Which caps do you folks recommend for dc blocking in mic pres? I built
one using large polypropylene caps, and would like to do one using
smaller parts.

Panasonic SU non polars?
Multiple film caps in parallel?

Thanks in advance

Roger Foote

R. Foote wrote:
Which caps do you folks recommend for dc blocking in mic pres? I built
one using large polypropylene caps, and would like to do one using
smaller parts.

Depends entirely on how small they need to be. I assume we are talking
about phantom power blocking on the front end? Any other blocking caps
inside are going to depend completely on the technology of the active
electronics you are using.

Panasonic SU non polars?
Multiple film caps in parallel?

I can recommend the Panasonic FC series caps if you have to go electrolytic.
Maybe with a film bypass cap if there is enough room. 20 uF of the Panasonic
Mylar caps, though, will sound better than an electrolytic and won't take up
quite as much room as polypropylene.

There is no reason to go nonpolar, since there is considerable DC offset.
--scott
--
"C'est un Nagra. C'est suisse, et tres, tres precis."

www.solen.ca

Solen, manufactures one of the best polypropylene capacitors in my opinion.
They are used by many loudspeaker manufactures, and they have a great web
page.


Marko.



"R. Foote" wrote in message
om...
Which caps do you folks recommend for dc blocking in mic pres? I built
one using large polypropylene caps, and would like to do one using
smaller parts.

Panasonic SU non polars?
Multiple film caps in parallel?

Thanks in advance

Roger Foote

Marko wrote:
www.solen.ca

Solen, manufactures one of the best polypropylene capacitors in my opinion.
They are used by many loudspeaker manufactures, and they have a great web
page.

Unfortunately a 100 uF dc blocking cap in polypropylene is about the size
of a Mack truck.
--scott


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

Scott Dorsey wrote in message
...
Marko wrote:
www.solen.ca

Solen, manufactures one of the best polypropylene capacitors in my
opinion.
They are used by many loudspeaker manufactures, and they have a great web
page.

Unfortunately a 100 uF dc blocking cap in polypropylene is about the size
of a Mack truck.
--scott

Yeah, but if you're designing the thing yourself you can do it so as not to
need 100uF. Make the resistance on the right-hand side of the cap 10k and
you can get away with 10uF. That's doable in polypropylene.

Peace,
Paul

In article ,
Paul Stamler wrote:

Scott Dorsey wrote in message
...
Marko wrote:
www.solen.ca

Solen, manufactures one of the best polypropylene capacitors in my
opinion.
They are used by many loudspeaker manufactures, and they have a great web
page.

Unfortunately a 100 uF dc blocking cap in polypropylene is about the size
of a Mack truck.

Yeah, but if you're designing the thing yourself you can do it so as not to
need 100uF. Make the resistance on the right-hand side of the cap 10k and
you can get away with 10uF. That's doable in polypropylene.

That's where the rub is. You want a low input impedance on your preamp,
so that it properly loads dynamic microphones. BUT, you want a high input
impedance, so your phantom blocking caps aren't enormous.

With a 600 ohm input Z, 10 uF gives you a -3dB point at 20 Hz or so, which
is kind of high, I think. But not insane.
--scott


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

"Paul Stamler" wrote in message

Scott Dorsey wrote in message
...
Marko wrote:
www.solen.ca

Solen, manufactures one of the best polypropylene capacitors in my
opinion. They are used by many loudspeaker manufactures, and they
have a great web page.

Unfortunately a 100 uF dc blocking cap in polypropylene is about the
size of a Mack truck.
--scott

Yeah, but if you're designing the thing yourself you can do it so as
not to need 100uF. Make the resistance on the right-hand side of the
cap 10k and you can get away with 10uF. That's doable in
polypropylene.

It's not free. The reactance of the cap at low frequencies adds to the
impedance seen by the amplifier, and low frequency noise increases
accordingly.

In article ,
Arny Krueger wrote:
"Paul Stamler" wrote in message

Yeah, but if you're designing the thing yourself you can do it so as
not to need 100uF. Make the resistance on the right-hand side of the
cap 10k and you can get away with 10uF. That's doable in
polypropylene.

