Re: Should I build this simple Mosfet follower amp?

[CJT]

Arny Krueger wrote:

> "Kevin McMurtrie" > wrote in message
> ...
<snip>
>>A nearby electronics surplus store is the key.
>
>
> I live in a large metro area (Detroit) and the last good local electronics
> surplus store has been gone for about 20 years.
<snip>

Was that Silverstein's on the east side?

[Kevin McMurtrie]

In article >,
"Arny Krueger" > wrote:

>"Kevin McMurtrie" > wrote in message
...
[snip]
>> A nearby electronics surplus store is the key.
>
>I live in a large metro area (Detroit) and the last good local electronics
>surplus store has been gone for about 20 years. Besides, relying on surplus
>is condemning yourself to outdated technology. However since power amps are
>a mature technology, that isn't the problem it might be. However, since
>power amps are a mature technology, what's to be gained by avoiding finished
>products that are often sold for far less than what the parts would cost,
>even from a good supplier (not Radio Shack).
[snip]

Maybe you're thinking of an electronics junkyard. The surplus stores
I've seen get parts left over from assembly line prototypes,
discontinued products, and stock from factories that closes down. I
usually find enough modern unused components to build 95% of a project.

[Bob-Stanton]

"Arny Krueger" > wrote in message news:<5A6dnS1fBuizjpGiU-


> Solid state amplifiers have been around for about 40 years and after 40
> years of development, why shouldn't most issues be pretty well settled?
>

That is because the FTC has never mandated a set of specifications
that tells very well how an amplifier or CD player performs. If the
FTC were to mandate a set of tests (similar to the tests used on your
website) it would eliminate much of the fraud and confusion in the
industry.

For example, a simple two tone test (using 17 KHz and 19 KHz) would
tell much more about the quality of amplifiers and CD players than
the presently used "THD" tests. The potential buyer could simply look
at the level of the second order beat falling at 2 KHz, and of two
third order beats falling at 15 KHz and 21 KHz, to help him determine
which amplifier or CD player had the lower distortion.

The consumer doesn't have much comfidence in the way amplifiers and CD
players are presently tested and specified. That leaves him open to
fraud and hype on the one hand, and to overlooking perfect good (low
cost) products on the other hand. Better tests and specifications
would go a long way toward eliminating fraud, deception, and comsumer
confusion in the audio industry.

Bob Stanton

[CJT]

Arny Krueger wrote:

> "CJT" > wrote in message
>
>
>>Arny Krueger wrote:
>>
>>
>>>"Kevin McMurtrie" > wrote in message
...
>>
>><snip>
>>
>>>>A nearby electronics surplus store is the key.
>>>
>>>
>>>I live in a large metro area (Detroit) and the last good local
>>>electronics surplus store has been gone for about 20 years.
>>
>><snip>
>
>
>>Was that Silverstein's on the east side?
>
>
> No, they died more than 30 years ago. At its peak it was quite the place,
> incluing the Duck that they actually started up and ran from time to time.
>
> Then there was Aaron on Chene that died earlier. Hershel's on Grand River,
> that died a little later.

Thanks, I was trying to remember the name of Hershel's; I spent many
hours there sifting through stuff.

>
> I was thinking of Lee's at 10 mile and Gratiot.
>
>

[CJT]

Bob-Stanton wrote:

> "Arny Krueger" > wrote in message news:<5A6dnS1fBuizjpGiU-
>
>
>
>>Solid state amplifiers have been around for about 40 years and after 40
>>years of development, why shouldn't most issues be pretty well settled?
>>
>
>
> That is because the FTC has never mandated a set of specifications
> that tells very well how an amplifier or CD player performs. If the
> FTC were to mandate a set of tests (similar to the tests used on your
> website) it would eliminate much of the fraud and confusion in the
> industry.

I'm not sure the FTC should be doing much more than setting minimal
standards to allow prosecuting false advertising. The IEEE or some
such body seems more appropriate for setting test standards.

>
> For example, a simple two tone test (using 17 KHz and 19 KHz) would
> tell much more about the quality of amplifiers and CD players than
> the presently used "THD" tests. The potential buyer could simply look
> at the level of the second order beat falling at 2 KHz, and of two
> third order beats falling at 15 KHz and 21 KHz, to help him determine
> which amplifier or CD player had the lower distortion.
>
> The consumer doesn't have much comfidence in the way amplifiers and CD
> players are presently tested and specified. That leaves him open to
> fraud and hype on the one hand, and to overlooking perfect good (low
> cost) products on the other hand. Better tests and specifications
> would go a long way toward eliminating fraud, deception, and comsumer
> confusion in the audio industry.

I don't think the consumer cares. If he did, I would have thought
manufacturers would satisfy his need as a result of competitive
pressure.

>
> Bob Stanton

[Alotbsol]

Not if you believe Doug Self - try reading his power amp design hand-book.
He has a few designs, which you might use as a basis.

Neil McBride http://www.neilmcbride.co.uk/ has a Naim 135 clone with
layouts. I have built these and they sound pretty good. Cost is way
cheaper than getting them from Naim, especially if you cannibalise an old
amp for tranformer, chassis and caps, which are the most expensive parts.


Andy > wrote in message
om...
> Hi,
> Can someone comment on this amp?
>
> http://www.pha.inecnet.cz/macura/follower_e.html
>
> Seems very simple and elegant. Can I expect good hi-fi sound out
> of this? How will it compare, says, to a Hafler DH200?
>
> Thanks
> Andy

[normanstrong]

>
> Andy > wrote in message
> om...
> > Hi,
> > Can someone comment on this amp?
> >
> > http://www.pha.inecnet.cz/macura/follower_e.html
> >
> > Seems very simple and elegant. Can I expect good hi-fi sound
out
> > of this? How will it compare, says, to a Hafler DH200?

It's reasonably simple, but certainly not elegant. It's extremely
inefficient. It has no gain, so you'll probably have to find a
substantial preamplifier just to drive it. It has poor power supply
rejection. It has no feedback (because it has no gain) and this is
listed as an advantage; why, I don't know. It contains a 3.3v zener
diode, which is about as poor as a zener gets.

Since I know nothing about the Hafler DH200, I can't comment reliably
on how it would compare. It's been my experience, however, that
Hafler is a superb designer. I can't imagine him designing anything
as crummy as that Macura design.

Norm Strong

[Bob-Stanton]

CJT > wrote in message
..
>
> I'm not sure the FTC should be doing much more than setting minimal
> standards to allow prosecuting false advertising. The IEEE or some
> such body seems more appropriate for setting test standards.
>

I agree. We really don't want the goverment poking around in the audio
industry.

> I don't think the consumer cares. If he did, I would have thought
> manufacturers would satisfy his need as a result of competitive
> pressure.
>

I agree with that as well.

I don't expect consumers in Wal-Mart, buying a boom-box, to care much
about intermodulation distortion. But, even in this audio group,
people seem to have little interest in what the actual measured
distortion of a product is. For example, I've seen threads on: SACD vs
DVD-A vs CD. There were sometimes hundreds of messages, in the
threads, but no one even mentioned actual *measured* performance of
the various types of systems.

Bob Stanton

[Andy]

I do not know much, but I do know this from 1st hand experience:

Comparison of Sony DA5ES, Rotel RB-980, and a Mark-5 TA-477 (diy amp)

Sony:
Type: 5.1 channel receiver
paid: ~$900 in 2001
Music production: tiny, digital artifacts (breaking up?), high
female voices sounded like a cheap clock radio..
(exaggeratted a bit..)

Rotel:
Type: 2 channel power amp
paid: ~$700 in 1992
Music production: warm, full. sounded good to me.

Mark-5 TA-477:
Type: diy amp.
paid ~$70/channel in 1992 (not including
casing and power supply)
Music production: warm, full. sounded good to me. Sounded the same
as the Rotel... I'm sure there's subtle differences,
but I didn't hear any.

