.
>

** Correct - so ****ing what?

Consider that your fake results are CONTRADICTED by everyone else !!

BTW:

You snipped my info on electro ESR meters working at 100kHz.

It alone PROVES you are wrong.

You are one stubborn POS aren't you ?

..... Phil

PStamler

February 18th 17, 01:49 AM

On Friday, February 17, 2017 at 7:44:12 PM UTC-6, Phil Allison wrote:
> PStamler wrote:
> >
> > Phil Allison wrote:
> >
> > > ** All the resonance values are wrong because you included connection lead inductance in the tests.
> >
> >
> > > The inductance is the same as* 1 inch of wire * and you did not take that fact into account.
> >
> >
> > I offer the following results in rebuttal.
> >
>
>
>
> >
> > Incidentally, the inductance of 1" of wire, if it's 22 AWG, is 0.022µH;
> > if it's 12AWG the inductance is 0.0162µH.
> >
>
> ** Correct - so ****ing what?
>
> Consider that your fake results are CONTRADICTED by everyone else !!

They weren't fake, just wrong, which I freely acknowledge.

> BTW:
>
> You snipped my info on electro ESR meters working at 100kHz.
>
> It alone PROVES you are wrong.

It's actually a side issue.

> You are one stubborn POS aren't you ?

Yes -- it's a useful survival skill.

Peace,
The Other Paul
>
>
> .... Phil

geoff

February 18th 17, 02:00 AM

On 18/02/2017 2:13 PM, Scott Dorsey wrote:
> In article >, Paul > wrote:
>> So let's assume the leaded capacitor has about 15nH of self
>> inductance.
>>
>> 270-15= 255 nH.
>>
>> And let's guesstimate 6nH of inductance per cm of lead length.
>>
>> 42.5 cm of added lead length???
>
> Sounds about right, since you have more than a meter of foil wrapped up
> inside that thing, and you have mutual coupling between winds.
>
>> THAT WOULD BE SLOPPY ENGINEERING!!!!
>
> Don't like it? Use a stacked film type!
> --scott
>

Isn't a wonder that one can hear anything vaguely coherent at all out of
nearly all audio gear ....

geoff

Phil Allison[_4_]

February 18th 17, 02:00 AM

PStamler wrote:

>
> >
> > Consider that your fake results are CONTRADICTED by everyone else !!
>
> They weren't fake, just wrong, which I freely acknowledge.
>

** Really - when did that happen ??

> > BTW:
> >
> > You snipped my info on electro ESR meters working at 100kHz.
> >
> > It alone PROVES you are wrong.
>
> It's actually a side issue.
>

** No, it proves how naïve and stubborn you are.

That electro ESR meters typically work at 100kHz *contradicts* all your mad assertions.

It PROVES that ESL has no effect on impedance until well above that frequency.

There are none so blind as those who will not see.

> > You are one stubborn POS aren't you ?
>
> Yes -- it's a useful survival skill.
>

** No round here pal.

..... Phil

Scott Dorsey

February 18th 17, 02:26 AM

geoff > wrote:
>
>Isn't a wonder that one can hear anything vaguely coherent at all out of
>nearly all audio gear ....

Well, thats part of the problem with measurement. If you measure carefully
enough, you'll find all kinds of weird stuff going on. Whether it actually
matters or not is a different issue.
--scott
--
"C'est un Nagra. C'est suisse, et tres, tres precis."

Paul[_13_]

February 18th 17, 03:57 AM

On 2/17/2017 6:13 PM, Scott Dorsey wrote:
> In article >, Paul > wrote:
>> So let's assume the leaded capacitor has about 15nH of self
>> inductance.
>>
>> 270-15= 255 nH.
>>
>> And let's guesstimate 6nH of inductance per cm of lead length.
>>
>> 42.5 cm of added lead length???
>
> Sounds about right, since you have more than a meter of foil wrapped up
> inside that thing, and you have mutual coupling between winds.
>

Incorrect.

270nH would be a ridiculous amount of parasitic inductance.

The Other Paul admitted his leads were 6", so he added
about a foot of lead length!

And look here again:

http://www.ti.com/lit/an/slta055/slta055.pdf

Check plots 4, 5, and 7.

Calculate the ESL from the resonant frequencies.

You'll see most are in the tens of nH.

>> THAT WOULD BE SLOPPY ENGINEERING!!!!
>
> Don't like it? Use a stacked film type!
> --scott
>

PStamler

February 18th 17, 05:23 AM

On Friday, February 17, 2017 at 8:00:45 PM UTC-6, Phil Allison wrote:
> PStamler wrote:
>
> >
> > >
> > > Consider that your fake results are CONTRADICTED by everyone else !!
> >
> > They weren't fake, just wrong, which I freely acknowledge.
> >
>
> ** Really - when did that happen ??

Just now -- see above.

Peace,
The Other Paul

Mat Nieuwenhoven

February 18th 17, 06:43 PM

On Fri, 17 Feb 2017 13:11:59 -0800 (PST), wrote:

>..
>>
>> The capacity of some capacitors (especially multi layer ceramic) is
>> dependent on the voltage across them; in some cases the value gets
>> halved! You don't want such in an audio path if the audio voltage is
>> a significant part of the blocking voltage (if any). See
>> http://www.eetimes.com/author.asp?section_id=30&doc_id=1330877& .
>>
>> Also
>> http://www.intersil.com/content/dam/Intersil/documents/an13/an1325.pdf
>> has a nice table of different capacitor types and their trade-offs.
>>
>> Mat Nieuwenhoven
>
>no it's not that bad
>
>if the cap "in the audio path" is a coupling cap
>then it's job is to have as little as possible audio voltage drop
>across it... and it will be sized accordingly.

In general yes, but with non-pro equipment cost is an important issue
and even coupling caps are likely to be as small as possible.
Fortunately, they are unlike to be the multilayer ceramics.
Electrolytics are much better in this respect.

>So then only at very high amplitude and very low frequency bass, i.e. below the - 3 dB point, will there be any significant voltage drop ACROSS the cap.
>
>In that special case, there may be added distortion.

It's not just the voltage. If the cap is voltage dependent, then the
current through it as result of an AC voltage will be distorted. See
for instance
http://www.cliftonlaboratories.com/capacitor_voltage_change.htm , a
little bit down you see a plot of 922 Hz voltage / current
over/through a Y5U type cap (about the worst case) . Even though it's
not so real-life, I found this shocking.
The same page also shows that electrolytic and tantalum caps are much
better in this respect.

>The other case is when the cap is used as part of a filter and there is
>large audio voltage ACROSS (not through) the cap.
>
>In most ordinary coupling cap applications, it is not a problem.

True. Most higher audio voltages, with apologies to our
thermionic-loving friends :-) , will nowadays be found in internal
passive loudspeaker filters, and there high-quality caps are
typically used.

Mat Nieuwenhoven

Mat Nieuwenhoven

February 18th 17, 06:52 PM

On 17 Feb 2017 20:11:57 -0500, Scott Dorsey wrote:

>Mat Nieuwenhoven > wrote:
>>The capacity of some capacitors (especially multi layer ceramic) is
>>dependent on the voltage across them; in some cases the value gets
>>halved! You don't want such in an audio path if the audio voltage is
>>a significant part of the blocking voltage (if any). See
>>http://www.eetimes.com/author.asp?section_id=30&doc_id=1330877& .
>
>Hey! I bet I could use that as a compressor!

:-) As a distortion thingy, more likely, albeit frequency and volume
dependent. See
http://www.cliftonlaboratories.com/capacitor_voltage_change.htm ,
where a sine voltage across the cap results in a more or less
triangular current through it. The distortion would be more if the AC
was higher. The graph show the tops more or less OK, but the flanks
are distorted. I bet this would sound very different from
top-limiting, such as in guitar amps or overdriven tubes.

Mat Nieuwenhoven

Mat Nieuwenhoven

February 18th 17, 07:36 PM

On Fri, 17 Feb 2017 17:24:55 -0600, John Hardy wrote:

>On 2/17/17 11:58 AM, Mat Nieuwenhoven wrote:
>> On 17 Feb 2017 09:26:36 -0500, Scott Dorsey wrote:
>
>>
>> The capacity of some capacitors (especially multi layer ceramic) is
>> dependent on the voltage across them; in some cases the value gets
>> halved! You don't want such in an audio path if the audio voltage is
>> a significant part of the blocking voltage (if any). See
>> http://www.eetimes.com/author.asp?section_id=30&doc_id=1330877& .
>>
>> Also
>> http://www.intersil.com/content/dam/Intersil/documents/an13/an1325.pdf
>> has a nice table of different capacitor types and their trade-offs.
>>
>> Mat Nieuwenhoven
>>
>
>Both of those references seem to discuss the shortcomings of the
>crappier ceramic capacitors, with just a passing reference to the
>premium ceramic capacitors known as the "COG" or "NP0" types. The
>COG/NP0 type deserves special consideration. If anyone is interested,
>page 8 of my 990 data package describes some of the differences between
>the three most common types of ceramic capacitors, the COG/NP0, X7R and Z5U.
>
>http://www.johnhardyco.com/pdf/990.pdf

Very interesting document, thanks. Indeed, the COG/NP0 caps are fine
in this respect.

Are transformer-based mic amps still used? I can see that a
transformer-based gain is essentially noise-free, but aren't they
sensitive to microphone impedance?

One question about the MPC-1 mic pre-amp schematic, if I may. For the
+/- 15V the 78L15/79L15 regulators are used. I thought that these
were quite noisy? I've seen recommenations to use adjustable
regulators ones instead.

Mat Nieuwenhoven

Mat Nieuwenhoven

February 18th 17, 08:26 PM

On Sat, 18 Feb 2017 20:45:19 GMT, Don Pearce wrote:

<snip>

>>>Both of those references seem to discuss the shortcomings of the
>>>crappier ceramic capacitors, with just a passing reference to the
>>>premium ceramic capacitors known as the "COG" or "NP0" types. The
>>>COG/NP0 type deserves special consideration. If anyone is interested,
>>>page 8 of my 990 data package describes some of the differences between
>>>the three most common types of ceramic capacitors, the COG/NP0, X7R and Z5U.
>>>
>>>http://www.johnhardyco.com/pdf/990.pdf
>>
>>Very interesting document, thanks. Indeed, the COG/NP0 caps are fine
>>in this respect.
>>
>>Are transformer-based mic amps still used? I can see that a
>>transformer-based gain is essentially noise-free, but aren't they
>>sensitive to microphone impedance?
>>
>>One question about the MPC-1 mic pre-amp schematic, if I may. For the
>>+/- 15V the 78L15/79L15 regulators are used. I thought that these
>>were quite noisy? I've seen recommenations to use adjustable
>>regulators ones instead.
>>
>>Mat Nieuwenhoven
>>
>>
>>
>Transformer based gain? It isn't gain - it is just impedance
>transformation.
Yes, but at the same time transforming voltage. And because the gain
stages after that are only interested in voltage, not power, why
wouldn't it be a noise-free voltage gain?