It's not free. The reactance of the cap at low frequencies adds to the
impedance seen by the amplifier, and low frequency noise increases
accordingly.

Yes, but it's low frequency noise, and only that component caused by
the front end's input current noise, so it's really pretty benign.
Sure, it's nice to not have it there at all, but using 10uF isn't such
a horrible compromise. 10uF has a reactance of 150 ohms at 106Hz, and
that's not all that high up there. Use 20uF and you move the turnover
to 50Hz, which is probably lower than the 1/f noise from most
semiconductors used in front ends anyway.


Regards,

Monte McGuire

(Scott Dorsey) wrote in message ...
R. Foote wrote:
Which caps do you folks recommend for dc blocking in mic pres? I built
one using large polypropylene caps, and would like to do one using
smaller parts.

Depends entirely on how small they need to be. I assume we are talking
about phantom power blocking on the front end? Any other blocking caps
inside are going to depend completely on the technology of the active
electronics you are using.

Panasonic SU non polars?
Multiple film caps in parallel?

I can recommend the Panasonic FC series caps if you have to go electrolytic.
Maybe with a film bypass cap if there is enough room. 20 uF of the Panasonic
Mylar caps, though, will sound better than an electrolytic and won't take up
quite as much room as polypropylene.

There is no reason to go nonpolar, since there is considerable DC offset.
--scott

Thanks Scott,

The preamp I just finished used 2 - 20uF polypropylene's and sounds
pretty good. Also takes up an enormous amount of room... It was a
challenge getting 2 pres into a 1u enclosure.

I would like to try the electrolytics as a comparison (front end dc
blocking) and , yes I would like to use FC's. But, lots of the data
sheets and such say that if phantom power is to be disabled, use a non
polar cap. Man, I would much rather use FC's if I can, but will be
turning phantom power on and off.

What do you think of the "wisdom" of using non polars for these
applications.

Roger Foote

"Marko" wrote in message ...
www.solen.ca

Solen, manufactures one of the best polypropylene capacitors in my opinion.
They are used by many loudspeaker manufactures, and they have a great web
page.


Marko.


Thanks Marko

Yes, Solens are a beautiful cap. I have some that I used for a
loudspeaker crossover and I might try them too. Right now though, I
want to design a single channel pre in a half rack box, hence the
question about electrolytics.

Roger Foote

In article ,
R. Foote wrote:
The preamp I just finished used 2 - 20uF polypropylene's and sounds
pretty good. Also takes up an enormous amount of room... It was a
challenge getting 2 pres into a 1u enclosure.

Nice! The other great thing is you'll never need to replace those
caps, unlike with an electrolytic. Which brand of cap(s) did you use?

I would like to try the electrolytics as a comparison (front end dc
blocking) and , yes I would like to use FC's. But, lots of the data
sheets and such say that if phantom power is to be disabled, use a non
polar cap. Man, I would much rather use FC's if I can, but will be
turning phantom power on and off.

Get some logic and a relay or two in there and remove the coupling
caps when P48 is turned off. It's not hard to do, good relays aren't
a fortune, and the whole shebang is pretty small compared to film
caps. You can use one shots like LS123s and some simple logic to
avoid using a clock (and getting noise from a clock) while still doing
some sequencing.

What do you think of the "wisdom" of using non polars for these
applications.

Nonpolar electrolytics are still essentially polarized electrolytics,
just two of them back to back. The chemistry is no different. In
some ways, they're worse since one is always reverse biased; at least
with a polar cap, you know which way to polarize it.


Best of luck,

Monte McGuire

Monte P McGuire wrote:

In article ,
R. Foote wrote:

I would like to try the electrolytics as a comparison (front end dc
blocking) and , yes I would like to use FC's. But, lots of the data
sheets and such say that if phantom power is to be disabled, use a non
polar cap. Man, I would much rather use FC's if I can, but will be
turning phantom power on and off.


Get some logic and a relay or two in there and remove the coupling
caps when P48 is turned off.

Or add another XLR in that bypasses the caps and phantom...

Scott Dorsey wrote in message
...

Yeah, but if you're designing the thing yourself you can do it so as not
to
need 100uF. Make the resistance on the right-hand side of the cap 10k and
you can get away with 10uF. That's doable in polypropylene.