You draw your own conclusions... :-)

Andy

"Arny Krueger" > wrote in message
...
....
...
> Building power amps and hoping to save money and get superior sound is
> futile at this time. If nothing else you now have to contend with the
> Chinese who can sell you a finished power amp for less than you will pay at
> a good electronics parts place for output devices, heat sink, power
> electrolytics and power transformer.
>
> You mentioned the Hafler DH-200. It's a fine amp. There are a number of
> DH-200 and DH-220 amps up for auction on eBay right now. Current bid prices
> are in the $200 range. You say that you can fix things - if the amp you get
> from eBay is a fixer-upper, have fun and save even more money!
>
> > If so, can someone point to a source
> > for a great diy amp? My kit building skill is pretty good.
> > I can etch my own boards if I have the circuit layout
> > patterns, but I don't have the skills to do a layout from
> > scratch.
>
> If it was worth the effort, I'd say do it. But, it isn't. Power amps are a
> solved problem - the only thing left for them to do is to get smaller,
> cheaper, lighter...

[Trevor]

"Bob-Stanton" > wrote in message
om...
>But, even in this audio group,
> people seem to have little interest in what the actual measured
> distortion of a product is. For example, I've seen threads on: SACD vs
> DVD-A vs CD. There were sometimes hundreds of messages, in the
> threads, but no one even mentioned actual *measured* performance of
> the various types of systems.

You obviously didn't look too hard then.

Trevor.

[Kevin McMurtrie]

In article >,
(Andy) wrote:

>I do not know much, but I do know this from 1st hand experience:
>
> Comparison of Sony DA5ES, Rotel RB-980, and a Mark-5 TA-477 (diy amp)
>
>Sony:
> Type: 5.1 channel receiver
> paid: ~$900 in 2001
> Music production: tiny, digital artifacts (breaking up?), high
> female voices sounded like a cheap clock radio..
> (exaggeratted a bit..)

[snip]

I've noticed that consumer grade 5.1 receivers lack the loudness
compensation that's traditionally in a power amplifier. Loudness
compensation may be a mess of guesswork that's never perfect but you
sorely miss it when it's gone.

[Andy]

I compared the three at moderate listening levels. The Sony
simply drove me to turn the volume down while the Rotel and
Mark-V was a joy to listen to. I didn't compare the three at
low listening levels...
The Sony was the top of the line ES model as such it is
targeted for audiophiles...

Andy

Kevin McMurtrie > wrote in message >...
> In article >,
> (Andy) wrote:
>
> >I do not know much, but I do know this from 1st hand experience:
> >
> > Comparison of Sony DA5ES, Rotel RB-980, and a Mark-5 TA-477 (diy amp)
> >
> >Sony:
> > Type: 5.1 channel receiver
> > paid: ~$900 in 2001
> > Music production: tiny, digital artifacts (breaking up?), high
> > female voices sounded like a cheap clock radio..
> > (exaggeratted a bit..)
>
> [snip]
>
> I've noticed that consumer grade 5.1 receivers lack the loudness
> compensation that's traditionally in a power amplifier. Loudness
> compensation may be a mess of guesswork that's never perfect but you
> sorely miss it when it's gone.

[Bob-Stanton]

"Trevor" <trevor@home> wrote in message >...
> "Bob-Stanton" > wrote in message
> om...
> >But, even in this audio group,
> > people seem to have little interest in what the actual measured
> > distortion of a product is. For example, I've seen threads on: SACD vs
> > DVD-A vs CD. There were sometimes hundreds of messages, in the
> > threads, but no one even mentioned actual *measured* performance of
> > the various types of systems.
>
> You obviously didn't look too hard then.
>
> Trevor.

I'm interested in a comparison of SACD player vs CD player non-linear
distortion. Do you know any typical numbers for the two systems?

Bob Stanton

[Bob-Stanton]

"Trevor" <trevor@home> wrote in message news:<3f14e6da$0$1209

> > I'm interested in a comparison of SACD player vs CD player non-linear
> > distortion. Do you know any typical numbers for the two systems?
>
> Typical THD figures for both systems are < 0.001%, so I'm assuming this is
> not what you are after :-)

No, that is exactly what I'm after.

>
> My point was that many people mentioned the lack of actual test results that
> show *audible* benefits for SACD or DVDA, just as you are doing.

Actually I'm inclined to believe SACD and DVDA have no audible
benefits over CD. I was looking for some actual distortion
measurements to back up that opinion.


> However since the real debate is whether the *theoretical* differences are
> even audible, actual measurements are rather pointless until some agreement
> of aural capabilities is reached.


Measuring distortion is not pointless. There have been many tests,
over the years, that show what level of THD is audible. (Vitually no
one has ever said that 0.001% distortion is audible.)

> Until then we can just say that SACD and DVDA
> do have more overkill margin.
>

They all seem to have the same amount of overkill (0.001%). I seen no
reason to switch to DADA or SACD, except it is hard to find a CD
player with better than 0.01% distortion, but there are some.

Bob Stanton

[Trevor]

"Bob-Stanton" > wrote in message
om...

> No, that is exactly what I'm after.

I'm glad then.

> Actually I'm inclined to believe SACD and DVDA have no audible
> benefits over CD. I was looking for some actual distortion
> measurements to back up that opinion.

IMO they have no audible benefits either, but the theoretical benefits are
mostly the wider bandwidth and to a lesser degree S/N+D.

> Measuring distortion is not pointless. There have been many tests,
> over the years, that show what level of THD is audible. (Vitually no
> one has ever said that 0.001% distortion is audible.)

True, but a simple THD measurement is not all there is to it.

> They all seem to have the same amount of overkill (0.001%).

No, I said < 0.001%, obviously an SACD or DVDA can *theoretically* be much
less.

> I seen no reason to switch to DADA or SACD, except it is hard to find a CD
> player with better than 0.01% distortion, but there are some.

Even cheap Chinese CD players can usually do better than 0.01% THD these
days, when working properly.

Trevor.

[Bob-Stanton]

"Trevor" <trevor@home> wrote in message news:<3f17a3de$0$1209

> True, but a simple THD measurement is not all there is to it.
>

I agree. THD is a very old (and obsolete) measurement. There are much
better ways of specifing (measuring) distortion, but they seem never
to have never trickled down to the somewhat backwards audio industry.

> > They all seem to have the same amount of overkill (0.001%).
>
> No, I said < 0.001%, obviously an SACD or DVDA can *theoretically* be much
> less.

CD players can also theoritically have much less than 0.001%
distortion.

Bob Stanton

[Richard D Pierce]

In article >,
Bob-Stanton > wrote:
>CD players can also theoritically have much less than 0.001%
>distortion.

Well, let's see. If we're using the conventional THD+N method
for distportion (which most "distortion" meters give), over a 20
kHz bandwidth, no, CD's are NOT capable of "much less than
0.001% distortion. "Much less than 0.001% distortion" requires a
total residual over the bandwidth to be 100 dB down, and that's
simply not doable in a 16 bit system with dither. Perfectly
dithered 16 bit CD is capable of 93 dB broadband residual floor
to maximum output.

If you want to propose something like very narrow effective
frequency resolution and the like, maybe you can get there. But
if you use the broad term "distortion" no, you cannot get to
0.001% with a CD.

--
| Dick Pierce |
| Professional Audio Development |
| 1-781/826-4953 Voice and FAX |
| |

[Bob-Stanton]

(Richard D Pierce) wrote in message >...
> In article >,
> Bob-Stanton > wrote:
> >CD players can also theoritically have much less than 0.001%
> >distortion.
>
> Well, let's see. If we're using the conventional THD+N method
> for distportion (which most "distortion" meters give), over a 20
> kHz bandwidth, no, CD's are NOT capable of "much less than
> 0.001% distortion. "Much less than 0.001% distortion" requires a
> total residual over the bandwidth to be 100 dB down, and that's
> simply not doable in a 16 bit system with dither. Perfectly
> dithered 16 bit CD is capable of 93 dB broadband residual floor
> to maximum output.
>
> If you want to propose something like very narrow effective
> frequency resolution and the like, maybe you can get there. But
> if you use the broad term "distortion" no, you cannot get to
> 0.001% with a CD.

I agree.

I was refering the THD by itself, without noise. In the audio
industry, because of the limitations of early test equipment, the
standard specification is: THD+N.

Distortion by itself, however, can be much lower than the broadband
noise floor. It is true that you can't get to <-93 dB THD+N with a 16
bit CD, but with perfect dithering, the THD (alone), can be much lower
than -93 dB.

Bob Stanton

[Arny Krueger]

"Bob-Stanton" > wrote in message
m

> CD players can also theoretically have much less than 0.001%
> distortion.