>And most transformers have a loss about 1dB, which
>equates to a 1dB added noise figure.
The 990 opamp pdf lists a transformer gain of 5.6 dB.

>If you want a quiet preamp you do
>away with the transformer and design the front end to present a noise
>match to the mic.
Understood, so such a preamp will have a recommended mic impedance?

>This will also provide the lowest distortion, which
>transformers won't manage at the low end.
I didn't know transformers have more distortion at low levels. Do
current top-of-the line mic preamps in mixers use discrete or
integrated circruitry?

>As for regulator noise, if your preamp has even a half-way decent
>PSRR, it simply isn't a factor.

Unless it's set to a high gain, which it isn't here.

Mat Nieuwenhoven

Don Pearce[_3_]

February 18th 17, 08:45 PM

On Sat, 18 Feb 2017 20:36:32 +0100 (CET), "Mat Nieuwenhoven"
> wrote:

>On Fri, 17 Feb 2017 17:24:55 -0600, John Hardy wrote:
>
>>On 2/17/17 11:58 AM, Mat Nieuwenhoven wrote:
>>> On 17 Feb 2017 09:26:36 -0500, Scott Dorsey wrote:
>>
>>>
>>> The capacity of some capacitors (especially multi layer ceramic) is
>>> dependent on the voltage across them; in some cases the value gets
>>> halved! You don't want such in an audio path if the audio voltage is
>>> a significant part of the blocking voltage (if any). See
>>> http://www.eetimes.com/author.asp?section_id=30&doc_id=1330877& .
>>>
>>> Also
>>> http://www.intersil.com/content/dam/Intersil/documents/an13/an1325.pdf
>>> has a nice table of different capacitor types and their trade-offs.
>>>
>>> Mat Nieuwenhoven
>>>
>>
>>Both of those references seem to discuss the shortcomings of the
>>crappier ceramic capacitors, with just a passing reference to the
>>premium ceramic capacitors known as the "COG" or "NP0" types. The
>>COG/NP0 type deserves special consideration. If anyone is interested,
>>page 8 of my 990 data package describes some of the differences between
>>the three most common types of ceramic capacitors, the COG/NP0, X7R and Z5U.
>>
>>http://www.johnhardyco.com/pdf/990.pdf
>
>Very interesting document, thanks. Indeed, the COG/NP0 caps are fine
>in this respect.
>
>Are transformer-based mic amps still used? I can see that a
>transformer-based gain is essentially noise-free, but aren't they
>sensitive to microphone impedance?
>
>One question about the MPC-1 mic pre-amp schematic, if I may. For the
>+/- 15V the 78L15/79L15 regulators are used. I thought that these
>were quite noisy? I've seen recommenations to use adjustable
>regulators ones instead.
>
>Mat Nieuwenhoven
>
>
>
Transformer based gain? It isn't gain - it is just impedance
transformation. And most transformers have a loss about 1dB, which
equates to a 1dB added noise figure. If you want a quiet preamp you do
away with the transformer and design the front end to present a noise
match to the mic. This will also provide the lowest distortion, which
transformers won't manage at the low end.

As for regulator noise, if your preamp has even a half-way decent
PSRR, it simply isn't a factor.
d

---
This email has been checked for viruses by Avast antivirus software.
https://www.avast.com/antivirus

Scott Dorsey

February 18th 17, 09:07 PM

Mat Nieuwenhoven > wrote:
>
>:-) As a distortion thingy, more likely, albeit frequency and volume
>dependent. See
>http://www.cliftonlaboratories.com/capacitor_voltage_change.htm ,
>where a sine voltage across the cap results in a more or less
>triangular current through it. The distortion would be more if the AC
>was higher. The graph show the tops more or less OK, but the flanks
>are distorted. I bet this would sound very different from
>top-limiting, such as in guitar amps or overdriven tubes.

Sure, but more interestingly I could put a large DC bias voltage across
them, and a small AC signal voltage. As I adjust the DC bias, the low
frequency corner moves up and down.

A little like the whole magnetic amplifier trick.
--scott

--
"C'est un Nagra. C'est suisse, et tres, tres precis."

Don Pearce[_3_]

February 18th 17, 09:32 PM

On Sat, 18 Feb 2017 21:26:05 +0100 (CET), "Mat Nieuwenhoven"
> wrote:

>On Sat, 18 Feb 2017 20:45:19 GMT, Don Pearce wrote:
>
><snip>
>
>>>>Both of those references seem to discuss the shortcomings of the
>>>>crappier ceramic capacitors, with just a passing reference to the
>>>>premium ceramic capacitors known as the "COG" or "NP0" types. The
>>>>COG/NP0 type deserves special consideration. If anyone is interested,
>>>>page 8 of my 990 data package describes some of the differences between
>>>>the three most common types of ceramic capacitors, the COG/NP0, X7R and Z5U.
>>>>
>>>>http://www.johnhardyco.com/pdf/990.pdf
>>>
>>>Very interesting document, thanks. Indeed, the COG/NP0 caps are fine
>>>in this respect.
>>>
>>>Are transformer-based mic amps still used? I can see that a
>>>transformer-based gain is essentially noise-free, but aren't they
>>>sensitive to microphone impedance?
>>>
>>>One question about the MPC-1 mic pre-amp schematic, if I may. For the
>>>+/- 15V the 78L15/79L15 regulators are used. I thought that these
>>>were quite noisy? I've seen recommenations to use adjustable
>>>regulators ones instead.
>>>
>>>Mat Nieuwenhoven
>>>
>>>
>>>
>>Transformer based gain? It isn't gain - it is just impedance
>>transformation.
>Yes, but at the same time transforming voltage. And because the gain
>stages after that are only interested in voltage, not power, why
>wouldn't it be a noise-free voltage gain?

No, not noise free. Transformers have a loss figure, which translates
directly into noise.

>
>>And most transformers have a loss about 1dB, which
>>equates to a 1dB added noise figure.
>The 990 opamp pdf lists a transformer gain of 5.6 dB.
>

That is the transformation ratio they recommend to bring the mic
impedance up to the optimum noise match impedance. You still have to
add the transformer loss though.

>>If you want a quiet preamp you do
>>away with the transformer and design the front end to present a noise
>>match to the mic.
>Understood, so such a preamp will have a recommended mic impedance?
>
Yes, although it is a fairly broad peak.

>>This will also provide the lowest distortion, which
>>transformers won't manage at the low end.
>I didn't know transformers have more distortion at low levels. Do
>current top-of-the line mic preamps in mixers use discrete or
>integrated circruitry?
>
Not low levels, low frequencies.

>>As for regulator noise, if your preamp has even a half-way decent
>>PSRR, it simply isn't a factor.
>
>Unless it's set to a high gain, which it isn't here.
>
The gain shouldn't change the PSRR, but it does imply a low signal
level, so proportionately the effect will be greater. A resistor and a
fat electrolytic will kill the noise.

d

---
This email has been checked for viruses by Avast antivirus software.
https://www.avast.com/antivirus

Scott Dorsey

February 18th 17, 11:03 PM

Mat Nieuwenhoven > wrote:
>Are transformer-based mic amps still used? I can see that a
>transformer-based gain is essentially noise-free, but aren't they
>sensitive to microphone impedance?

Yes, that's where the free gain comes from. You trade admittance on
the input for it. It's not noise free, though, because of the thermal
noise of the transformer which can be substantial if you want a high
step-up.

There is a classic JAES paper from the 1980s by Marshall Leach which
is probably still on his website, which does the math for various input
topologies. Sometimes it is a win for noise and sometimes it is not.

But more importantly.. the transformer buys you ENORMOUS CMRR, and a
free RF low-pass filter. I once worked in a club where my remote truck
ground was 65V away from the stage ground... but there was no hum because
that's how good the Jensen splitter transformers were.

>One question about the MPC-1 mic pre-amp schematic, if I may. For the
>+/- 15V the 78L15/79L15 regulators are used. I thought that these
>were quite noisy? I've seen recommenations to use adjustable
>regulators ones instead.

They are very noisy, but it's all high frequency noise so a filter is not
hard. Requires careful layout, though. You can spend a lot more money for
expensive LT regulators if you are tight on space though.
--scott

--
"C'est un Nagra. C'est suisse, et tres, tres precis."

John Hardy

February 18th 17, 11:40 PM

On 2/18/17 1:36 PM, Mat Nieuwenhoven wrote:
> On Fri, 17 Feb 2017 17:24:55 -0600, John Hardy wrote:
>
>> On 2/17/17 11:58 AM, Mat Nieuwenhoven wrote:
>>> On 17 Feb 2017 09:26:36 -0500, Scott Dorsey wrote:
>>
>>>
>>> The capacity of some capacitors (especially multi layer ceramic) is
>>> dependent on the voltage across them; in some cases the value gets
>>> halved! You don't want such in an audio path if the audio voltage is
>>> a significant part of the blocking voltage (if any). See
>>> http://www.eetimes.com/author.asp?section_id=30&doc_id=1330877& .
>>>
>>> Also
>>> http://www.intersil.com/content/dam/Intersil/documents/an13/an1325.pdf
>>> has a nice table of different capacitor types and their trade-offs.
>>>
>>> Mat Nieuwenhoven
>>>
>>
>> Both of those references seem to discuss the shortcomings of the
>> crappier ceramic capacitors, with just a passing reference to the
>> premium ceramic capacitors known as the "COG" or "NP0" types. The
>> COG/NP0 type deserves special consideration. If anyone is interested,
>> page 8 of my 990 data package describes some of the differences between
>> the three most common types of ceramic capacitors, the COG/NP0, X7R and Z5U.
>>
>> http://www.johnhardyco.com/pdf/990.pdf
>
> Very interesting document, thanks. Indeed, the COG/NP0 caps are fine
> in this respect.
>
> Are transformer-based mic amps still used? I can see that a
> transformer-based gain is essentially noise-free, but aren't they
> sensitive to microphone impedance?
>
> One question about the MPC-1 mic pre-amp schematic, if I may. For the
> +/- 15V the 78L15/79L15 regulators are used. I thought that these
> were quite noisy? I've seen recommenations to use adjustable
> regulators ones instead.
>
> Mat Nieuwenhoven
>

The 78L15A and 79L15A are only used for the DC servo op-amp, which is
now the OP97FP. The regulators do not add any noise to that circuit. The
main regulators for the +/-24V power supplies for the 990 op-amps are
LM317 and LM337 with lots of filtering after the regulators, 1000uF per
side on the power supply card and 1000uF per side on each mic preamp card.