That's where the rub is. You want a low input impedance on your preamp,
so that it properly loads dynamic microphones. BUT, you want a high input
impedance, so your phantom blocking caps aren't enormous.

With a 600 ohm input Z, 10 uF gives you a -3dB point at 20 Hz or so, which
is kind of high, I think. But not insane.

So put your 600 ohm or whatever oad resistor on the left side of the
capacitors, between the two input pins. They have the same DC voltage on
them, if your phantom resistors are matched properly, so no current will
flow. Total AC load will be the resistor across the pins in parallel with
the 10k on the right side of the caps.

Peace,
Paul

Arny Krueger wrote in message
...

It's not free. The reactance of the cap at low frequencies adds to the
impedance seen by the amplifier, and low frequency noise increases
accordingly.

Yeah, but just looking at Johnson noise (okay, not realistic, but bear with
me for a moment) my back-of-the-envelope calculations say that the total
noise contributed by the cap will be about 12dB below the Johnson noise of a
150-ohm microphone, and that's not significant, particularly in a region
where Fletcher and Munson say you don't hear so well to begin with.

Peace,
Paul

"Paul Stamler" wrote in message

Arny Krueger wrote in message
...

It's not free. The reactance of the cap at low frequencies adds to
the impedance seen by the amplifier, and low frequency noise
increases accordingly.

Yeah, but just looking at Johnson noise (okay, not realistic, but
bear with me for a moment) my back-of-the-envelope calculations say
that the total noise contributed by the cap will be about 12dB below
the Johnson noise of a 150-ohm microphone, and that's not
significant, particularly in a region where Fletcher and Munson say
you don't hear so well to begin with.

my Excel spreadsheet comes up with the following reactances for a 10 uF
cap:

Freq Hz Cap, uF X(C)
1 10 15915.5
20 10 795.8
50 10 318.3
100 10 159.2

It seems to me like at 100 Hz and below the cap is contributing more noise
than the mic.

R. Foote wrote:

The preamp I just finished used 2 - 20uF polypropylene's and sounds
pretty good. Also takes up an enormous amount of room... It was a
challenge getting 2 pres into a 1u enclosure.

That's very small. What is your input impedance? If you have a nice low
input Z, you might benefit from bigger caps.

I would like to try the electrolytics as a comparison (front end dc
blocking) and , yes I would like to use FC's. But, lots of the data
sheets and such say that if phantom power is to be disabled, use a non
polar cap. Man, I would much rather use FC's if I can, but will be
turning phantom power on and off.

Yes, you absolutely have to have DC offset across a conventional cap.

What do you think of the "wisdom" of using non polars for these
applications.

I think that a nonpolar will still sound better if you can keep a DC
offset across it. Why can't you live with permanently on phantom power?
--scott
--
"C'est un Nagra. C'est suisse, et tres, tres precis."

On Tue, 9 Mar 2004 07:26:56 -0500, "Arny Krueger"
wrote:

my Excel spreadsheet comes up with the following reactances for a 10 uF
cap:

Freq Hz Cap, uF X(C)
1 10 15915.5
20 10 795.8
50 10 318.3
100 10 159.2

It seems to me like at 100 Hz and below the cap is contributing more noise
than the mic.

Reactances don't contribute noise; you doubtless know this; so what
does this mean?

Also, for everyone, remember that the two caps are in series for
signal and noise.

Chris Hornbeck

"Second star to the right,
Then straight on 'til morning."

R. Foote wrote:

The preamp I just finished used 2 - 20uF polypropylene's and sounds
pretty good. Also takes up an enormous amount of room... It was a
challenge getting 2 pres into a 1u enclosure.

Don't forget you can construct parallel arrays of capacitors if cap
diameter is a problem. 5 times 4.7uF per channel will take up more "floor
space" but less height.

--
Mark.
http://tranchant.plus.com/

"Chris Hornbeck" wrote in message

On Tue, 9 Mar 2004 07:26:56 -0500, "Arny Krueger"
wrote:

my Excel spreadsheet comes up with the following reactances for a
10 uF cap:

Freq Hz Cap, uF X(C)
1 10 15915.5
20 10 795.8
50 10 318.3
100 10 159.2

It seems to me like at 100 Hz and below the cap is contributing more
noise than the mic.