Since a properly-dithered digital audio system has no linear or nonlinear
distortion whatsoever, that would have to be true.

Of course, in practice CD players add linear and nonlinear distortion via
their analog circuitry, including the analog side of the DAC.

So, the only CD players with zero linear and nonlinear distortion are those
that have digital outputs. You can see a practical example of this at

http://www.pcavtech.com/play-rec/dv525/index.htm#SNR_DA

I'll leave the SNR+N issue to Mr. Pierce, except to observe that its fairly
common to find practical digital systems that reduce all nonlinear
distortion products to well below the broadband noise floor.

You can see a practical example of this at

http://www.pcavtech.com/soundcards/CardDDeluxe/index.htm#SNR_1644-a

All nonlinear distortion products are at -100 dB or below, but the broadband
noise floor is at -94 dB per
http://www.pcavtech.com/soundcards/CardDDeluxe/index.htm#ZS_1644-a .

[Trevor]

"Bob-Stanton" > wrote in message
om...

> Distortion by itself, however, can be much lower than the broadband
> noise floor. It is true that you can't get to <-93 dB THD+N with a 16
> bit CD, but with perfect dithering, the THD (alone), can be much lower
> than -93 dB.

Exactly, I specified THD not S/N+D. Regardless it is inaudible for either
CD, DVDA or SACD, which was the topic of discussion.
Why you would care about such vanishingly small amounts of THD is not
readily apparent to me.

Trevor.

[Bob-Stanton]

"Trevor" <trevor@home> wrote in message news:<3f1cb911$0$31921


> Exactly, I specified THD not S/N+D. Regardless it is inaudible for either
> CD, DVDA or SACD, which was the topic of discussion.
> Why you would care about such vanishingly small amounts of THD is not
> readily apparent to me.
>
> Trevor.

Many people claim (two channel) SACD and DVD-A sounds better than the
old 16-bit CD's. I'm looking to see if there is any reason why this
should be so. The more I look into it, the more I see *no* meaningful
advantage for two channel SACD and DVD-A, over 16-bit CD. I see no
reason why CD's couldn't sound just as good as two channel SACD's!

I know that you believe there is no reason to know what the distortion
of a CD player or amplifier is, as long as it is "below audability". I
prefer to know what the level of distortion is, and make my own
judgment as to audiblity.

Bob Stanton

[Arny Krueger]

"Bob-Stanton" > wrote in message
om
> "Trevor" <trevor@home> wrote in message news:<3f1cb911$0$31921
>
>
>> Exactly, I specified THD not S/N+D. Regardless it is inaudible for
>> either CD, DVDA or SACD, which was the topic of discussion.
>> Why you would care about such vanishingly small amounts of THD is not
>> readily apparent to me.
>>
>> Trevor.
>
> Many people claim (two channel) SACD and DVD-A sounds better than the
> old 16-bit CD's. I'm looking to see if there is any reason why this
> should be so.

You've got the cart way in front of the horse. First find out if there's any
facts to justify. That would be the scientific approach.

>The more I look into it, the more I see *no* meaningful
> advantage for two channel SACD and DVD-A, over 16-bit CD. I see no
> reason why CD's couldn't sound just as good as two channel SACD's!

Guess what, there's plenty of evidence that higher sample rates and longer
data words offer no audible advantage over 16/44 as a distribution format.

> I know that you believe there is no reason to know what the distortion
> of a CD player or amplifier is, as long as it is "below audibility".

Again, what is the problem with this?

> I prefer to know what the level of distortion is, and make my own
> judgment as to audibility.

False causality. I'm not against knowing what the level of distortion is for
other reasons, but judgments about audibility should be based on actual
listening tests.

[Trevor]

"Bob-Stanton" > wrote in message
om...
> Many people claim (two channel) SACD and DVD-A sounds better than the
> old 16-bit CD's. I'm looking to see if there is any reason why this
> should be so. The more I look into it, the more I see *no* meaningful
> advantage for two channel SACD and DVD-A, over 16-bit CD. I see no
> reason why CD's couldn't sound just as good as two channel SACD's!

Agreed.

> I know that you believe there is no reason to know what the distortion
> of a CD player or amplifier is, as long as it is "below audability". I
> prefer to know what the level of distortion is, and make my own
> judgment as to audiblity.

Actually I think measurements are very important! But my point was that you
are looking in the wrong spot. *IF* ANY difference can be heard from an
otherwise identical recording, it is probably NOT due to the THD
performance. You have to determine what *IS* audible before you can prove
that is the cause of any supposed benefit.

Easier to stay in the 24/96 format (or 24/192) and introduce distortions,
subtract bandwidth etc. and PROVE beyond statistical doubt with double blind
testing, what you can or can't hear. Then repeat for a few thousand other
listeners of all ages, sexes, races etc. I look forward to the definitive
results :-)

Trevor.

[Bob-Stanton]

"Trevor" <trevor@home> wrote in message news:<3f1e2365$0$1208



> Actually I think measurements are very important! But my point was that you
> are looking in the wrong spot. *IF* ANY difference can be heard from an
> otherwise identical recording, it is probably NOT due to the THD
> performance.

That's right. But, you won't know whether the differences are caused
by THD or by something else, if you don't measure THD.


> You have to determine what *IS* audible before you can prove
> that is the cause of any supposed benefit.
>

What is audible has been known for many years.


> Easier to stay in the 24/96 format (or 24/192) and introduce distortions,
> subtract bandwidth etc. and PROVE beyond statistical doubt with double blind
> testing, what you can or can't hear. Then repeat for a few thousand other
> listeners of all ages, sexes, races etc. I look forward to the definitive
> results :-)

Again, most of that kind of work was done years ago. The S/N
audibility studys that were done long ago (for tape and records) still
apply to the the audibility of noise on 24/192.

I *don't have to do* a double blind evaluation of 16-bit CD's to know
that a 93 dB S/N will be inaudible to me. I don't have to do a double
bind evaluation of 24/96, to know that a >110 dB S/N will also be
inaudible to me. I don't have to do a double bind evaluation of SACD
or 16-bit CD, to know that a 0.0001% THD is inaudible to me.

The same is true for the other measurable parameters.

Bob Stanton

[Arny Krueger]

"Bob-Stanton" > wrote in message
om
> "Arny Krueger" > wrote in message
>
>>> Many people claim (two channel) SACD and DVD-A sounds better than
>>> the old 16-bit CD's. I'm looking to see if there is any reason why
>>> this should be so.
>>
>> You've got the cart way in front of the horse. First find out if
>> there's any facts to justify. That would be the scientific approach.
>
>
> The facts are, as stated above: many claim that SACD and DVD-A sound
> better than 16-bit CD's.

So what? People claim lots of crazy things. Flying saucers, anybody?

>> Guess what, there's plenty of evidence that higher sample rates and
>> longer data words offer no audible advantage over 16/44 as a
>> distribution format.

> I'm looking for something more accurate than just listening tests.

When it comes to audibility questions, there is no known better way.

> With listening tests, any imperfections in the signal source can mask
> the difference between systems.

More significantly, the listener's ears mask the differences between
systems, big time.

> For (an extreme) example, Suppose we recorded Thomas Edison's first
> recording, "Mary had a little lamb...", on SACD, DVD-A, CD-16 bit, and
> recorded it on an old 78 RPM wax record cutter. Listening tests would
> show that the old wax record system sounds just the same as SACD.

Perhaps. OTOH there might be quite a bit of ultrasonic content to the
various spurious responses in that wax cutter.

> Listening tests are not easly repeatable.

The seem to converge to consistent results pretty quickly.

> Listening tests are not nearly as sensitive as test instruments.

Of course, but this was about audibility, not measureability, no?

>>> I know that you believe there is no reason to know what the
>>> distortion of a CD player or amplifier is, as long as it is "below
>>> audibility".

>> Again, what is the problem with this?

> Nothing, as long as you don't care to know very much.

What else is there that is relevant to know?

>>> I prefer to know what the level of distortion is, and make my own
>>> judgment as to audibility.

>> False causality. I'm not against knowing what the level of
>> distortion is for other reasons, but judgments about audibility
>> should be based on actual listening tests.

> If listening tests are not correlated to measured performance, what do
> they tell you?

They tell me something we both seem to agree about. Listening tests are not
nearly as sensitive as test instruments.