Deane Jensen designed the 990 to have very low noise when dealing with
low source impedances. Here is an excerpt from the JE-990 paper that
Deane wrote:

=======
Its application may be considered where some of these parameters are to
be improved:
1) Input stage for any application where the source impedance is 2500
ohms or less,
2) Line output amplifier for driving a 75 ohm load up to an rms
voltage level re 0.775 V of +25 dB, which is an rms voltage of 13.8 V
and a peak-to-peak voltage of 39 V,
3) Summing amplifier,
4) Active filters requiring a high degree of stability,
5)Laboratory preamplifier for extending the sensitivity of noise or
distortion measurements.
=======

Contact Jensen for a copy (www.jensen-transformers.com).

I am increasingly emphasizing the importance of the use of the
lowest-ratio mic input transformer with the 990, the Jensen JT-16-B (or
"A"):

http://www.jensen-transformers.com/wp-content/uploads/2014/08/jt-16-a1.pdf

Jensen makes several ratios of mic-input transformers, each one the best
it can be for the ratio that it has. A summary of the specs for those
transformers is here, with links to pdf files for each model:

http://www.jensen-transformers.com/transformers/mic-input/

The laws of physics dictate that the lower the ratio, the better the
transformer will perform: lower distortion, wider bandwidth, linear
phase response over a wider bandwidth. The trade-off is, the low-ratio
transformer provides less voltage gain than a higher-ratio transformer.
I am sure that this is why Deane came up with the two-stage design (two
990 op-amps in series), known as the Jensen Twin Servo 990 Mic Preamp.
The JT-16 input transformer provides 5.7 dB of voltage gain (I'll update
my specs some day). If you need 60 dB of gain for a particular situation
(ribbon mic, etc.), a preamp with a high-ratio transformer such as the
Jensen JT-115K-E which provides 20 dB of voltage gain would require one
op-amp that provides 40 dB of gain to provide a total gain of 60 dB.
With the JT-16 you get 5.7 dB of voltage gain, so a single 990 would
have to provide 54.3 dB of gain to provide a total of 60 dB. The
two-stage design of the Jensen Twin Servo has each of the two 990
op-amps providing 27.15 dB of gain to get to the total of 60 dB of gain.

In terms of overall noise, the combination of the JT-16 mic-input
transformer and the 990 op-amp is about as quiet as you can get. The
typical voltage gain of 5.7 dB would suggest that you only lose 0.3 dB
along the way. The distortion specs are shown in the pdf for the JT-16
and they are quite low at low frequencies. In the world of mic-input
transformers, the JT-16 is as good as it gets.

Also note that when I converted the 990 to surface-mount in 2013 (except
for the output transistors, which remain in the TO-225AA through-hole
package), I changed the two 0.1 uF power supply bypass capacitors to the
COG/NP0 type. The constant current source filter capacitor was also
changed to the COG/NP0 type. Capacitor manufacturers finally introduced
0.1 uF COG/NP0 caps in the 1206 package at a very reasonable price.

Thank you.

John Hardy
The John Hardy Co.
www.johnhardyco.com

Paul[_13_]

February 19th 17, 01:35 AM

On 2/18/2017 1:45 PM, Don Pearce wrote:
> On Sat, 18 Feb 2017 20:36:32 +0100 (CET), "Mat Nieuwenhoven"
> > wrote:
>
>> On Fri, 17 Feb 2017 17:24:55 -0600, John Hardy wrote:
>>
>>> On 2/17/17 11:58 AM, Mat Nieuwenhoven wrote:
>>>> On 17 Feb 2017 09:26:36 -0500, Scott Dorsey wrote:
>>>
>>>>
>>>> The capacity of some capacitors (especially multi layer ceramic) is
>>>> dependent on the voltage across them; in some cases the value gets
>>>> halved! You don't want such in an audio path if the audio voltage is
>>>> a significant part of the blocking voltage (if any). See
>>>> http://www.eetimes.com/author.asp?section_id=30&doc_id=1330877& .
>>>>
>>>> Also
>>>> http://www.intersil.com/content/dam/Intersil/documents/an13/an1325.pdf
>>>> has a nice table of different capacitor types and their trade-offs.
>>>>
>>>> Mat Nieuwenhoven
>>>>
>>>
>>> Both of those references seem to discuss the shortcomings of the
>>> crappier ceramic capacitors, with just a passing reference to the
>>> premium ceramic capacitors known as the "COG" or "NP0" types. The
>>> COG/NP0 type deserves special consideration. If anyone is interested,
>>> page 8 of my 990 data package describes some of the differences between
>>> the three most common types of ceramic capacitors, the COG/NP0, X7R and Z5U.
>>>
>>> http://www.johnhardyco.com/pdf/990.pdf
>>
>> Very interesting document, thanks. Indeed, the COG/NP0 caps are fine
>> in this respect.
>>
>> Are transformer-based mic amps still used? I can see that a
>> transformer-based gain is essentially noise-free, but aren't they
>> sensitive to microphone impedance?
>>
>> One question about the MPC-1 mic pre-amp schematic, if I may. For the
>> +/- 15V the 78L15/79L15 regulators are used. I thought that these
>> were quite noisy? I've seen recommenations to use adjustable
>> regulators ones instead.
>>
>> Mat Nieuwenhoven
>>
>>
>>
> Transformer based gain? It isn't gain - it is just impedance
> transformation. And most transformers have a loss about 1dB, which
> equates to a 1dB added noise figure. If you want a quiet preamp you do
> away with the transformer and design the front end to present a noise
> match to the mic. This will also provide the lowest distortion, which
> transformers won't manage at the low end.
>
> As for regulator noise, if your preamp has even a half-way decent
> PSRR, it simply isn't a factor.

Some interesting reading here:

http://www.ti.com/lit/an/slaa414/slaa414.pdf

gregz

February 19th 17, 08:34 AM

Scott Dorsey > wrote:
> In article >, Paul > wrote:
>>
>> In contrast, you can listen anywhere in front of the Yammies, and
>> although they may sound bright, and have tons of "presence", they never
>> get HARSH like the piezos!
>
> Okay, you saw that 4kc peak on the piezos. Now, assuming you're using a
> 12dB/octave filter, if you crossed over at 8kc then the peak would only be
> 12dB down. You'd probably want to cross over at 16kc in order to really
> control the problem, or use a sharper filter. Which kind of makes it useless.
>
> You'd think maybe you could use a Zobel network like you would with an
> electrodynamic tweeter to control the peak, but really nobody has ever been
> able to make that work well. It might not be minimum phase.
>
> It MIGHT be possible to use an acoustic network in order to deal with the
> problem, but it's hard to do that and not screw up the pattern.
>
> In the 1980s some grad student built a PA speaker system using a horizontal
> array of piezo tweeters and phase-shift networks, which somehow wound up in
> the EE department auditorium at gatech, probably because nobody else wanted
> them. They had oversized Motorola drivers with the worst of the ugliness
> around 1kc, and they were crossed over around 5kc using a conventional
> midrange driver. Even though the plot wasn't so horrible, there was still
> severe harshness due to the nonlinearities.
>
> Now... Jon Dhalquist with the DQ-10 actually did use a piezo tweeter and
> did actually get some benefit from it. But he was crossing them over at
> 18 KHz or so and using them only as a supertweeter in order to add a little
> more air.
> --scott

Arrays tend to emphasize the lower harsh peak. Additive phase. I once cut
off half the horn length. That helped. Use a 2nd order crossover above 8
kHz.

Greg

gregz

February 19th 17, 08:34 AM

Richard Kuschel > wrote:
> On Wednesday, February 15, 2017 at 9:13:13 AM UTC-7, Scott Dorsey wrote:
> Snip
>
>> Now... Jon Dhalquist with the DQ-10 actually did use a piezo tweeter and
>> did actually get some benefit from it. But he was crossing them over at
>> 18 KHz or so and using them only as a supertweeter in order to add a little
>> more air.
>> --scott
>>
>> --
>> "C'est un Nagra. C'est suisse, et tres, tres precis."
>
> A friend on mine had a pair of those Dahlquists and they were a good sounding system.
> I knew that they had piezo tweeters, but wasn't awate that they were crossed that high.
> A piezo will respond to a 40kHz signal but all that is going to do is annoy the dog.

I have a piezo in a mini box, 555 oscillator, used to test animals and
people.

Greg

Paul[_13_]