Reactances don't contribute noise; you doubtless know this; so what
does this mean?

What does contribute noise is the effective source resistance for the
amplifier. Normally a microphone will bring this way down. However, these
caps are decoupling the input stage from this nice low source resistance out
in the mic. So, the amp sees a higher source resistance, and this increases
the noise that is generated by the things that can generate noise. This
includes thermal noise in any resistors, and it includes the consequences of
any noise source that is modeled as a current source. When a current source
acts on an increased resistance, there is more noise.

Also, for everyone, remember that the two caps are in series for
signal and noise.

In a sense that is pretty meaningful . Or, there are two noise sources, one
for the balanced and unbalanced input, and they each have one cap as part of
their input circuit.

On Tue, 9 Mar 2004 10:37:16 -0500, "Arny Krueger"
wrote:

What does contribute noise is the effective source resistance for the
amplifier. Normally a microphone will bring this way down. However, these
caps are decoupling the input stage from this nice low source resistance out
in the mic. So, the amp sees a higher source resistance, and this increases
the noise that is generated by the things that can generate noise. This
includes thermal noise in any resistors, and it includes the consequences of
any noise source that is modeled as a current source. When a current source
acts on an increased resistance, there is more noise.

Excellent. And for our purposes, two 10uF coupling caps present
a series source impedance of 5uF. And will have the low frequency
rolloff of a 5uF cap.

Chris Hornbeck

"Second star to the right,
Then straight on 'til morning."

Chris Hornbeck wrote:
On Tue, 9 Mar 2004 07:26:56 -0500, "Arny Krueger"
wrote:

my Excel spreadsheet comes up with the following reactances for a 10 uF
cap:

Freq Hz Cap, uF X(C)
1 10 15915.5
20 10 795.8
50 10 318.3
100 10 159.2

It seems to me like at 100 Hz and below the cap is contributing more noise
than the mic.

Reactances don't contribute noise; you doubtless know this; so what
does this mean?

It's handy for figuring out how much low end roll-off you will get from
the series capacitor, into a given impedance load. You know your load
impedance and the capacitor impedance form a resistive divider, so you
can figure out what the voltage loss throught the cap is at any given
frequency.

Also, for everyone, remember that the two caps are in series for
signal and noise.

Depends on how you want to model the differential input stage.....
--scott
--
"C'est un Nagra. C'est suisse, et tres, tres precis."

In article ,
my Excel spreadsheet comes up with the following reactances for a
10 uF cap:

Freq Hz Cap, uF X(C)
1 10 15915.5
20 10 795.8
50 10 318.3
100 10 159.2

It seems to me like at 100 Hz and below the cap is contributing more
noise than the mic.

The reactances are correct, but the meaning of that is not quite what
you think...

Reactances don't contribute noise; you doubtless know this; so what
does this mean?

What does contribute noise is the effective source resistance for the
amplifier. Normally a microphone will bring this way down. However, these
caps are decoupling the input stage from this nice low source resistance out
in the mic. So, the amp sees a higher source resistance, and this increases
the noise that is generated by the things that can generate noise. This
includes thermal noise in any resistors,

No, the Johnson noise from a resistor, such as the effective source
resistance of a mike, will be unchanged by any loading.

and it includes the consequences of
any noise source that is modeled as a current source. When a current source
acts on an increased resistance, there is more noise.

Yes, this is the _only_ thing that will cause increased noise from a
higher source impedance caused by excess reactance: the input current
noise from the amplifier's input stage devices.

There are three main sources of noise in a mike amp: the Johnson noise
caused by the resistance of the source (the mike), the input voltage
noise caused by the input stage devices and the input current noise
caused by the input stage devices. Only one of these three sources is
affected by excess reactance at a mike amp input, and the significance
of this noise source also depends on the type of device used for the
input stage.

For example, JFET input devices often have much lower current noise
contributions than bipolar input devices that have similar total noise
performance. So, whether this increase is significant or not really
depends upon the particular devices used.