>Virtually nothing. Only that two things sound the "same".

Which suggests an important truth - the problems with sound recording and
playback are elsewhere.

> Suppose I wanted to know if the temperature in my living room was the
> same as the temperature in my dining room. I could take ten people,
> blind fold them and have them stand in my living room, move them back
> and forth randomly between the two rooms, and see if they could tell a
> difference. After a while I could make a determination that the
> difference in temperature is or is not, subjectively detectable.

OK.

> Yes, the double blind test method "works" and is scientific. But,
> wouldn't it be a lot easier (and more accurate), to just put a
> thermometer in each room?

That's part of the charm of measurements. They are easy, but their relevance
to human perception is not always a given.

> First comes the measurement, then comes the subjective evaluation. All
> you want to do is a subjective evaluation. You didn't put the cart in
> front the horse. You have a cart *without a horse*.

Depends how you define the goal of audio reproduction. Most would define it
as being the perception of a life-like if not live performance.

There's that nasty "perception" word again.

[Phil]

"Bob-Stanton" > wrote in message
om...
> That's right. But, you won't know whether the differences are caused
> by THD or by something else, if you don't measure THD.

And probably not even if you do, unless one is broken.

> > You have to determine what *IS* audible before you can prove
> > that is the cause of any supposed benefit.

> What is audible has been known for many years.

What level of THD is audible in your opinion then?

> > Easier to stay in the 24/96 format (or 24/192) and introduce
distortions,
> > subtract bandwidth etc. and PROVE beyond statistical doubt with double
blind
> > testing, what you can or can't hear. Then repeat for a few thousand
other
> > listeners of all ages, sexes, races etc. I look forward to the
definitive
> > results :-)
>
> Again, most of that kind of work was done years ago. The S/N
> audibility studys that were done long ago (for tape and records) still
> apply to the the audibility of noise on 24/192.

But ANY audible noise on 24/192 is obviously *NOT* due to limitations of the
format.
What are your figures? What is the required listening environment.

> I *don't have to do* a double blind evaluation of 16-bit CD's to know
> that a 93 dB S/N will be inaudible to me. I don't have to do a double
> bind evaluation of 24/96, to know that a >110 dB S/N will also be
> inaudible to me. I don't have to do a double bind evaluation of SACD
> or 16-bit CD, to know that a 0.0001% THD is inaudible to me.
>
> The same is true for the other measurable parameters.

Exactly, so what *ARE* you measuring that you believe will make a
difference?

Trevor.

[Bob-Stanton]

"Phil" > wrote in message news:<3f221df9$0$20081


> > What is audible has been known for many years.
>
> What level of THD is audible in your opinion then?
>

I use the term "THD" reluctantly, because that is what most people, in
this form, think of as *the* distortion test. Actually THD won't tell
if an amplifier or CD player, has audible distortion. THD will give
the level of the 2nd and 3rd harmonics, for various frequencies. That
is not enough information to predict the audiblity of distortion
generated by a complex waveform, such as music.

To predict audible distortion, we need to measure the level of beats
created by all possible combinations of tones.

> > Again, most of that kind of work was done years ago. The S/N
> > audibility studys that were done long ago (for tape and records) still
> > apply to the the audibility of noise on 24/192.
>
> But ANY audible noise on 24/192 is obviously *NOT* due to limitations of the
> format.
> What are your figures? What is the required listening environment.
>

I'm not going to give figures. Do a Google Audio Group seach on:
'audibility noise'.

As testing methods get better, future listening tests will be required
to evaluate the new types of measurements.

Evaluation of amplifiers or CD players should be on the basis of
measurements, not listening tests. Measurements such as: frequency
response, phase shift, transient response, impulse response, group
delay, white noise, spurious noise (such as hum), jitter, 2nd order
intermodulation distortion, 3rd order intermodulation distortion,
crossmodulation distortion, transient intermodulation distortion,
overload recovery, clipping characteristics, and probably a couple
other things. Of course, after measurements have been done, a
listening test should be the final check.


> Exactly, so what *ARE* you measuring that you believe will make a
> difference?

I just started puting together the software and the hardware necessary
to perform a few, simple, measurments of audio distortion. I haven't
invented any new distortion measuring techniques. I'm just taking
existing techniques, and applying them to audio equipment. Look on
Arny's website and you will see he uses nonstandard (better) measuring
techniques.

Using measurments similar to Arny's, my first test was on the
computer's CD-player through an Audigy-2 sound card . The Audigy-2 sum
and difference beats measured more than 100 dB down (20 to 7000 Hz)!
My $49 CD player, on the other hand, has beats that measured of only
80 dB down. Listening to the two players, they sounded the same. Go
figure :-)

What new tests do I think will make a difference? I like the idea of
testing with forty simultainous tones, one at every 500 Hz, from 500
Hz to 20kHz, and measuring the amount of intermodulaton beat power
that lies "under" each tone. Until I acutally do this test, I will say
only that it looks promising.

Bob Stanton

[Richard D Pierce]

In article >,
Bob-Stanton > wrote:
>"Phil" > wrote in message news:<3f221df9$0$20081
>
>
>> > What is audible has been known for many years.
>>
>> What level of THD is audible in your opinion then?
>>
>
>I use the term "THD" reluctantly, because that is what most people, in
>this form, think of as *the* distortion test. Actually THD won't tell
>if an amplifier or CD player, has audible distortion. THD will give
>the level of the 2nd and 3rd harmonics, for various frequencies.

No it does not. THD gives the sum of ALL non-fundamental
harmonic components, 2dn, 3rd, 4th, and so on. And, when
measured by most analog THD meters, it's actually the sum of ALL
non-fundamental components, including noise, power supply hum,
RFI and so forth.

Granted, there are a couple of computer-based measurement system
that sum only the 2nd and 3rd harmonics, but that's NOT what THD
is.

>That
>is not enough information to predict the audiblity of distortion
>generated by a complex waveform, such as music.

Given what we have learned from Mr. Fourier, can you show any
instance where a component designed for use in a high fidelity
audio application had no THD, yet had large amounts of signal
degradation due to non-linear properties in the amplitude
domain?

>To predict audible distortion, we need to measure the level of beats
>created by all possible combinations of tones.

And, once again, where is the studies that support this
predictability?

>Using measurments similar to Arny's, my first test was on the
>computer's CD-player through an Audigy-2 sound card . The Audigy-2 sum
>and difference beats measured more than 100 dB down (20 to 7000 Hz)!
>My $49 CD player, on the other hand, has beats that measured of only
>80 dB down. Listening to the two players, they sounded the same. Go
>figure :-)

THank you, so you have now quoted a study that seems to refute
your own hypothesis.

>What new tests do I think will make a difference? I like the idea of
>testing with forty simultainous tones, one at every 500 Hz, from 500
>Hz to 20kHz, and measuring the amount of intermodulaton beat power
>that lies "under" each tone. Until I acutally do this test, I will say
>only that it looks promising.

So the beats between tones separated by 500 Hz will be separated
by multiples of 500 Hz, right? How does one, then, distiguish
the original tones from them evil beats that are lurking "under"
each tone?

Multi-tone IM tests, such as spectral contamination and purity
tests, are done using multiple dones that are at "relatively
prime" frequencies, having no common factors or harmonics. At
that point, the effects of non=linearity become obvious:
components in the output appear where there are no components in
the input.

--
| Dick Pierce |
| Professional Audio Development |
| 1-781/826-4953 Voice and FAX |
| |

[Bob-Stanton]

(Richard D Pierce) wrote in message >...
> In article >,


> No it does not. THD gives the sum of ALL non-fundamental
> harmonic components, 2dn, 3rd, 4th, and so on. And, when
> measured by most analog THD meters, it's actually the sum of ALL
> non-fundamental components, including noise, power supply hum,
> RFI and so forth.
>

Yes, I know. Thank you for filling in the little details that I left
out.


> Given what we have learned from Mr. Fourier, can you show any
> instance where a component designed for use in a high fidelity
> audio application had no THD, yet had large amounts of signal
> degradation due to non-linear properties in the amplitude
> domain?
>

Yes.
I have a CD player that measures THD in the 100 dB down range, using
any THD test frequency from 20 Hz to 7500 Hz. Above 7500 we lose the
3rd harmonic so the tests show low THD. With a THD test tone of 13
kHz, the CD player shows very low THD, because the 2nd and 3rd (and
4th and 5th, etc) harmonics are all gone.