February 19th 17, 10:17 AM

On 2/18/2017 4:40 PM, John Hardy wrote:
> On 2/18/17 1:36 PM, Mat Nieuwenhoven wrote:
>> On Fri, 17 Feb 2017 17:24:55 -0600, John Hardy wrote:
>>
>>> On 2/17/17 11:58 AM, Mat Nieuwenhoven wrote:
>>>> On 17 Feb 2017 09:26:36 -0500, Scott Dorsey wrote:
>>>
>>>>
>>>> The capacity of some capacitors (especially multi layer ceramic) is
>>>> dependent on the voltage across them; in some cases the value gets
>>>> halved! You don't want such in an audio path if the audio voltage is
>>>> a significant part of the blocking voltage (if any). See
>>>> http://www.eetimes.com/author.asp?section_id=30&doc_id=1330877& .
>>>>
>>>> Also
>>>> http://www.intersil.com/content/dam/Intersil/documents/an13/an1325.pdf
>>>> has a nice table of different capacitor types and their trade-offs.
>>>>
>>>> Mat Nieuwenhoven
>>>>
>>>
>>> Both of those references seem to discuss the shortcomings of the
>>> crappier ceramic capacitors, with just a passing reference to the
>>> premium ceramic capacitors known as the "COG" or "NP0" types. The
>>> COG/NP0 type deserves special consideration. If anyone is interested,
>>> page 8 of my 990 data package describes some of the differences between
>>> the three most common types of ceramic capacitors, the COG/NP0, X7R
>>> and Z5U.
>>>
>>> http://www.johnhardyco.com/pdf/990.pdf
>>
>> Very interesting document, thanks. Indeed, the COG/NP0 caps are fine
>> in this respect.
>>
>> Are transformer-based mic amps still used? I can see that a
>> transformer-based gain is essentially noise-free, but aren't they
>> sensitive to microphone impedance?
>>
>> One question about the MPC-1 mic pre-amp schematic, if I may. For the
>> +/- 15V the 78L15/79L15 regulators are used. I thought that these
>> were quite noisy? I've seen recommenations to use adjustable
>> regulators ones instead.
>>
>> Mat Nieuwenhoven
>>
>
> The 78L15A and 79L15A are only used for the DC servo op-amp, which is
> now the OP97FP. The regulators do not add any noise to that circuit. The
> main regulators for the +/-24V power supplies for the 990 op-amps are
> LM317 and LM337 with lots of filtering after the regulators, 1000uF per
> side on the power supply card and 1000uF per side on each mic preamp card.
>
> Deane Jensen designed the 990 to have very low noise when dealing with
> low source impedances. Here is an excerpt from the JE-990 paper that
> Deane wrote:
>
> =======
> Its application may be considered where some of these parameters are to
> be improved:
> 1) Input stage for any application where the source impedance is 2500
> ohms or less,
> 2) Line output amplifier for driving a 75 ohm load up to an rms
> voltage level re 0.775 V of +25 dB, which is an rms voltage of 13.8 V
> and a peak-to-peak voltage of 39 V,
> 3) Summing amplifier,
> 4) Active filters requiring a high degree of stability,
> 5)Laboratory preamplifier for extending the sensitivity of noise or
> distortion measurements.
> =======
>
> Contact Jensen for a copy (www.jensen-transformers.com).
>
> I am increasingly emphasizing the importance of the use of the
> lowest-ratio mic input transformer with the 990, the Jensen JT-16-B (or
> "A"):
>
> http://www.jensen-transformers.com/wp-content/uploads/2014/08/jt-16-a1.pdf
>
> Jensen makes several ratios of mic-input transformers, each one the best
> it can be for the ratio that it has. A summary of the specs for those
> transformers is here, with links to pdf files for each model:
>
> http://www.jensen-transformers.com/transformers/mic-input/
>
> The laws of physics dictate that the lower the ratio, the better the
> transformer will perform: lower distortion, wider bandwidth, linear
> phase response over a wider bandwidth. The trade-off is, the low-ratio
> transformer provides less voltage gain than a higher-ratio transformer.
> I am sure that this is why Deane came up with the two-stage design (two
> 990 op-amps in series), known as the Jensen Twin Servo 990 Mic Preamp.
> The JT-16 input transformer provides 5.7 dB of voltage gain (I'll update
> my specs some day). If you need 60 dB of gain for a particular situation
> (ribbon mic, etc.), a preamp with a high-ratio transformer such as the
> Jensen JT-115K-E which provides 20 dB of voltage gain would require one
> op-amp that provides 40 dB of gain to provide a total gain of 60 dB.
> With the JT-16 you get 5.7 dB of voltage gain, so a single 990 would
> have to provide 54.3 dB of gain to provide a total of 60 dB. The
> two-stage design of the Jensen Twin Servo has each of the two 990
> op-amps providing 27.15 dB of gain to get to the total of 60 dB of gain.
>
> In terms of overall noise, the combination of the JT-16 mic-input
> transformer and the 990 op-amp is about as quiet as you can get. The
> typical voltage gain of 5.7 dB would suggest that you only lose 0.3 dB
> along the way. The distortion specs are shown in the pdf for the JT-16
> and they are quite low at low frequencies. In the world of mic-input
> transformers, the JT-16 is as good as it gets.
>
> Also note that when I converted the 990 to surface-mount in 2013 (except
> for the output transistors, which remain in the TO-225AA through-hole
> package), I changed the two 0.1 uF power supply bypass capacitors to the
> COG/NP0 type. The constant current source filter capacitor was also
> changed to the COG/NP0 type. Capacitor manufacturers finally introduced
> 0.1 uF COG/NP0 caps in the 1206 package at a very reasonable price.
>
> Thank you.
>

I don't mean to open a big can of worms here but...

What is the current state of the Transformerless Vs. Transformer
mic preamp debate?

Here's what someone said: "Transformers reject RF much better than
transformerless inputs. Transformers exibibit certain non-linear
loading characteristics that some mics like which give a characteristic
sound that you just can't get otherwise (NEVE). Transformers are
generally less transparent. Transformer-less can be more sterile"

Surely the Friis Noise Figure equation applies to any signal
chain, whether RF or audio, but I assume since in actual audio usage,
long XLR cables will be used, so you don't really care too much
about losses in the front end anyways? (this is why cell phone
towers have the pre-amps in the towers, close to the antennas, for
improved signal/noise ratios)

Just curious....

Don Pearce[_3_]

February 19th 17, 10:38 AM

On Sun, 19 Feb 2017 03:17:16 -0700, Paul > wrote:

>On 2/18/2017 4:40 PM, John Hardy wrote:
>> On 2/18/17 1:36 PM, Mat Nieuwenhoven wrote:
>>> On Fri, 17 Feb 2017 17:24:55 -0600, John Hardy wrote:
>>>
>>>> On 2/17/17 11:58 AM, Mat Nieuwenhoven wrote:
>>>>> On 17 Feb 2017 09:26:36 -0500, Scott Dorsey wrote:
>>>>
>>>>>
>>>>> The capacity of some capacitors (especially multi layer ceramic) is
>>>>> dependent on the voltage across them; in some cases the value gets
>>>>> halved! You don't want such in an audio path if the audio voltage is
>>>>> a significant part of the blocking voltage (if any). See
>>>>> http://www.eetimes.com/author.asp?section_id=30&doc_id=1330877& .
>>>>>
>>>>> Also
>>>>> http://www.intersil.com/content/dam/Intersil/documents/an13/an1325.pdf
>>>>> has a nice table of different capacitor types and their trade-offs.
>>>>>
>>>>> Mat Nieuwenhoven
>>>>>
>>>>
>>>> Both of those references seem to discuss the shortcomings of the
>>>> crappier ceramic capacitors, with just a passing reference to the
>>>> premium ceramic capacitors known as the "COG" or "NP0" types. The
>>>> COG/NP0 type deserves special consideration. If anyone is interested,
>>>> page 8 of my 990 data package describes some of the differences between
>>>> the three most common types of ceramic capacitors, the COG/NP0, X7R
>>>> and Z5U.
>>>>
>>>> http://www.johnhardyco.com/pdf/990.pdf
>>>
>>> Very interesting document, thanks. Indeed, the COG/NP0 caps are fine
>>> in this respect.
>>>
>>> Are transformer-based mic amps still used? I can see that a
>>> transformer-based gain is essentially noise-free, but aren't they
>>> sensitive to microphone impedance?
>>>
>>> One question about the MPC-1 mic pre-amp schematic, if I may. For the
>>> +/- 15V the 78L15/79L15 regulators are used. I thought that these
>>> were quite noisy? I've seen recommenations to use adjustable
>>> regulators ones instead.
>>>
>>> Mat Nieuwenhoven
>>>
>>
>> The 78L15A and 79L15A are only used for the DC servo op-amp, which is
>> now the OP97FP. The regulators do not add any noise to that circuit. The
>> main regulators for the +/-24V power supplies for the 990 op-amps are
>> LM317 and LM337 with lots of filtering after the regulators, 1000uF per
>> side on the power supply card and 1000uF per side on each mic preamp card.
>>
>> Deane Jensen designed the 990 to have very low noise when dealing with
>> low source impedances. Here is an excerpt from the JE-990 paper that
>> Deane wrote:
>>
>> =======
>> Its application may be considered where some of these parameters are to
>> be improved:
>> 1) Input stage for any application where the source impedance is 2500
>> ohms or less,
>> 2) Line output amplifier for driving a 75 ohm load up to an rms
>> voltage level re 0.775 V of +25 dB, which is an rms voltage of 13.8 V
>> and a peak-to-peak voltage of 39 V,
>> 3) Summing amplifier,
>> 4) Active filters requiring a high degree of stability,
>> 5)Laboratory preamplifier for extending the sensitivity of noise or
>> distortion measurements.
>> =======
>>
>> Contact Jensen for a copy (www.jensen-transformers.com).
>>
>> I am increasingly emphasizing the importance of the use of the
>> lowest-ratio mic input transformer with the 990, the Jensen JT-16-B (or
>> "A"):
>>
>> http://www.jensen-transformers.com/wp-content/uploads/2014/08/jt-16-a1.pdf
>>
>> Jensen makes several ratios of mic-input transformers, each one the best
>> it can be for the ratio that it has. A summary of the specs for those
>> transformers is here, with links to pdf files for each model:
>>
>> http://www.jensen-transformers.com/transformers/mic-input/
>>
>> The laws of physics dictate that the lower the ratio, the better the
>> transformer will perform: lower distortion, wider bandwidth, linear
>> phase response over a wider bandwidth. The trade-off is, the low-ratio
>> transformer provides less voltage gain than a higher-ratio transformer.
>> I am sure that this is why Deane came up with the two-stage design (two
>> 990 op-amps in series), known as the Jensen Twin Servo 990 Mic Preamp.
>> The JT-16 input transformer provides 5.7 dB of voltage gain (I'll update
>> my specs some day). If you need 60 dB of gain for a particular situation
>> (ribbon mic, etc.), a preamp with a high-ratio transformer such as the
>> Jensen JT-115K-E which provides 20 dB of voltage gain would require one
>> op-amp that provides 40 dB of gain to provide a total gain of 60 dB.
>> With the JT-16 you get 5.7 dB of voltage gain, so a single 990 would
>> have to provide 54.3 dB of gain to provide a total of 60 dB. The
>> two-stage design of the Jensen Twin Servo has each of the two 990
>> op-amps providing 27.15 dB of gain to get to the total of 60 dB of gain.
>>
>> In terms of overall noise, the combination of the JT-16 mic-input
>> transformer and the 990 op-amp is about as quiet as you can get. The
>> typical voltage gain of 5.7 dB would suggest that you only lose 0.3 dB
>> along the way. The distortion specs are shown in the pdf for the JT-16
>> and they are quite low at low frequencies. In the world of mic-input
>> transformers, the JT-16 is as good as it gets.
>>
>> Also note that when I converted the 990 to surface-mount in 2013 (except
>> for the output transistors, which remain in the TO-225AA through-hole
>> package), I changed the two 0.1 uF power supply bypass capacitors to the
>> COG/NP0 type. The constant current source filter capacitor was also
>> changed to the COG/NP0 type. Capacitor manufacturers finally introduced
>> 0.1 uF COG/NP0 caps in the 1206 package at a very reasonable price.
>>
>> Thank you.
>>
>
> I don't mean to open a big can of worms here but...
>
> What is the current state of the Transformerless Vs. Transformer
>mic preamp debate?
>
> Here's what someone said: "Transformers reject RF much better than
>transformerless inputs. Transformers exibibit certain non-linear
>loading characteristics that some mics like which give a characteristic
>sound that you just can't get otherwise (NEVE). Transformers are
>generally less transparent. Transformer-less can be more sterile"
>
> Surely the Friis Noise Figure equation applies to any signal
>chain, whether RF or audio, but I assume since in actual audio usage,
>long XLR cables will be used, so you don't really care too much
>about losses in the front end anyways? (this is why cell phone
>towers have the pre-amps in the towers, close to the antennas, for
>improved signal/noise ratios)
>
> Just curious....
>
>
>

At audio frequencies and impedances that are typically 150 ohms source
to 1500 ohms load, the losses in a mike cable are essentially zero. I
have no idea what an adjective like sterile means in an audio context
unless it is a word of praise - the sound of the performer doesn't get
messed with.