Regards,

Monte McGuire

(Monte P McGuire) wrote in message ...
In article ,
R. Foote wrote:
The preamp I just finished used 2 - 20uF polypropylene's and sounds
pretty good. Also takes up an enormous amount of room... It was a
challenge getting 2 pres into a 1u enclosure.

Nice! The other great thing is you'll never need to replace those
caps, unlike with an electrolytic. Which brand of cap(s) did you use?

I used Vishay (Sprague), the only other brand I had access to at the
time was Cornell Dubileer... The pre uses an INA217 and it likes to
have 2.2K from each input to gnd, so a 20 uf is pretty much required.

Get some logic and a relay or two in there and remove the coupling
caps when P48 is turned off. It's not hard to do, good relays aren't
a fortune, and the whole shebang is pretty small compared to film
caps. You can use one shots like LS123s and some simple logic to
avoid using a clock (and getting noise from a clock) while still doing
some sequencing.

Or maybe use a timer. P48 off, time out, then short caps. That would
sure make it sweet for using non condensor mics!

Good Idea, maybe next time


What do you think of the "wisdom" of using non polars for these
applications.

Nonpolar electrolytics are still essentially polarized electrolytics,
just two of them back to back. The chemistry is no different. In
some ways, they're worse since one is always reverse biased; at least
with a polar cap, you know which way to polarize it.

Check, that's what I thought. Oh well, so much for np's.

Thanks Monte

Best of luck,

Monte McGuire

Once again, I'm just blown away by the value of this newsgroup as an info
resource. Thanks, all, for contributing your remarkable knowledge so
freely!

George Reiswig
Song of the River Music

"R. Foote" wrote in message
om...
(Monte P McGuire) wrote in message
...
In article ,
R. Foote wrote:
The preamp I just finished used 2 - 20uF polypropylene's and sounds
pretty good. Also takes up an enormous amount of room... It was a
challenge getting 2 pres into a 1u enclosure.

Nice! The other great thing is you'll never need to replace those
caps, unlike with an electrolytic. Which brand of cap(s) did you use?

I used Vishay (Sprague), the only other brand I had access to at the
time was Cornell Dubileer... The pre uses an INA217 and it likes to
have 2.2K from each input to gnd, so a 20 uf is pretty much required.

Get some logic and a relay or two in there and remove the coupling
caps when P48 is turned off. It's not hard to do, good relays aren't
a fortune, and the whole shebang is pretty small compared to film
caps. You can use one shots like LS123s and some simple logic to
avoid using a clock (and getting noise from a clock) while still doing
some sequencing.

Or maybe use a timer. P48 off, time out, then short caps. That would
sure make it sweet for using non condensor mics!

Good Idea, maybe next time


What do you think of the "wisdom" of using non polars for these
applications.

Nonpolar electrolytics are still essentially polarized electrolytics,
just two of them back to back. The chemistry is no different. In
some ways, they're worse since one is always reverse biased; at least
with a polar cap, you know which way to polarize it.

Check, that's what I thought. Oh well, so much for np's.

Thanks Monte

Best of luck,

Monte McGuire

In article ,
R. Foote wrote:
(Monte P McGuire) wrote in message ...
Get some logic and a relay or two in there and remove the coupling
caps when P48 is turned off. It's not hard to do, good relays aren't
a fortune, and the whole shebang is pretty small compared to film
caps. You can use one shots like LS123s and some simple logic to
avoid using a clock (and getting noise from a clock) while still doing
some sequencing.

Or maybe use a timer. P48 off, time out, then short caps. That would
sure make it sweet for using non condensor mics!

I'd recommend you use a DPDT (2 form C) type relay and not short the
caps, but merely route the input jacks to the P48 side or the preamp
side of the input blocking caps. Let the other side of the cap float
when it's not in circuit. Expecting a relay to take the huge current
pulse of shorting a big cap charged to 48V and also performing well
with microscopic signals is a bit too much to ask!! After one or two
discharge cycles, the contacts will have small craters blasted into
them, and they won't work well anymore.

Waiting for P48 to bleed down is possible, but it's not a constant
time - it depends on the mike being powered, and it may take quite a
bit for the voltage to get down to even a few volts. Also, it annoys
the cap to have to short it out. Let it float...