But, measure this CD player with two tones (17 kHz and 19 kHz), and
you will see 3rd order beats only 60 dB down. The standard THD test
completely fails to detect this flaw in the player's performance.


> >To predict audible distortion, we need to measure the level of beats
> >created by all possible combinations of tones.
>
> And, once again, where is the studies that support this
> predictability?
>


I think that my statement should have been self evident.

Clue) Each different combination of tones produces a slightly
different level of beats. Typically 17 kHz and 19 kHz, produce higher
level beats than, 17 Hz and 19 Hz.


> >Using measurments similar to Arny's, my first test was on the
> >computer's CD-player through an Audigy-2 sound card . The Audigy-2 sum
> >and difference beats measured more than 100 dB down (20 to 7000 Hz)!
> >My $49 CD player, on the other hand, has beats that measured of only
> >80 dB down. Listening to the two players, they sounded the same. Go
> >figure :-)
>
> THank you, so you have now quoted a study that seems to refute
> your own hypothesis.

Yes, I know. That is why I wrote "Go figure." (Why is it that the
universe won't bend itself, to conform to my theorys?)


> So the beats between tones separated by 500 Hz will be separated
> by multiples of 500 Hz, right? How does one, then, distiguish
> the original tones from them evil beats that are lurking "under"
> each tone?

I know how to do this, but I won't tell. :-)

>
> Multi-tone IM tests, such as spectral contamination and purity
> tests, are done using multiple dones that are at "relatively
> prime" frequencies, having no common factors or harmonics. At
> that point, the effects of non=linearity become obvious:
> components in the output appear where there are no components in
> the input.

Agreed, but that doesn't make it the best possible multi-tone test.

Bob Stanton

[Arny Krueger]

"Bob-Stanton" > wrote in message
m...

> I just started putting together the software and the hardware necessary
> to perform a few, simple, measurements of audio distortion. I haven't
> invented any new distortion measuring techniques. I'm just taking
> existing techniques, and applying them to audio equipment. Look on
> Arny's website and you will see he uses nonstandard (better) measuring
> techniques.

The tests at www.pcavtech.com are informed by about 25 years of DBTs.

> Using measurements similar to Arny's, my first test was on the
> computer's CD-player through an Audigy-2 sound card . The Audigy-2 sum
> and difference beats measured more than 100 dB down (20 to 7000 Hz)!
> My $49 CD player, on the other hand, has beats that measured of only
> 80 dB down. Listening to the two players, they sounded the same. Go
> figure :-)

I know from past experience that it's very hard (impossible?) to set up a
reasonable listening test involving music or natural sounds where a -80 dB
nonlinearity is audible. The practical limit of the detection of nonlinear
distortion under these conditions is more like -60 dB.

If you want to try to hear high frequency nonlinearities, pick a test signal
with lots of high frequency content (say the keys jangling from
www.pcabx.com ) and high pass filter it at say 6-8 KHz. Then record the say
8-22 KHz slice of sound at a high level (CD burner) and play it back on the
suspect CD player.

Nonlinearities in the 10-20 KHz range will cause spurious responses in the
far more audible 0-10 KHz range, hopefully including something in the 3-5
KHz range where the ear is most sensitive and oh, by the way we artificially
ensured that there is no test signal to mask any distortion products.

With all this contrivance and careful mouth holding, your detection
threshold will still be around -60 dB, YMMV. Your fairly crappy -80 dB
player will probably ride free.

I know of no more sensitive listening test to contrive.


> What new tests do I think will make a difference? I like the idea of
> testing with forty simultaneous tones, one at every 500 Hz, from 500
> Hz to 20kHz, and measuring the amount of intermediation beat power
> that lies "under" each tone. Until I actually do this test, I will say
> only that it looks promising.

The problem with test signals that have uniform spacing is that there's this
nasty tendency for the effects of many artifacts from different parts of the
spectrum to show up at the same frequencies, no matter what part of the
spectrum the nonlinearity is in.

From a diagnostic viewpoint, and from the viewpoint of evaluating the
audible significance of a problem, its nice to know in which part(s) of the
spectrum problems the problems lie.

One interesting (and highly workable) approach is to record pink or white
noise that has segments of it filtered out. Record/play it and look for the
segments to be filled in by spurious responses due to nonlinear distortion.
You can make it more diagnostic by only recording noise in certain ranges,
like from 15-20 KHz.

A good tool for making and analyzing test signals like these is Cool Edit
(www.syntrillium.com) . Us the FFT filter with a modest (4096) samples. CE
also has a FFT analyzer that you can use to evaluate your results. You can
also use CE's FFT filter to filter out just the spurious responses, and
then use CE's analysis tool to measure the amplitudes.

[Richard D Pierce]

In article >,
Fred <Fred@once> wrote:
>
>"Richard D Pierce" > wrote in message
...
>> In article >,
>> Bob-Stanton > wrote:
>
>> No it does not. THD gives the sum of ALL non-fundamental
>> harmonic components, 2dn, 3rd, 4th, and so on. And, when
>> measured by most analog THD meters, it's actually the sum of ALL
>> non-fundamental components, including noise, power supply hum,
>> RFI and so forth.
>
>True, but that is a limitation of the measuring device rather than a proper
>definition of THD.

That may be, but until recently ALL THD measurements were THD+N
as a matter of course.

>> Granted, there are a couple of computer-based measurement system
>> that sum only the 2nd and 3rd harmonics, but that's NOT what THD
>> is.
>
>True, and some FFT systems that DO sum all in band harmonics without the
>noise.
>If you measure ALL other components and subtract the fundamental, then you
>are making a S/N+D measurement surely.

Yes, true, and that's what traditional analog-based THD
measurement systems did as a matter of course (common examples
being the venerable HP330, HP334, ST1700, AP System One, and
similar devices by Marconi and the like. The operation was
essentially as described, but in a slightly different order: a
single sine tone was applied to the DUT, then the output was put
through a high-Q notch filter tuned to the frequency of the sine
tone. Whatever was left over was considered distortion.

Now, it may not be a theoretically precise definition of THD, it
is, however, the de facto definition.
--
| Dick Pierce |
| Professional Audio Development |
| 1-781/826-4953 Voice and FAX |
| |

[Bob-Stanton]

"Arny Krueger" > wrote in message


> If you want to try to hear high frequency nonlinearities, pick a test signal
> with lots of high frequency content (say the keys jangling from
> www.pcabx.com ) and high pass filter it at say 6-8 KHz. Then record the say
> 8-22 KHz slice of sound at a high level (CD burner) and play it back on the
> suspect CD player.
>
> Nonlinearities in the 10-20 KHz range will cause spurious responses in the
> far more audible 0-10 KHz range, hopefully including something in the 3-5
> KHz range where the ear is most sensitive and oh, by the way we artificially
> ensured that there is no test signal to mask any distortion products.
>
> With all this contrivance and careful mouth holding, your detection
> threshold will still be around -60 dB, YMMV. Your fairly crappy -80 dB
> player will probably ride free.
>
> I know of no more sensitive listening test to contrive.
>
>



I set up a test for evaluating my collection of old tweeters. Doing
this I accidently stumbled upon an excellent amplifier test.

I made a fourth order, 2 kHz, highpass filter for the tweeters. I ran
my amplifier into an 8 Ohm dummy load and also put the tweeter on the
amplifier. Put (music) test signal into the amplifier and listened to
the output of the tweeter alone. This test work well, showing that
each tweeter had it's own "voice".

Then I decided to test how loud the tweeters would play without
distortion. I turned up the level until I could hear distortion from
the tweeter. The problem was, distortion was audible at only moderate
output power levels. At first, I assumed it was the tweeters causing
the distortion, but I decided to do a second test.

I built an active fourth-order 2kHz, filter and input it in front of
the amplifier. Now the tweeters could play, at an ear piercing levels,
without audible distortion. So, it was the amplifier that was
originally creating the distortion, not the tweeters!



Here is a simple test for amplifier distortion:

---CD player------Amplifier---------------high pass filter--
| |
8 Ohms Tweeter
| |
------------------------------------------------------------
Gnd

(One also could add an L-pad in front of the tweeter.)