Ignoring the transformer (and CMRR these days is more than good enough
without them), the noise of a preamp is set by two factors - voltage
noise and current noise. They are produced by independent mechanisms,
but sum to a noise power at the input. The effects are represented by
two opposite slopes on a graph of noise vs impedance, and the lowest
noise is where they cross. You choose the preamp devices to put that
crossing at the impedance of the mikes you are designing the preamp
for. In other words, a low noise preamp for a high impedance mike is
totally different to one for a low impedance mike. Voltage noise
inserts more noise power into a low impedance, while current noise
puts more into a high impedance.

d

---
This email has been checked for viruses by Avast antivirus software.
https://www.avast.com/antivirus

Paul[_13_]

February 19th 17, 01:32 PM

On 2/19/2017 3:38 AM, Don Pearce wrote:
> On Sun, 19 Feb 2017 03:17:16 -0700, Paul > wrote:
>
>> On 2/18/2017 4:40 PM, John Hardy wrote:
>>> On 2/18/17 1:36 PM, Mat Nieuwenhoven wrote:
>>>> On Fri, 17 Feb 2017 17:24:55 -0600, John Hardy wrote:
>>>>
>>>>> On 2/17/17 11:58 AM, Mat Nieuwenhoven wrote:
>>>>>> On 17 Feb 2017 09:26:36 -0500, Scott Dorsey wrote:
>>>>>
>>>>>>
>>>>>> The capacity of some capacitors (especially multi layer ceramic) is
>>>>>> dependent on the voltage across them; in some cases the value gets
>>>>>> halved! You don't want such in an audio path if the audio voltage is
>>>>>> a significant part of the blocking voltage (if any). See
>>>>>> http://www.eetimes.com/author.asp?section_id=30&doc_id=1330877& .
>>>>>>
>>>>>> Also
>>>>>> http://www.intersil.com/content/dam/Intersil/documents/an13/an1325.pdf
>>>>>> has a nice table of different capacitor types and their trade-offs.
>>>>>>
>>>>>> Mat Nieuwenhoven
>>>>>>
>>>>>
>>>>> Both of those references seem to discuss the shortcomings of the
>>>>> crappier ceramic capacitors, with just a passing reference to the
>>>>> premium ceramic capacitors known as the "COG" or "NP0" types. The
>>>>> COG/NP0 type deserves special consideration. If anyone is interested,
>>>>> page 8 of my 990 data package describes some of the differences between
>>>>> the three most common types of ceramic capacitors, the COG/NP0, X7R
>>>>> and Z5U.
>>>>>
>>>>> http://www.johnhardyco.com/pdf/990.pdf
>>>>
>>>> Very interesting document, thanks. Indeed, the COG/NP0 caps are fine
>>>> in this respect.
>>>>
>>>> Are transformer-based mic amps still used? I can see that a
>>>> transformer-based gain is essentially noise-free, but aren't they
>>>> sensitive to microphone impedance?
>>>>
>>>> One question about the MPC-1 mic pre-amp schematic, if I may. For the
>>>> +/- 15V the 78L15/79L15 regulators are used. I thought that these
>>>> were quite noisy? I've seen recommenations to use adjustable
>>>> regulators ones instead.
>>>>
>>>> Mat Nieuwenhoven
>>>>
>>>
>>> The 78L15A and 79L15A are only used for the DC servo op-amp, which is
>>> now the OP97FP. The regulators do not add any noise to that circuit. The
>>> main regulators for the +/-24V power supplies for the 990 op-amps are
>>> LM317 and LM337 with lots of filtering after the regulators, 1000uF per
>>> side on the power supply card and 1000uF per side on each mic preamp card.
>>>
>>> Deane Jensen designed the 990 to have very low noise when dealing with
>>> low source impedances. Here is an excerpt from the JE-990 paper that
>>> Deane wrote:
>>>
>>> =======
>>> Its application may be considered where some of these parameters are to
>>> be improved:
>>> 1) Input stage for any application where the source impedance is 2500
>>> ohms or less,
>>> 2) Line output amplifier for driving a 75 ohm load up to an rms
>>> voltage level re 0.775 V of +25 dB, which is an rms voltage of 13.8 V
>>> and a peak-to-peak voltage of 39 V,
>>> 3) Summing amplifier,
>>> 4) Active filters requiring a high degree of stability,
>>> 5)Laboratory preamplifier for extending the sensitivity of noise or
>>> distortion measurements.
>>> =======
>>>
>>> Contact Jensen for a copy (www.jensen-transformers.com).
>>>
>>> I am increasingly emphasizing the importance of the use of the
>>> lowest-ratio mic input transformer with the 990, the Jensen JT-16-B (or
>>> "A"):
>>>
>>> http://www.jensen-transformers.com/wp-content/uploads/2014/08/jt-16-a1.pdf
>>>
>>> Jensen makes several ratios of mic-input transformers, each one the best
>>> it can be for the ratio that it has. A summary of the specs for those
>>> transformers is here, with links to pdf files for each model:
>>>
>>> http://www.jensen-transformers.com/transformers/mic-input/
>>>
>>> The laws of physics dictate that the lower the ratio, the better the
>>> transformer will perform: lower distortion, wider bandwidth, linear
>>> phase response over a wider bandwidth. The trade-off is, the low-ratio
>>> transformer provides less voltage gain than a higher-ratio transformer.
>>> I am sure that this is why Deane came up with the two-stage design (two
>>> 990 op-amps in series), known as the Jensen Twin Servo 990 Mic Preamp.
>>> The JT-16 input transformer provides 5.7 dB of voltage gain (I'll update
>>> my specs some day). If you need 60 dB of gain for a particular situation
>>> (ribbon mic, etc.), a preamp with a high-ratio transformer such as the
>>> Jensen JT-115K-E which provides 20 dB of voltage gain would require one
>>> op-amp that provides 40 dB of gain to provide a total gain of 60 dB.
>>> With the JT-16 you get 5.7 dB of voltage gain, so a single 990 would
>>> have to provide 54.3 dB of gain to provide a total of 60 dB. The
>>> two-stage design of the Jensen Twin Servo has each of the two 990
>>> op-amps providing 27.15 dB of gain to get to the total of 60 dB of gain.
>>>
>>> In terms of overall noise, the combination of the JT-16 mic-input
>>> transformer and the 990 op-amp is about as quiet as you can get. The
>>> typical voltage gain of 5.7 dB would suggest that you only lose 0.3 dB
>>> along the way. The distortion specs are shown in the pdf for the JT-16
>>> and they are quite low at low frequencies. In the world of mic-input
>>> transformers, the JT-16 is as good as it gets.
>>>
>>> Also note that when I converted the 990 to surface-mount in 2013 (except
>>> for the output transistors, which remain in the TO-225AA through-hole
>>> package), I changed the two 0.1 uF power supply bypass capacitors to the
>>> COG/NP0 type. The constant current source filter capacitor was also
>>> changed to the COG/NP0 type. Capacitor manufacturers finally introduced
>>> 0.1 uF COG/NP0 caps in the 1206 package at a very reasonable price.
>>>
>>> Thank you.
>>>
>>
>> I don't mean to open a big can of worms here but...
>>
>> What is the current state of the Transformerless Vs. Transformer
>> mic preamp debate?
>>
>> Here's what someone said: "Transformers reject RF much better than
>> transformerless inputs. Transformers exibibit certain non-linear
>> loading characteristics that some mics like which give a characteristic
>> sound that you just can't get otherwise (NEVE). Transformers are
>> generally less transparent. Transformer-less can be more sterile"
>>
>> Surely the Friis Noise Figure equation applies to any signal
>> chain, whether RF or audio, but I assume since in actual audio usage,
>> long XLR cables will be used, so you don't really care too much
>> about losses in the front end anyways? (this is why cell phone
>> towers have the pre-amps in the towers, close to the antennas, for
>> improved signal/noise ratios)
>>
>> Just curious....
>>
>>
>>
>
> At audio frequencies and impedances that are typically 150 ohms source
> to 1500 ohms load, the losses in a mike cable are essentially zero.

From here:

https://www.bhphotovideo.com/explora/audio/buying-guide/xlr-cable-just-cable-right%3F

"The issue of frequency-response degradation in cables is sometimes
brought up, and while there is a small potential for a cable to do this,
the problem is largely immaterial except when dealing with extremely
long cable lengths. It takes close to four hundred feet of cable to
produce 1 dB of attenuation at 20 kHz (nominal), which in most live
situations, is virtually inaudible."

This I did not know!