While you're at it, you could add a muting relay to short each preamp
input to ground to avoid passing the large pulse that happens when you
connect/disconnect P48. It won't prevent hot plugging disasters, but
it will certainly lighten the load on your normal input voltage
clamping system and it'll let you turn P48 on/off without getting a
very large click (or bang... ;-)

Good Idea, maybe next time

I built an otherwise non-special preamp that used logic and relays to
handle P48 switching and it works well. I built mine out of clockless
garbage logic (4 HC chips, versions of the old TTL stuff), but I'd
probably do it again with a PIC controller that stays in sleep mode
most of the time to avoid clock noise. The PIC would save on PC board
real estate and allow the last values to be stored in EEPROM for the
next startup. They're pretty slick little devices!


Regards,

Monte McGuire

(Scott Dorsey) wrote:

There is no reason to go nonpolar, since there is considerable DC offset.

Except when you turn off the phantom power. Then there's very little if any DC
offset.

ulysses

(Scott Dorsey) wrote in message ...
R. Foote wrote:

The preamp I just finished used 2 - 20uF polypropylene's and sounds
pretty good. Also takes up an enormous amount of room... It was a
challenge getting 2 pres into a 1u enclosure.

That's very small. What is your input impedance? If you have a nice low
input Z, you might benefit from bigger caps.

The input impedance is 2.21K from each input to gnd. So, if I was
correct, I would be down .1 or .2 db at 20hz. (INA217 preamp chip)


I would like to try the electrolytics as a comparison (front end dc
blocking) and , yes I would like to use FC's. But, lots of the data
sheets and such say that if phantom power is to be disabled, use a non
polar cap. Man, I would much rather use FC's if I can, but will be
turning phantom power on and off.

Yes, you absolutely have to have DC offset across a conventional cap.

What do you think of the "wisdom" of using non polars for these
applications.

I think that a nonpolar will still sound better if you can keep a DC
offset across it. Why can't you live with permanently on phantom power?
--scott

Well, I might want to use dynamics etc... That and I guess I was
incorporating features I thought were sorta "standard".

Thanks again
rf

"R. Foote" wrote:

"Marko" wrote in message ...
www.solen.ca

Solen, manufactures one of the best polypropylene capacitors in my opinion.
They are used by many loudspeaker manufactures, and they have a great web
page.

Marko.

Thanks Marko

Yes, Solens are a beautiful cap. I have some that I used for a
loudspeaker crossover and I might try them too. Right now though, I
want to design a single channel pre in a half rack box, hence the
question about electrolytics.

Quite different requirements for a crossover cap, esp when considering the current rating !

Recent thinking suggests that using a low HF impedance cap such as designed for switch mode power
supply output filters is a very good choice for your application.

The link someone posted regarding various types of caps is kinda out of date. It didn't even
mention multi-later ceramics for example but did mention the almost unobtainable ( these days )
mica types.

Component qualities move on - what seemed right 20 yrs ago isn't necessarily good advice any more.

Use at least 100 uF btw if you want LF noise to remain low.


Graham

Marko wrote:

www.solen.ca

they have a great web page.

That clearly sorts it then !

Graham

Paul Stamler wrote:

Yeah, but if you're designing the thing yourself you can do it so as not to
need 100uF. Make the resistance on the right-hand side of the cap 10k and
you can get away with 10uF. That's doable in polypropylene.

Tssk @ you Paul.

Surely you know that the noise figure for a mic pre is determined in part by the
impedance the pre *sees*.

10uF = 160 ohms @ 100Hz. Two in series obviously are 320 ohms. Source R = 150 ?
- coupling Z should be much lower, e.g. 2x100uF in series = 32 ohms.

Poor flicker noise otherwise.

Graham

Arny Krueger wrote:

"Paul Stamler" wrote in message

Arny Krueger wrote in message
...

It's not free. The reactance of the cap at low frequencies adds to
the impedance seen by the amplifier, and low frequency noise
increases accordingly.