Turn up the level (of the music) until you hear distortion coming from
the tweeter. Then put a voltmeter across the 8 Ohm dummy load to get
an idea of the amplifier's output power. (With some amplifiers, you
may be surprised at how low the undistorted output power level is.)

Bob Stanton

[Arny Krueger]

"Bob-Stanton" > wrote in message
om
> "Arny Krueger" > wrote in message
>
>
>> If you want to try to hear high frequency nonlinearities, pick a
>> test signal with lots of high frequency content (say the keys
>> jangling from www.pcabx.com ) and high pass filter it at say 6-8
>> KHz. Then record the say 8-22 KHz slice of sound at a high level (CD
>> burner) and play it back on the suspect CD player.
>>
>> Nonlinearities in the 10-20 KHz range will cause spurious responses
>> in the far more audible 0-10 KHz range, hopefully including
>> something in the 3-5 KHz range where the ear is most sensitive and
>> oh, by the way we artificially ensured that there is no test signal
>> to mask any distortion products.
>>
>> With all this contrivance and careful mouth holding, your detection
>> threshold will still be around -60 dB, YMMV. Your fairly crappy -80
>> dB player will probably ride free.
>>
>> I know of no more sensitive listening test to contrive.

> I set up a test for evaluating my collection of old tweeters. Doing
> this I accidentally stumbled upon an excellent amplifier test.

> I made a fourth order, 2 kHz, highpass filter for the tweeters. I ran
> my amplifier into an 8 Ohm dummy load and also put the tweeter on the
> amplifier. Put (music) test signal into the amplifier and listened to
> the output of the tweeter alone. This test work well, showing that
> each tweeter had it's own "voice".

OK that seems pretty straight-forward.

> Then I decided to test how loud the tweeters would play without
> distortion. I turned up the level until I could hear distortion from
> the tweeter. The problem was, distortion was audible at only moderate
> output power levels. At first, I assumed it was the tweeters causing
> the distortion, but I decided to do a second test.

> I built an active fourth-order 2kHz, filter and input it in front of
> the amplifier. Now the tweeters could play, at an ear piercing levels,
> without audible distortion. So, it was the amplifier that was
> originally creating the distortion, not the tweeters!

I don't see clear causality here because the second active filter was
causing several changes. It was effectively upping the tweeter crossover
from 4th order to 8th order. Furthermore, depending on the impedance curves
of the tweeters, the 4th order passive filter might not have always produced
the expected filter characteristic in terms of voltage at the input
terminals of the tweeter. It might provide a lot less cutoff than expected.

> Here is a simple test for amplifier distortion:

> ---CD player------Amplifier---------------high pass filter--
> | |
> 8 Ohms Tweeter
> | |
> ------------------------------------------------------------
> Gnd

> (One also could add an L-pad in front of the tweeter.)

> Turn up the level (of the music) until you hear distortion coming from
> the tweeter. Then put a voltmeter across the 8 Ohm dummy load to get
> an idea of the amplifier's output power. (With some amplifiers, you
> may be surprised at how low the undistorted output power level is.)

If there's any question, the test I suggested was:

---broadband sound--->--- High pass filter--->---CD
player--->---amplifier-->--- full range speaker

This is then a test of high frequency nonlinear distortion in the CD player
and the amplifier. Nonlinear distortion will cause spurious tones that land
square in the range where the ear is most sensitive, and there is
essentially no music or other sound to interfere with hearing the distortion
products.

If the amount of audible distortion is dependent on amplifier volume control
settings, and actual volume levels remain high enough that Fletcher-Munson
effects remain constant, then the distortion is in the amplifier.

This turns out to be a very sensitive test for amplifier clipping. With 2
KHz and higher high pass filters, it's easy to surprise yourself by clipping
the amplifier at rated power, and yet have sound levels that don't seem to
be that high.

[Bob-Stanton]

"Arny Krueger" > wrote in message news:<Oy-


> I don't see clear causality here because the second active filter was
> causing several changes. It was effectively upping the tweeter crossover
> from 4th order to 8th order. Furthermore, depending on the impedance curves
> of the tweeters, the 4th order passive filter might not have always produced
> the expected filter characteristic in terms of voltage at the input
> terminals of the tweeter. It might provide a lot less cutoff than expected.
>

Actually, I did take out the first filter, when I inserted the second
(active) filter in front of the amplifier. So, the response remained
rolled off by the same 24dB/octave.

>
> ---broadband sound--->--- High pass filter--->---CD
> player--->---amplifier-->--- full range speaker
>

The above not really necessary, since we are looking for a *change* in
audible distortion. Also, I doubt that the CD-player contributes very
much distortion.


> If the amount of audible distortion is dependent on amplifier volume control
> settings, and actual volume levels remain high enough that Fletcher-Munson
> effects remain constant, then the distortion is in the amplifier.

That is why I suggested an using and (optional) L-pad in front of the
tweeter. It's a little bit more work, but the level out of the tweeter
can be kept constant as the volume of the amplifier increases.

> This turns out to be a very sensitive test for amplifier clipping. With 2
> KHz and higher high pass filters, it's easy to surprise yourself by clipping
> the amplifier at rated power, and yet have sound levels that don't seem to
> be that high.

That is right. The change in sound quality from the tweeter is
probably caused by clipping. This test shows what the amplifier can
put out, before clipping. However, you might be surprised to find that
some amplifiers don't put out full rated power before distortion is
audible.

Clipping is not the only thing that could cause this distortion. TIM
could cause it as well. (You say, modern amplfiers don't have TIM?
Good, than they will pass the test with no problem.)

Bob Stanton

[Arny Krueger]

"Bob-Stanton" > wrote in message
om
> "Arny Krueger" > wrote in message news:<Oy-
>
>
>> I don't see clear causality here because the second active filter was
>> causing several changes. It was effectively upping the tweeter
>> crossover from 4th order to 8th order. Furthermore, depending on the
>> impedance curves of the tweeters, the 4th order passive filter might
>> not have always produced the expected filter characteristic in terms
>> of voltage at the input terminals of the tweeter. It might provide a
>> lot less cutoff than expected.

> Actually, I did take out the first filter, when I inserted the second
> (active) filter in front of the amplifier. So, the response remained
> rolled off by the same 24dB/octave.

Providing that the passive filter had the identical same response, which I
will doubt until measurements are shown.

>> ---broadband sound--->--- High pass filter--->---CD
>> player--->---amplifier-->--- full range speaker

> The above not really necessary, since we are looking for a *change* in
> audible distortion.

It is necessary because we're looking for a change that we can attribute to
some UUT, either amplifier, CD player or both.

> Also, I doubt that the CD-player contributes very
> much distortion.

I doubt that either a good amp or a CD player contributes much distortion
since I've done the experiment outlined above with several amps including an
OEM car audio system.

>> If the amount of audible distortion is dependent on amplifier volume
>> control settings, and actual volume levels remain high enough that
>> Fletcher-Munson effects remain constant, then the distortion is in
>> the amplifier.

> That is why I suggested an using and (optional) L-pad in front of the
> tweeter. It's a little bit more work, but the level out of the tweeter
> can be kept constant as the volume of the amplifier increases.

The L-pad changes the impedance of the load on the amplifier, greatly.

>> This turns out to be a very sensitive test for amplifier clipping.
>> With 2 KHz and higher high pass filters, it's easy to surprise
>> yourself by clipping the amplifier at rated power, and yet have
>> sound levels that don't seem to be that high.

> That is right. The change in sound quality from the tweeter is
> probably caused by clipping. This test shows what the amplifier can
> put out, before clipping. However, you might be surprised to find that
> some amplifiers don't put out full rated power before distortion is
> audible.

There are some issues related to measuring peak power versus average power.

> Clipping is not the only thing that could cause this distortion. TIM
> could cause it as well. (You say, modern amplifiers don't have TIM?
> Good, than they will pass the test with no problem.)

No, I say that TIM is an obscure way to talk about high frequency nonlinear
distortion.

[Bob-Stanton]

"Arny Krueger" > wrote in message

> > Actually, I did take out the first filter, when I inserted the second
> > (active) filter in front of the amplifier. So, the response remained
> > rolled off by the same 24dB/octave.
>
> Providing that the passive filter had the identical same response, which I
> will doubt until measurements are shown.