Thanks! :)

I
> have no idea what an adjective like sterile means in an audio context
> unless it is a word of praise - the sound of the performer doesn't get
> messed with.
>
> Ignoring the transformer (and CMRR these days is more than good enough
> without them), the noise of a preamp is set by two factors - voltage
> noise and current noise. They are produced by independent mechanisms,
> but sum to a noise power at the input. The effects are represented by
> two opposite slopes on a graph of noise vs impedance, and the lowest
> noise is where they cross. You choose the preamp devices to put that
> crossing at the impedance of the mikes you are designing the preamp
> for. In other words, a low noise preamp for a high impedance mike is
> totally different to one for a low impedance mike. Voltage noise
> inserts more noise power into a low impedance, while current noise
> puts more into a high impedance.
>
> d
>
> ---
> This email has been checked for viruses by Avast antivirus software.
> https://www.avast.com/antivirus
>

Don Pearce[_3_]

February 19th 17, 08:26 PM

On Sun, 19 Feb 2017 06:32:04 -0700, Paul > wrote:

> "The issue of frequency-response degradation in cables is sometimes
>brought up, and while there is a small potential for a cable to do this,
>the problem is largely immaterial except when dealing with extremely
>long cable lengths. It takes close to four hundred feet of cable to
>produce 1 dB of attenuation at 20 kHz (nominal), which in most live
>situations, is virtually inaudible."
>
> This I did not know!
>
> Thanks! :)

This is where simplistic cables models fall over. Four hundred feet of
cable is enough that at 20kHz a real, distributed model will give a
correct answer, but the lumped C/L/C model has failed. Four hundred
feet of typical 300 ohm mike cable connecting a 150 ohm mike to a 1500
ohm preamp will actually result in about 0.25dB RISE at 20kHz, not a
drop. The reason for this is that at high frequency the cable is
starting to act as a transformer, slightly improving the match between
150 ohms and 1500 ohms.

d

---
This email has been checked for viruses by Avast antivirus software.
https://www.avast.com/antivirus

Scott Dorsey

February 19th 17, 08:43 PM

In article >, Paul > wrote:
>https://www.bhphotovideo.com/explora/audio/buying-guide/xlr-cable-just-cable-right%3F
>
> "The issue of frequency-response degradation in cables is sometimes
>brought up, and while there is a small potential for a cable to do this,
>the problem is largely immaterial except when dealing with extremely
>long cable lengths. It takes close to four hundred feet of cable to
>produce 1 dB of attenuation at 20 kHz (nominal), which in most live
>situations, is virtually inaudible."
>
> This I did not know!

This is why audio folks use low-Z balanced systems. That estimate is kind
of silly, though, as you really need to specify the source and load impedances
as well as the cable. With the Schoeps mikes that have a 50 ohm source Z,
you can run longer than 400 ft. before getting 1dB down. Which is good for
classical work where running a thousand feet of cable from suspended mikes
out to the truck is not unusual.

But... there are two cases you will encounter frequently in the studio where
this isn't necessarily the case, and both have to do with high-Z systems.
The first, obviously, is guitar amp cables. Very high impedance line, so
it becomes a huge issue not only the shunt capacitance of the cable but with
some weird cable designs even the series inductance.

But the second one isn't so obvious, and it's the ribbon microphone. Ribbons
have pretty weird source impedances which can be pretty high at some
frequencies, so a lot of people like to use preamps with very high load Z
so that source impedance is not a problem. But, when you do that, the cable
then does start to be a problem, especially cables with a lot of shunt
capacitance like star-quad types.
--scott
--
"C'est un Nagra. C'est suisse, et tres, tres precis."

Phil Allison[_4_]

February 20th 17, 12:55 AM

Don Pearce wrote:

>
>
> This is where simplistic cables models fall over. Four hundred feet of
> cable is enough that at 20kHz a real, distributed model will give a
> correct answer, but the lumped C/L/C model has failed. Four hundred
> feet of typical 300 ohm mike cable ...
>
>

** Who sells "300 ohm mic cable" ??.

Or are you saying typical twisted pair mic cables have a characteristic impedance of 300ohms in the audio range ?

IME, 400 feet of common or garden mic cable will significantly attenuate high frequencies from a mic like the SM58 and most others - assuming there is the usual 1500 ohms load at the other end.

The rated impedance of a mic cable is not defined anywhere I can find, but IMO ought to be the value of terminating resistor that minimises or eliminate shunt capacitance in and somewhat beyond the audio range.

Users would at least then know how to get the best HF response with long runs.

..... Phil

connecting a 150 ohm mike to a 1500
> ohm preamp will actually result in about 0.25dB RISE at 20kHz, not a
> drop. The reason for this is that at high frequency the cable is
> starting to act as a transformer, slightly improving the match between
> 150 ohms and 1500 ohms.
>
>

Paul[_13_]

February 20th 17, 01:17 AM

On 2/19/2017 5:55 PM, Phil Allison wrote:
> Don Pearce wrote:
>
>>
>>
>> This is where simplistic cables models fall over. Four hundred feet of
>> cable is enough that at 20kHz a real, distributed model will give a
>> correct answer, but the lumped C/L/C model has failed. Four hundred
>> feet of typical 300 ohm mike cable ...
>>
>>
>
>
> ** Who sells "300 ohm mic cable" ??.
>
> Or are you saying typical twisted pair mic cables have a characteristic impedance of 300ohms in the audio range ?
>
> IME, 400 feet of common or garden mic cable will significantly attenuate high frequencies from a mic like the SM58 and most others - assuming there is the usual 1500 ohms load at the other end.
>
> The rated impedance of a mic cable is not defined anywhere I can find, but IMO ought to be the value of terminating resistor that minimises or eliminate shunt capacitance in and somewhat beyond the audio range.
>

Time Domain Reflectometry (TDR) is used in telecommunications to
find impedance discontinuities in cables.

RF engineers use the stand wave ratio (SWR) to find the least
amount of reflected power with a given termination load.

Not sure if audio frequency people use these techniques....

>
>
>
>
>
>
> connecting a 150 ohm mike to a 1500
>> ohm preamp will actually result in about 0.25dB RISE at 20kHz, not a
>> drop. The reason for this is that at high frequency the cable is
>> starting to act as a transformer, slightly improving the match between
>> 150 ohms and 1500 ohms.
>>
>>

Scott Dorsey

February 20th 17, 01:18 AM

Phil Allison > wrote:
>Don Pearce wrote:
>>
>> This is where simplistic cables models fall over. Four hundred feet of
>> cable is enough that at 20kHz a real, distributed model will give a
>> correct answer, but the lumped C/L/C model has failed. Four hundred
>> feet of typical 300 ohm mike cable ...
>
>** Who sells "300 ohm mic cable" ??.
>
>Or are you saying typical twisted pair mic cables have a characteristic impedance of 300ohms in the audio range ?

Actually, typical 24 ga mike cable tends to run around 110 ohms or so, which
is why that was picked as the impedance for AES/EBU, since they wanted to be
able to run AES/EBU over existing cable plant.

Older 18 ga mike cable with thicker rubber comes in around 150 ohms.

If you wanted to get up to 300 ohms you'd have to separate the conductors
more, or make them tiny.

>IME, 400 feet of common or garden mic cable will significantly attenuate high frequencies from a mic like the SM58 and most others - assuming there is the usual 1500 ohms load at the other end.

If you're using an SM58, you probably want a load of around 600 ohms in order
to make it happy. If you're using a condenser microphone, the output Z is
likely far lower than the load and so if you're thinking about it as a lumped
sum system with the impedances in parallel the source impedance is dominant.

The loading issues are more of an issue than the cable issues.

>The rated impedance of a mic cable is not defined anywhere I can find, but IMO ought to be the value of terminating resistor that minimises or eliminate shunt capacitance in and somewhat beyond the audio range.
>
>Users would at least then know how to get the best HF response with long runs.

The problem is that microphone noise and microphone dynamic behaviour due to
electrical damping (in the case of dynamic mikes) are usually much more
important than any HF behaviour.

Still, the HF behaviour gets more interesting with cables like star quad that
have far higher shunt capacitance without any change in series inductance.

I'd like to see a model of the system. Shouldn't be hard to do, and I would
not be surprised to see an HF peak appear in the top octave if you get the
loading right for the cable.
--scott

--
"C'est un Nagra. C'est suisse, et tres, tres precis."

Scott Dorsey

February 20th 17, 02:07 AM

In article >, Paul > wrote:
> RF engineers use the stand wave ratio (SWR) to find the least
>amount of reflected power with a given termination load.
>
> Not sure if audio frequency people use these techniques....

They do, only when cable lengths are long enough for transmission line effects
to be significant. Which is to say many miles.

Back in the days of the analogue telephone plant, looking at long distance
lines with a tdr would let you see discontinuities like loading coils and
line damage. These days you seldom see even telephone loops more than a
few miles before the signal is digitized.

Most of the time audio people view cables as lumped-sum networks because
cables are far shorter than a quarter-wave.

And, as Phil notes, microphone cables aren't even specified for characteristic
impedance, and usually the diameter and dielectric thickness aren't controlled
as tightly as a cable intended to be constant-Z.
--scott

--
"C'est un Nagra. C'est suisse, et tres, tres precis."

Phil Allison[_4_]

February 20th 17, 02:07 AM

Scott Dorsey wrote:
> Phil Allison
> >Don Pearce wrote:
> >>
> >> This is where simplistic cables models fall over. Four hundred feet of
> >> cable is enough that at 20kHz a real, distributed model will give a
> >> correct answer, but the lumped C/L/C model has failed. Four hundred
> >> feet of typical 300 ohm mike cable ...
> >
> >** Who sells "300 ohm mic cable" ??.
> >
> >Or are you saying typical twisted pair mic cables have a characteristic impedance of 300ohms in the audio range ?
>
> Actually, typical 24 ga mike cable tends to run around 110 ohms or so, which
> is why that was picked as the impedance for AES/EBU, since they wanted to be
> able to run AES/EBU over existing cable plant.
>
> Older 18 ga mike cable with thicker rubber comes in around 150 ohms.
>
> If you wanted to get up to 300 ohms you'd have to separate the conductors
> more, or make them tiny.
>
> >IME, 400 feet of common or garden mic cable will significantly
> attenuate high frequencies from a mic like the SM58 and most others -
> assuming there is the usual 1500 ohms load at the other end.
>
> If you're using an SM58, you probably want a load of around 600 ohms in order
> to make it happy. If you're using a condenser microphone, the output Z is
> likely far lower than the load
>

** But not very likely - plenty of condenser mics have 250ohm or so impedance.

> The loading issues are more of an issue than the cable issues.
>

** They are inseparable.

> The rated impedance of a mic cable is not defined anywhere I can find,
> but IMO ought to be the value of terminating resistor that minimises
> or eliminate shunt capacitance in and somewhat beyond the audio range.
>
> Users would at least then know how to get the best HF response with
> long runs.
>
>
> The problem is that microphone noise and microphone dynamic behaviour due to
> electrical damping (in the case of dynamic mikes) are usually much more
> important than any HF behaviour.
>

** Sorry, that is gobbldegook.

Alluding to your superior knowledge of god knows what is a sure way to **** readers off. Being very specific is far more useful.

..... Phil

Scott Dorsey

February 20th 17, 02:22 AM

Phil Allison > wrote:
>>
>> The problem is that microphone noise and microphone dynamic behaviour due to
>> electrical damping (in the case of dynamic mikes) are usually much more
>> important than any HF behaviour.
>
>** Sorry, that is gobbldegook.

Okay, here we go....

A lot of dynamic microphones, most especially the SM-58, rely upon the
load to provide mechanical damping of the capsule.