Yeah, but just looking at Johnson noise (okay, not realistic, but
bear with me for a moment) my back-of-the-envelope calculations say
that the total noise contributed by the cap will be about 12dB below
the Johnson noise of a 150-ohm microphone, and that's not
significant, particularly in a region where Fletcher and Munson say
you don't hear so well to begin with.

my Excel spreadsheet comes up with the following reactances for a 10 uF
cap:

Freq Hz Cap, uF X(C)
1 10 15915.5
20 10 795.8
50 10 318.3
100 10 159.2

It seems to me like at 100 Hz and below the cap is contributing more noise
than the mic.

You are absolutely right. Learnt this one back in 1979.

Only experienced mic pre designers know this stuff. ( apologies to anyone who
may feel 'miffed' )

Graham


p.s. it's *doubly bad* due to the 1/f noise corner of active devices.

Chris Hornbeck wrote:

On Tue, 9 Mar 2004 07:26:56 -0500, "Arny Krueger"
wrote:

my Excel spreadsheet comes up with the following reactances for a 10 uF
cap:

Freq Hz Cap, uF X(C)
1 10 15915.5
20 10 795.8
50 10 318.3
100 10 159.2

It seems to me like at 100 Hz and below the cap is contributing more noise
than the mic.

Reactances don't contribute noise; you doubtless know this; so what
does this mean?

Noise = noise voltage. Are you saying that current passing through a cap
produces no voltage across it ?

The fact that the component is imaginary makes the impedance ( reactance ) no
less *real* - lol.

Graham

Monte P McGuire wrote:

snip

What does contribute noise is the effective source resistance for the
amplifier. Normally a microphone will bring this way down. However, these
caps are decoupling the input stage from this nice low source resistance out
in the mic. So, the amp sees a higher source resistance, and this increases
the noise that is generated by the things that can generate noise. This
includes thermal noise in any resistors,

No, the Johnson noise from a resistor, such as the effective source
resistance of a mike, will be unchanged by any loading.

Sure, the resistor itself isn't the problem. You're forgetting the caps in the
way.

and it includes the consequences of
any noise source that is modeled as a current source. When a current source
acts on an increased resistance, there is more noise.

Yes, this is the _only_ thing that will cause increased noise from a
higher source impedance caused by excess reactance: the input current
noise from the amplifier's input stage devices.

Yes, it's the input noise current - even specced on high end op-amps, although
you won't be using an op-amp for a mic.


There are three main sources of noise in a mike amp: the Johnson noise
caused by the resistance of the source (the mike), the input voltage
noise caused by the input stage devices and the input current noise
caused by the input stage devices.

Flowing in the source *impedance*.

Actually, you missed the emitter dynamic impedance as a noise source, although
some like to lump this into the published noise figures. That's why you see no
mic pres operating with 100uA flowing through the input devices. Or maybe you do
( as in noisy ones - lol ).

The typical inter-emitter gain set resitor is there too.

Only one of these three sources is
affected by excess reactance at a mike amp input, and the significance
of this noise source also depends on the type of device used for the
input stage.

Yes.

For example, JFET input devices often have much lower current noise
contributions than bipolar input devices that have similar total noise
performance. So, whether this increase is significant or not really
depends upon the particular devices used.

Sadly, FETs typically have worse 1/f noise than bipolars used for othis job.
Seems like you can't win.


Graham

"Pooh Bear" wrote in message
...
It seems to me like at 100 Hz and below the cap is contributing more
noise
than the mic.

You are absolutely right. Learnt this one back in 1979.
Only experienced mic pre designers know this stuff. ( apologies to anyone
who
may feel 'miffed' )

Yeah right, even most not so experienced microphone and phono pre-amp
designers knew it 40 years ago.
Didn't, and still doesn't stop many poor commercial designs of course.

TonyP.

"Pooh Bear" wrote in message

Monte P McGuire wrote:

snip

What does contribute noise is the effective source resistance for
the amplifier. Normally a microphone will bring this way down.
However, these caps are decoupling the input stage from this nice
low source resistance out in the mic. So, the amp sees a higher
source resistance, and this increases the noise that is generated
by the things that can generate noise. This includes thermal noise
in any resistors,

No, the Johnson noise from a resistor, such as the effective source
resistance of a mike, will be unchanged by any loading.

Sure, the resistor itself isn't the problem. You're forgetting the
caps in the way.

Exactly!