You don't need measurements if you know filter theory. Both the active
and passive versions of the 24 db/octave Linkwitz will have the same
response, except for two factors:
1) The finite Q of the coils and inductors.
2) The impedance change of the speaker load.

Zobels and other networks can make the tweeter impedance constant.

So, the only difference in the filter response will be due to the
finite Q of the inductors and capacitors.



Below is a chart of an ideal filter response, for a 24 dB/octave
Linkwitz high pass, and of the calculated actual response of a passive
Linkwitz:

The calculated actual response assumes you use Madisound Air-Core
inductors.


Frequency Ideal Attenuation Pratical Attenuation
in Hz. in dB in dB

2000 -6.02 -6.71
1750 -8.64 -9.31
1500 -12.38 -12.98
1250 -17.56 -18.09
1000 -24.61 -25.12
750 -34.25 -34.76
500 -48.20 -48.62
250 -72.25 -71.83

You can see the rolloff curve of the passive filter, is the same as
the ideal filter's, but with a added 0.7 insertion loss.




>
> >> ---broadband sound--->--- High pass filter--->---CD
> >> player--->---amplifier-->--- full range speaker
>
> > The above not really necessary, since we are looking for a *change* in
> > audible distortion.
>
> It is necessary because we're looking for a change that we can attribute to
> some UUT, either amplifier, CD player or both.
>

No, the distortion of the CD player and the tweeter stay the same
during the test. The only thing that changes is the distortion of the
amplifier.


> > Also, I doubt that the CD-player contributes very
> > much distortion.
>
> I doubt that either a good amp or a CD player contributes much distortion
> since I've done the experiment outlined above with several amps including an
> OEM car audio system.
>

Well, that is why we do the tests, to find out if the amplifier has
audible distortion. I'm glad your amplifiers did well.



> >> If the amount of audible distortion is dependent on amplifier volume
> >> control settings, and actual volume levels remain high enough that
> >> Fletcher-Munson effects remain constant, then the distortion is in
> >> the amplifier.
>
> > That is why I suggested an using and (optional) L-pad in front of the
> > tweeter. It's a little bit more work, but the level out of the tweeter
> > can be kept constant as the volume of the amplifier increases.
>
> The L-pad changes the impedance of the load on the amplifier, greatly.
>

No. L-pads don't change the impedance at all. Here are the values for
8 Ohm L-pads, for various attenuations.

Attenuation R(series) R(parallel)
in dB in Ohms in Ohms

1 0.87 65.0
3 2.34 19.4
6 4.0 8.0
10 5.47 3.7

If you chech this out, you will see the impedance stays 8 Ohms.



> No, I say that TIM is an obscure way to talk about high frequency nonlinear
> distortion.

That is because, it is an obscure form of distortion. :-)

Bob Stanton

[Arny Krueger]

"Bob-Stanton" > wrote in message
om...
> "Arny Krueger" > wrote in message
>
> > > Actually, I did take out the first filter, when I inserted the second
> > > (active) filter in front of the amplifier. So, the response remained
> > > rolled off by the same 24dB/octave.
> >
> > Providing that the passive filter had the identical same response, which
I
> > will doubt until measurements are shown.
>
> You don't need measurements if you know filter theory. Both the active
> and passive versions of the 24 db/octave Linkwitz will have the same
> response, except for two factors:

> 1) The finite Q of the coils and inductors.
> 2) The impedance change of the speaker load.

right. Hold that thought about the tweeter.

> Zobels and other networks can make the tweeter impedance constant.

Not necessarily perfectly constant in both magnitude and phase at all
frequencies without a ton of work. For example, modelling the inductance of
a tweeter as being a pure inductance, which is essentially what simple zobel
calcuations do, is not a particularly good approximation for most tweeters.

See http://users.ece.gatech.edu/~mleach/ece4445/downloads/zobel.pdf for
additional caveats. Notice that his plots only show the magnitude of the
impedance and not its phase angle. There's a slight dip and a downward slope
in most of his magnitude plots, and that suggests non-zero phase.

> So, the only difference in the filter response will be due to the
> finite Q of the inductors and capacitors.

Hopefully.

> Below is a chart of an ideal filter response, for a 24 dB/octave
> Linkwitz high pass, and of the calculated actual response of a passive
> Linkwitz:
>
> The calculated actual response assumes you use Madisound Air-Core
> inductors.
>
>
> Frequency Ideal Attenuation Pratical Attenuation
> in Hz. in dB in dB
>
> 2000 -6.02 -6.71
> 1750 -8.64 -9.31
> 1500 -12.38 -12.98
> 1250 -17.56 -18.09
> 1000 -24.61 -25.12
> 750 -34.25 -34.76
> 500 -48.20 -48.62
> 250 -72.25 -71.83
>
> You can see the rolloff curve of the passive filter, is the same as
> the ideal filter's, but with a added 0.7 insertion loss.
>
>
>
>
> >
> > >> ---broadband sound--->--- High pass filter--->---CD
> > >> player--->---amplifier-->--- full range speaker
> >
> > > The above not really necessary, since we are looking for a *change* in
> > > audible distortion.
> >
> > It is necessary because we're looking for a change that we can attribute
to
> > some UUT, either amplifier, CD player or both.
> >
>
> No, the distortion of the CD player and the tweeter stay the same
> during the test. The only thing that changes is the distortion of the
> amplifier.
>
>
> > > Also, I doubt that the CD-player contributes very
> > > much distortion.
> >
> > I doubt that either a good amp or a CD player contributes much
distortion
> > since I've done the experiment outlined above with several amps
including an
> > OEM car audio system.
> >
>
> Well, that is why we do the tests, to find out if the amplifier has
> audible distortion. I'm glad your amplifiers did well.
>
>
>
> > >> If the amount of audible distortion is dependent on amplifier volume
> > >> control settings, and actual volume levels remain high enough that
> > >> Fletcher-Munson effects remain constant, then the distortion is in
> > >> the amplifier.
> >
> > > That is why I suggested an using and (optional) L-pad in front of the
> > > tweeter. It's a little bit more work, but the level out of the tweeter
> > > can be kept constant as the volume of the amplifier increases.
> >
> > The L-pad changes the impedance of the load on the amplifier, greatly.

> No. L-pads don't change the impedance at all. Here are the values for
> 8 Ohm L-pads, for various attenuations.

> Attenuation R(series) R(parallel)
> in dB in Ohms in Ohms

> 1 0.87 65.0
> 3 2.34 19.4
> 6 4.0 8.0
> 10 5.47 3.7

But the load isn't a pure 8 ohms resistive. It's a tweeter with variable
impedance magnitude and phase, even with a zobel. The L-Pad isolates the
input impedance of the tweeter from the amplifier. This in turn causes the
attenuation of the filter to wander around.

> If you check this out, you will see the impedance stays 8 Ohms.

Presuming a pure 8 ohm resistive load, which is not what a practical tweeter
will be at all frequencies, zobel notwithstanding.

> > No, I say that TIM is an obscure way to talk about high frequency
nonlinear
> > distortion.

> That is because, it is an obscure form of distortion. :-)

It's actually very common and easy to understand if you call it by its
conventional name.

[Bob-Stanton]

"Arny Krueger" > wrote in message news:<vPadnb-


> > Zobels and other networks can make the tweeter impedance constant.
>
> Not necessarily perfectly constant in both magnitude and phase at all
> frequencies without a ton of work. For example, modelling the inductance of
> a tweeter as being a pure inductance, which is essentially what simple zobel
> calcuations do, is not a particularly good approximation for most tweeters.
>
> See http://users.ece.gatech.edu/~mleach/ece4445/downloads/zobel.pdf for
> additional caveats. Notice that his plots only show the magnitude of the
> impedance and not its phase angle. There's a slight dip and a downward slope
> in most of his magnitude plots, and that suggests non-zero phase.
>
> > So, the only difference in the filter response will be due to the
> > finite Q of the inductors and capacitors.
>
> Hopefully.
>

You are right that making a high-pass filter compensate for the
impedance changes of a driver, is a *hell of lot* of work.
Fortunately, it is unnessary. The problem can be obviated by starting
the test with the L-pad set to 10 dB (or 15 dB!). That way, the
impedance the filter sees, will always be 8 Ohms. (You probably would
want to do it anyway. Than you wouldn't need to use ear pluges.)