If you operate them into a high-Z load, you see all kinds of ringing and
overshoot, because the diaphragm becomes a poorly damped mass-spring system.

These microphones need a fairly low impedance load in order to avoid ringing.
That being the case, the load impedance across the cable means that cable
effects are pretty heavily minimized.

> Alluding to your superior knowledge of god knows what is a sure way to **** readers off. Being very specific is far more useful.

This isn't any superior knowledge, this is pretty commonly known. It is
one of the reasons why people praise transformer-input preamps without really
understanding why they are getting the sound they are, however.

On the other hand, that ringing on an unloaded or lightly loaded SM-57 is
a useful thing on snare mikes, to the point where people have removed the
step-up transformer from some SM-57s to accentuate the effect.
--scott

--
"C'est un Nagra. C'est suisse, et tres, tres precis."

Phil Allison[_4_]

February 20th 17, 05:19 AM

Scott Dorsey wrote:

> Phil Allison
> >>
> >> The problem is that microphone noise and microphone dynamic behaviour
> >> due to
> >> electrical damping (in the case of dynamic mikes) are usually much more
> >> important than any HF behaviour.
> >
> >** Sorry, that is gobbldegook.
>
> Okay, here we go....
>
> A lot of dynamic microphones, most especially the SM-58, rely upon the
> load to provide mechanical damping of the capsule.
>

** Resistance loading on the output *only* affects the LF resonance of the diaphragm and even then not by much.

> If you operate them into a high-Z load, you see all kinds of ringing and
> overshoot, because the diaphragm becomes a poorly damped mass-spring system.
>

** That is nonsense.

Capsule inductance combined with the leakage inductance of the internal transformer plus any cable capacitance form a network that can ring at specific frequencies - normally supersonic ones.

Adding the right value load resistance will damp this nicely. If Shure thought it valuable, they could easily add a resistor ( or RC combo) inside the mic.

> These microphones need a fairly low impedance load in order to avoid ringing.
> That being the case, the load impedance across the cable means that cable
> effects are pretty heavily minimized.
>

** But we are considering *unusually long* runs where cable capacitance can be a devil.

> > Alluding to your superior knowledge of god knows what is a sure way to **** readers off. Being very specific is far more useful.
>
> This isn't any superior knowledge, this is pretty commonly known.
>

** But instead of posting like you have now, you *alluded* to it.

..... Phil

Don Pearce[_3_]

February 20th 17, 05:46 AM

On Sun, 19 Feb 2017 16:55:31 -0800 (PST), Phil Allison
> wrote:

>Don Pearce wrote:
>
>>
>>
>> This is where simplistic cables models fall over. Four hundred feet of
>> cable is enough that at 20kHz a real, distributed model will give a
>> correct answer, but the lumped C/L/C model has failed. Four hundred
>> feet of typical 300 ohm mike cable ...
>>
>>
>
>
>** Who sells "300 ohm mic cable" ??.
>
>Or are you saying typical twisted pair mic cables have a characteristic impedance of 300ohms in the audio range ?
>
>IME, 400 feet of common or garden mic cable will significantly attenuate high frequencies from a mic like the SM58 and most others - assuming there is the usual 1500 ohms load at the other end.
>
>The rated impedance of a mic cable is not defined anywhere I can find, but IMO ought to be the value of terminating resistor that minimises or eliminate shunt capacitance in and somewhat beyond the audio range.
>
>Users would at least then know how to get the best HF response with long runs.
>
>
I've measured the impedance of a few cables. They came out in the
range of about 220 to s40 ohms. 300 seemed a reasonable average for
the calculation.

d

---
This email has been checked for viruses by Avast antivirus software.
https://www.avast.com/antivirus

geoff

February 20th 17, 05:59 AM

On 20/02/2017 6:19 PM, Phil Allison wrote:

>
> Adding the right value load resistance will damp this nicely. If
> Shure thought it valuable, they could easily add a resistor ( or RC
> combo) inside the mic.

They provide a specified load Z , 500 Ohms. They cannot control what it
is being plugged into, so surely best to leave any extra loading to
compensate for higher input impedances, damping, etc, to the user ?

geoff

Don Pearce[_3_]

February 20th 17, 06:57 AM

On Mon, 20 Feb 2017 05:46:21 GMT, (Don Pearce) wrote:

>On Sun, 19 Feb 2017 16:55:31 -0800 (PST), Phil Allison
> wrote:
>
>>Don Pearce wrote:
>>
>>>
>>>
>>> This is where simplistic cables models fall over. Four hundred feet of
>>> cable is enough that at 20kHz a real, distributed model will give a
>>> correct answer, but the lumped C/L/C model has failed. Four hundred
>>> feet of typical 300 ohm mike cable ...
>>>
>>>
>>
>>
>>** Who sells "300 ohm mic cable" ??.
>>
>>Or are you saying typical twisted pair mic cables have a characteristic impedance of 300ohms in the audio range ?
>>
>>IME, 400 feet of common or garden mic cable will significantly attenuate high frequencies from a mic like the SM58 and most others - assuming there is the usual 1500 ohms load at the other end.
>>
>>The rated impedance of a mic cable is not defined anywhere I can find, but IMO ought to be the value of terminating resistor that minimises or eliminate shunt capacitance in and somewhat beyond the audio range.
>>
>>Users would at least then know how to get the best HF response with long runs.
>>
>>
>I've measured the impedance of a few cables. They came out in the
>range of about 220 to s40 ohms. 300 seemed a reasonable average for
>the calculation.
>
>d
>
>---
>This email has been checked for viruses by Avast antivirus software.
>https://www.avast.com/antivirus

Here are a few calculations on the behaviour of longish cables in a
microphone scenario. It is interesting that at any frequency where the
cable is starting to have an effect, a calculation using just
capacitance is wrong. It is not only the wrong answer, but the answer
is backwards, showing a loss where there is in fact a gain. Anyways,
have a look and see what actually happens.

http://www.soundthoughts.co.uk/read/cable.html

d

---
This email has been checked for viruses by Avast antivirus software.
https://www.avast.com/antivirus

Paul[_13_]

February 20th 17, 10:25 AM

On 2/19/2017 11:57 PM, Don Pearce wrote:
> On Mon, 20 Feb 2017 05:46:21 GMT, (Don Pearce) wrote:
>
>> On Sun, 19 Feb 2017 16:55:31 -0800 (PST), Phil Allison
>> > wrote:
>>
>>> Don Pearce wrote:
>>>
>>>>
>>>>
>>>> This is where simplistic cables models fall over. Four hundred feet of
>>>> cable is enough that at 20kHz a real, distributed model will give a
>>>> correct answer, but the lumped C/L/C model has failed. Four hundred
>>>> feet of typical 300 ohm mike cable ...
>>>>
>>>>
>>>
>>>
>>> ** Who sells "300 ohm mic cable" ??.
>>>
>>> Or are you saying typical twisted pair mic cables have a characteristic impedance of 300ohms in the audio range ?
>>>
>>> IME, 400 feet of common or garden mic cable will significantly attenuate high frequencies from a mic like the SM58 and most others - assuming there is the usual 1500 ohms load at the other end.
>>>
>>> The rated impedance of a mic cable is not defined anywhere I can find, but IMO ought to be the value of terminating resistor that minimises or eliminate shunt capacitance in and somewhat beyond the audio range.
>>>
>>> Users would at least then know how to get the best HF response with long runs.
>>>
>>>
>> I've measured the impedance of a few cables. They came out in the
>> range of about 220 to s40 ohms. 300 seemed a reasonable average for
>> the calculation.
>>
>> d
>>
>> ---
>> This email has been checked for viruses by Avast antivirus software.
>> https://www.avast.com/antivirus
>
> Here are a few calculations on the behaviour of longish cables in a
> microphone scenario. It is interesting that at any frequency where the
> cable is starting to have an effect, a calculation using just
> capacitance is wrong. It is not only the wrong answer, but the answer
> is backwards, showing a loss where there is in fact a gain. Anyways,
> have a look and see what actually happens.
>
> http://www.soundthoughts.co.uk/read/cable.html
>

Interesting, but where is the series resistance and series
inductance per unit length in your lumped element model?

And where is the shunt conductance G per unit length in the
admittance? You only have the shunt capacitance.

Even if you assume a lossless cable, and R=0 and G=0, you
would still have the series inductance per unit length.

And what is being used for the distributed model? Some
sort of finite element analysis, using numerical methods?

Don Pearce[_3_]

February 20th 17, 10:52 AM

On Mon, 20 Feb 2017 03:25:42 -0700, Paul > wrote:

>On 2/19/2017 11:57 PM, Don Pearce wrote:
>> On Mon, 20 Feb 2017 05:46:21 GMT, (Don Pearce) wrote:
>>
>>> On Sun, 19 Feb 2017 16:55:31 -0800 (PST), Phil Allison
>>> > wrote:
>>>
>>>> Don Pearce wrote:
>>>>
>>>>>
>>>>>
>>>>> This is where simplistic cables models fall over. Four hundred feet of
>>>>> cable is enough that at 20kHz a real, distributed model will give a
>>>>> correct answer, but the lumped C/L/C model has failed. Four hundred
>>>>> feet of typical 300 ohm mike cable ...
>>>>>
>>>>>
>>>>
>>>>
>>>> ** Who sells "300 ohm mic cable" ??.
>>>>
>>>> Or are you saying typical twisted pair mic cables have a characteristic impedance of 300ohms in the audio range ?
>>>>
>>>> IME, 400 feet of common or garden mic cable will significantly attenuate high frequencies from a mic like the SM58 and most others - assuming there is the usual 1500 ohms load at the other end.
>>>>
>>>> The rated impedance of a mic cable is not defined anywhere I can find, but IMO ought to be the value of terminating resistor that minimises or eliminate shunt capacitance in and somewhat beyond the audio range.
>>>>
>>>> Users would at least then know how to get the best HF response with long runs.
>>>>
>>>>
>>> I've measured the impedance of a few cables. They came out in the
>>> range of about 220 to s40 ohms. 300 seemed a reasonable average for
>>> the calculation.
>>>
>>> d
>>>
>>> ---
>>> This email has been checked for viruses by Avast antivirus software.
>>> https://www.avast.com/antivirus
>>
>> Here are a few calculations on the behaviour of longish cables in a
>> microphone scenario. It is interesting that at any frequency where the
>> cable is starting to have an effect, a calculation using just
>> capacitance is wrong. It is not only the wrong answer, but the answer
>> is backwards, showing a loss where there is in fact a gain. Anyways,
>> have a look and see what actually happens.
>>
>> http://www.soundthoughts.co.uk/read/cable.html
>>
>
> Interesting, but where is the series resistance and series
>inductance per unit length in your lumped element model?
>
> And where is the shunt conductance G per unit length in the
>admittance? You only have the shunt capacitance.
>
> Even if you assume a lossless cable, and R=0 and G=0, you
>would still have the series inductance per unit length.
>
> And what is being used for the distributed model? Some
>sort of finite element analysis, using numerical methods?
>
>

I used just capacitance and source resistance, because that is what
people go to when they try to calculate the high frequency limit of a
cable. I could use a C-L-C model to make it appear more accurate, but
in fact it makes things worse. What you end up with is a multiple pole
lowpass filter.