I was looking for a good way to say this. Thanks!

"Pooh Bear" wrote in message

Arny Krueger wrote:

"Paul Stamler" wrote in message

Arny Krueger wrote in message
...

It's not free. The reactance of the cap at low frequencies adds to
the impedance seen by the amplifier, and low frequency noise
increases accordingly.

Yeah, but just looking at Johnson noise (okay, not realistic, but
bear with me for a moment) my back-of-the-envelope calculations say
that the total noise contributed by the cap will be about 12dB below
the Johnson noise of a 150-ohm microphone, and that's not
significant, particularly in a region where Fletcher and Munson say
you don't hear so well to begin with.

my Excel spreadsheet comes up with the following reactances for a
10 uF cap:

Freq Hz Cap, uF X(C)
1 10 15915.5
20 10 795.8
50 10 318.3
100 10 159.2

It seems to me like at 100 Hz and below the cap is contributing more
noise than the mic.

You are absolutely right. Learnt this one back in 1979.

Only experienced mic pre designers know this stuff. ( apologies to
anyone who may feel 'miffed' )

Experienced mic pre designers only? That tain't me. It lept right out the
schematic at me about a year ago when we were deconstructing one of Rane's
designs. Or maybe someone else who knew better commented on it.

Pooh Bear wrote:

For example, JFET input devices often have much lower current noise
contributions than bipolar input devices that have similar total noise
performance. So, whether this increase is significant or not really
depends upon the particular devices used.

Sadly, FETs typically have worse 1/f noise than bipolars used for othis job.
Seems like you can't win.

You can't. BUT, you can use an input transformer, which gives you some
free voltage gain even as it's giving you more thermal noise. This also
lets you eliminate the blocking capacitors completely. With a typical FET
or tube input stage, this gives you lower noise than you would get without
it. With a bipolar input stage it's a net loss. And of course you get
more distortion and pay a lot more money, but you also get RF rejection in
the bargain.
--scott

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

In article ,
Pooh Bear wrote:

Monte P McGuire wrote:
snip
What does contribute noise is the effective source resistance for the
amplifier. Normally a microphone will bring this way down. However, these
caps are decoupling the input stage from this nice low source resistance out
in the mic. So, the amp sees a higher source resistance, and this increases
the noise that is generated by the things that can generate noise. This
includes thermal noise in any resistors,

No, the Johnson noise from a resistor, such as the effective source
resistance of a mike, will be unchanged by any loading.

Sure, the resistor itself isn't the problem. You're forgetting the caps in the
way.

No, I'm methodically addressing each noise source, the point being
that the impedance of the coupling cap only affects _one_ of the noise
sources, the current noise generated at the amplifier input. Arny
stated or implied that the coupling cap would also affect the noise
generated by the resistive component of the source and that's wrong.


Regards,

Monte McGuire

In article ,
Pooh Bear wrote:

Sadly, FETs typically have worse 1/f noise than bipolars used for this job.
Seems like you can't win.


Graham

Nor can you break even. But when was the last time anyone complained
about preamp noise? Seriously.

The SSM-2220 seems to work, though

DC

--
Dave Collins Entropy just isn't what it used to be!


www.collinsaudio.com

Monte P McGuire wrote:

In article ,
Pooh Bear wrote:

Monte P McGuire wrote:
snip
What does contribute noise is the effective source resistance for the
amplifier. Normally a microphone will bring this way down. However, these
caps are decoupling the input stage from this nice low source resistance out
in the mic. So, the amp sees a higher source resistance, and this increases
the noise that is generated by the things that can generate noise. This
includes thermal noise in any resistors,

No, the Johnson noise from a resistor, such as the effective source
resistance of a mike, will be unchanged by any loading.

Sure, the resistor itself isn't the problem. You're forgetting the caps in the
way.

No, I'm methodically addressing each noise source, the point being
that the impedance of the coupling cap only affects _one_ of the noise
sources, the current noise generated at the amplifier input. Arny
stated or implied that the coupling cap would also affect the noise
generated by the resistive component of the source and that's wrong.

Just so I understand you right, are you saying that - but for input noise current -
the noise figure of the pre won't be affected by coupling cap value ?


Graham