The distributed model is the standard Spice transmission line - which
almost all RF simulators use. The parameters it needs are delay and
impedance.

I currently use normal coax cable at up to 60GHz (60,000,000,00 Hz) so
it is clear that no cable has any kind of upper frequency limit -
until it starts moding. That is when the inner diameter of the cable
approximates a half wavelength.

d

---
This email has been checked for viruses by Avast antivirus software.
https://www.avast.com/antivirus

Don Pearce[_3_]

February 20th 17, 11:18 AM

On Mon, 20 Feb 2017 10:52:34 GMT, (Don Pearce) wrote:

>On Mon, 20 Feb 2017 03:25:42 -0700, Paul > wrote:
>
>>On 2/19/2017 11:57 PM, Don Pearce wrote:
>>> On Mon, 20 Feb 2017 05:46:21 GMT, (Don Pearce) wrote:
>>>
>>>> On Sun, 19 Feb 2017 16:55:31 -0800 (PST), Phil Allison
>>>> > wrote:
>>>>
>>>>> Don Pearce wrote:
>>>>>
>>>>>>
>>>>>>
>>>>>> This is where simplistic cables models fall over. Four hundred feet of
>>>>>> cable is enough that at 20kHz a real, distributed model will give a
>>>>>> correct answer, but the lumped C/L/C model has failed. Four hundred
>>>>>> feet of typical 300 ohm mike cable ...
>>>>>>
>>>>>>
>>>>>
>>>>>
>>>>> ** Who sells "300 ohm mic cable" ??.
>>>>>
>>>>> Or are you saying typical twisted pair mic cables have a characteristic impedance of 300ohms in the audio range ?
>>>>>
>>>>> IME, 400 feet of common or garden mic cable will significantly attenuate high frequencies from a mic like the SM58 and most others - assuming there is the usual 1500 ohms load at the other end.
>>>>>
>>>>> The rated impedance of a mic cable is not defined anywhere I can find, but IMO ought to be the value of terminating resistor that minimises or eliminate shunt capacitance in and somewhat beyond the audio range.
>>>>>
>>>>> Users would at least then know how to get the best HF response with long runs.
>>>>>
>>>>>
>>>> I've measured the impedance of a few cables. They came out in the
>>>> range of about 220 to s40 ohms. 300 seemed a reasonable average for
>>>> the calculation.
>>>>
>>>> d
>>>>
>>>> ---
>>>> This email has been checked for viruses by Avast antivirus software.
>>>> https://www.avast.com/antivirus
>>>
>>> Here are a few calculations on the behaviour of longish cables in a
>>> microphone scenario. It is interesting that at any frequency where the
>>> cable is starting to have an effect, a calculation using just
>>> capacitance is wrong. It is not only the wrong answer, but the answer
>>> is backwards, showing a loss where there is in fact a gain. Anyways,
>>> have a look and see what actually happens.
>>>
>>> http://www.soundthoughts.co.uk/read/cable.html
>>>
>>
>> Interesting, but where is the series resistance and series
>>inductance per unit length in your lumped element model?
>>
>> And where is the shunt conductance G per unit length in the
>>admittance? You only have the shunt capacitance.
>>
>> Even if you assume a lossless cable, and R=0 and G=0, you
>>would still have the series inductance per unit length.
>>
>> And what is being used for the distributed model? Some
>>sort of finite element analysis, using numerical methods?
>>
>>
>
>I used just capacitance and source resistance, because that is what
>people go to when they try to calculate the high frequency limit of a
>cable. I could use a C-L-C model to make it appear more accurate, but
>in fact it makes things worse. What you end up with is a multiple pole
>lowpass filter.
>
>The distributed model is the standard Spice transmission line - which
>almost all RF simulators use. The parameters it needs are delay and
>impedance.
>
>I currently use normal coax cable at up to 60GHz (60,000,000,00 Hz) so
>it is clear that no cable has any kind of upper frequency limit -
>until it starts moding. That is when the inner diameter of the cable
>approximates a half wavelength.
>
I've just made a lumped/distributed model that uses 500 L-C-L elements
in series, and it reproduces the gain and cyclic variation very nicely
up to 1MHz. Above that it turns flaky.

I'm not bothering with admittance and conductance. For a low impedance
speaker cable they become important, but for this they are irrelevant.

d

---
This email has been checked for viruses by Avast antivirus software.
https://www.avast.com/antivirus

geoff

February 20th 17, 07:05 PM

On 20/02/2017 11:52 PM, Don Pearce wrote:
> On Mon, 20 Feb 2017 03:25:42 -0700, Paul > wrote:
>
> I currently use normal coax cable at up to 60GHz (60,000,000,00 Hz) so
> it is clear that no cable has any kind of upper frequency limit -
> until it starts moding. That is when the inner diameter of the cable
> approximates a half wavelength.

Jeeeze. 60gHz. My hearing isn't that good any more, but your bat must be
really ****ed off !

geoff

Don Pearce[_3_]

February 20th 17, 07:29 PM

On Tue, 21 Feb 2017 08:05:07 +1300, geoff >
wrote:

>On 20/02/2017 11:52 PM, Don Pearce wrote:
>> On Mon, 20 Feb 2017 03:25:42 -0700, Paul > wrote:
>>
>> I currently use normal coax cable at up to 60GHz (60,000,000,00 Hz) so
>> it is clear that no cable has any kind of upper frequency limit -
>> until it starts moding. That is when the inner diameter of the cable
>> approximates a half wavelength.
>
>Jeeeze. 60gHz. My hearing isn't that good any more, but your bat must be
>really ****ed off !
>
Yeah, I can't hear that when I have a head cold.

d

---
This email has been checked for viruses by Avast antivirus software.
https://www.avast.com/antivirus

Mat Nieuwenhoven

February 20th 17, 09:48 PM

On Sat, 18 Feb 2017 17:40:05 -0600, John Hardy wrote:

<snip>
>> Are transformer-based mic amps still used? I can see that a
>> transformer-based gain is essentially noise-free, but aren't they
>> sensitive to microphone impedance?
>>
>> One question about the MPC-1 mic pre-amp schematic, if I may. For the
>> +/- 15V the 78L15/79L15 regulators are used. I thought that these
>> were quite noisy? I've seen recommenations to use adjustable
>> regulators ones instead.
>>
>> Mat Nieuwenhoven
>>
>
>The 78L15A and 79L15A are only used for the DC servo op-amp, which is
>now the OP97FP. The regulators do not add any noise to that circuit. The
>main regulators for the +/-24V power supplies for the 990 op-amps are
>LM317 and LM337 with lots of filtering after the regulators, 1000uF per
>side on the power supply card and 1000uF per side on each mic preamp card.
>
>Deane Jensen designed the 990 to have very low noise when dealing with
>low source impedances. Here is an excerpt from the JE-990 paper that
>Deane wrote:
>
>=======
>Its application may be considered where some of these parameters are to
>be improved:
> 1) Input stage for any application where the source impedance is 2500
>ohms or less,
> 2) Line output amplifier for driving a 75 ohm load up to an rms
>voltage level re 0.775 V of +25 dB, which is an rms voltage of 13.8 V
>and a peak-to-peak voltage of 39 V,
> 3) Summing amplifier,
> 4) Active filters requiring a high degree of stability,
> 5)Laboratory preamplifier for extending the sensitivity of noise or
>distortion measurements.
>=======
>
>Contact Jensen for a copy (www.jensen-transformers.com).
>
>I am increasingly emphasizing the importance of the use of the
>lowest-ratio mic input transformer with the 990, the Jensen JT-16-B (or
>"A"):
>
>http://www.jensen-transformers.com/wp-content/uploads/2014/08/jt-16-a1.pdf
>
>Jensen makes several ratios of mic-input transformers, each one the best
>it can be for the ratio that it has. A summary of the specs for those
>transformers is here, with links to pdf files for each model:
>
>http://www.jensen-transformers.com/transformers/mic-input/
>
>The laws of physics dictate that the lower the ratio, the better the
>transformer will perform: lower distortion, wider bandwidth, linear
>phase response over a wider bandwidth. The trade-off is, the low-ratio
>transformer provides less voltage gain than a higher-ratio transformer.
>I am sure that this is why Deane came up with the two-stage design (two
>990 op-amps in series), known as the Jensen Twin Servo 990 Mic Preamp.
>The JT-16 input transformer provides 5.7 dB of voltage gain (I'll update
>my specs some day). If you need 60 dB of gain for a particular situation
>(ribbon mic, etc.), a preamp with a high-ratio transformer such as the
>Jensen JT-115K-E which provides 20 dB of voltage gain would require one
>op-amp that provides 40 dB of gain to provide a total gain of 60 dB.
>With the JT-16 you get 5.7 dB of voltage gain, so a single 990 would
>have to provide 54.3 dB of gain to provide a total of 60 dB. The
>two-stage design of the Jensen Twin Servo has each of the two 990
>op-amps providing 27.15 dB of gain to get to the total of 60 dB of gain.
>
>In terms of overall noise, the combination of the JT-16 mic-input
>transformer and the 990 op-amp is about as quiet as you can get. The
>typical voltage gain of 5.7 dB would suggest that you only lose 0.3 dB
>along the way. The distortion specs are shown in the pdf for the JT-16
>and they are quite low at low frequencies. In the world of mic-input
>transformers, the JT-16 is as good as it gets.
>
>Also note that when I converted the 990 to surface-mount in 2013 (except
>for the output transistors, which remain in the TO-225AA through-hole
>package), I changed the two 0.1 uF power supply bypass capacitors to the
>COG/NP0 type. The constant current source filter capacitor was also
>changed to the COG/NP0 type. Capacitor manufacturers finally introduced
>0.1 uF COG/NP0 caps in the 1206 package at a very reasonable price.

Thanks for taking the effort to explain all this. Most of it was new
to me, and it's interesting to learn about the technology behind the
audio.

Very impressive, these Jensen transformers.

Mat Nieuwenhoven