> > No. L-pads don't change the impedance at all. Here are the values for
> > 8 Ohm L-pads, for various attenuations.
>
> > Attenuation R(series) R(parallel)
> > in dB in Ohms in Ohms
>
> > 1 0.87 65.0
> > 3 2.34 19.4
> > 6 4.0 8.0
> > 10 5.47 3.7
>
> But the load isn't a pure 8 ohms resistive. It's a tweeter with variable
> impedance magnitude and phase, even with a zobel. The L-Pad isolates the
> input impedance of the tweeter from the amplifier.

> This in turn causes the
> attenuation of the filter to wander around.
>

Not if the L-pad starting attenuation is 15 dB.


> > That is because, it is an obscure form of distortion. :-)
>
> It's actually very common and easy to understand if you call it by its
> conventional name.

Than why didn't you call it by it's proper name?

Bob Stanton

[Arny Krueger]

"Bob-Stanton" > wrote in message
om
> "Arny Krueger" > wrote in message news:<vPadnb-

>>> Zobels and other networks can make the tweeter impedance constant.

>> Not necessarily perfectly constant in both magnitude and phase at all
>> frequencies without a ton of work. For example, modelling the
>> inductance of a tweeter as being a pure inductance, which is
>> essentially what simple zobel calcuations do, is not a particularly
>> good approximation for most tweeters.

>> See http://users.ece.gatech.edu/~mleach/ece4445/downloads/zobel.pdf
>> for additional caveats. Notice that his plots only show the
>> magnitude of the impedance and not its phase angle. There's a slight
>> dip and a downward slope in most of his magnitude plots, and that
>> suggests non-zero phase.

>>> So, the only difference in the filter response will be due to the
>>> finite Q of the inductors and capacitors.

>> Hopefully.

> You are right that making a high-pass filter compensate for the
> impedance changes of a driver, is a *hell of lot* of work.
> Fortunately, it is unnecessary. The problem can be obviated by starting
> the test with the L-pad set to 10 dB (or 15 dB!). That way, the
> impedance the filter sees, will always be 8 Ohms. (You probably would
> want to do it anyway. Than you wouldn't need to use ear plugs.)

Oh, I guess so. This is, I provided a more straightforward methodology some
posts back.

>>> No. L-pads don't change the impedance at all. Here are the values
>>> for 8 Ohm L-pads, for various attenuations.
>>
>>> Attenuation R(series) R(parallel)
>>> in dB in Ohms in Ohms
>>
>>> 1 0.87 65.0
>>> 3 2.34 19.4
>>> 6 4.0 8.0
>>> 10 5.47 3.7

>> But the load isn't a pure 8 ohms resistive. It's a tweeter with
>> variable impedance magnitude and phase, even with a zobel. The
>> L-Pad isolates the input impedance of the tweeter from the amplifier.

>> This in turn causes the
>> attenuation of the filter to wander around.

> Not if the L-pad starting attenuation is 15 dB.

I guess, but....

If we go back a number of posts, I provided a means that works with a lot
less fooling around.

>>> That is because, it is an obscure form of distortion. :-)
>>
>> It's actually very common and easy to understand if you call it by
>> its conventional name.
>
> Than why didn't you call it by it's proper name?

I do.

[Bob-Stanton]

"Arny Krueger" > wrote in message news:<N8udnSIGNbzHVLeiU-


> I guess, but....
>
> If we go back a number of posts, I provided a means that works with a lot
> less fooling around.
>
Which is what?



I don't know if anything could be simpler than what I proposed. The
best test signal is music. The best test instrument is the human ear.

I think I might build this distortion tester. If someone elso wants to
build it, here is the schemetic:

DISTORTION TESTER


| Dummy Load | Highpass Filter | L-pad |Speaker

Amplifier-----------------------10 uF-----22uF---10 Ohms--
(D.U.T.) | | | | | |
10 Ohms 10 Ohms 10 Ohms | -10 Ohms----------
| | | 0.55 mH | | |
| | | | 10 10 Spk
10 Ohms 10 Ohms 10 Ohms | | | |
| | | | | | |
--------------------------------------------------------------------

All resistors are 10 Ohm, 10 Watt, from Radio Shack ($0.99)
Capacitors are 50 V, N.P., from Radio Shack ($0.99)
Speaker is any small speaker the goes up to 8 kHz or 9 kHz.
Inductor is approximately 120 turns, 22 gage wire, on 1.5 in dia form.


I don't know what could be simpler, anyone could built it. (Even me.)





> > Than why didn't you call it by it's proper name?
>
> I do.

Which is? (Don't say, "slew-rate")

Bob Stanton

[Arny Krueger]

"Bob-Stanton" > wrote in message
om
> "Arny Krueger" > wrote in message
> news:<N8udnSIGNbzHVLeiU-
>
>
>> I guess, but....
>>
>> If we go back a number of posts, I provided a means that works with
>> a lot less fooling around.

> Which is what?


"Arny Krueger" > wrote in message


> If you want to try to hear high frequency nonlinearities, pick a test
signal
> with lots of high frequency content (say the keys jangling from
> www.pcabx.com ) and high pass filter it at say 6-8 KHz. Then record the
say
> 8-22 KHz slice of sound at a high level (CD burner) and play it back on
the
> suspect CD player.

If the CD player is good, then this becomes a test of the power amp and
speaker. Power amps are the topic of the thread.

> I don't know if anything could be simpler than what I proposed.

Sure, no equipment at all is simpler than any equipment.

>The best test signal is music.

Hence the reference to test sounds from pcabx.com

>The best test instrument is the human ear.

That's what I recommended using.

> I think I might build this distortion tester. If someone elso wants to
> build it, here is the schemetic:
>
> DISTORTION TESTER
>
>
> | Dummy Load | Highpass Filter | L-pad |Speaker
>
> Amplifier-----------------------10 uF-----22uF---10 Ohms--
> (D.U.T.) | | | | | |
> 10 Ohms 10 Ohms 10 Ohms | -10 Ohms----------
> | | | 0.55 mH | | |
> | | | | 10 10 Spk
> 10 Ohms 10 Ohms 10 Ohms | | | |
> | | | | | | |
> --------------------------------------------------------------------
>
> All resistors are 10 Ohm, 10 Watt, from Radio Shack ($0.99)
> Capacitors are 50 V, N.P., from Radio Shack ($0.99)
> Speaker is any small speaker the goes up to 8 kHz or 9 kHz.
> Inductor is approximately 120 turns, 22 gage wire, on 1.5 in dia form.

> I don't know what could be simpler, anyone could built it. (Even me.)

My solution requires building nothing at all.

[Bob-Stanton]

"Arny Krueger" > wrote in message

> > I don't know what could be simpler, anyone could built it. (Even me.)
>
> My solution requires building nothing at all.

Like the great man once said: "Things should be as simple as possible,
but not simpler."

Bob Stanton

[Arny Krueger]

"Tony Pearce" > wrote in message
u
> "Bob-Stanton" > wrote in message
> om...
>
>> The best test instrument is the human ear.
>
> A quaint notion, but not able to be substantiated these days.

Admittedly not the most sensitive solution, but still unbeatable in terms of
evaluating audibility.

>> Which is? (Don't say, "slew-rate")

> How about Slew Induced Distortion?

How about nonlinear distortion at high frequencies?

[Bob-Stanton]

"Tony Pearce" > wrote in message >...
> "Bob-Stanton" > wrote in message
> om...
>
> > The best test instrument is the human ear.
>
> A quaint notion, but not able to be substantiated these days.
>

Yes, a test instrument, such as a spectrum analyzer, can measure
things that are beyond the threshold of human ear. It can measure
noise , distortion, flatness, and transient response, beter than any
human ear.

Test instruments are more sensitive than human ears, but they don't
understand what they "hear". Can test instruments tell us the best
place to put a microphone, for the most muscically balanced sound? No.

Test instruments can "hear" the reverb of concerts halls. Can they
tell the difference between a great concert hall, and a less than
great concert hall? I don't think so. I don't think we have developed
our understanding of reverb data to that point. If we did, we could
make every hall concert hall, a "great" concert hall, with just a few
measurements and a few tweeks.

A spectrum analyzer can record all the data defining how something
sounds. But with a complex signal, such as an orchestra playing in a
concert hall, the data is overwelming. We simply can't look at the
data and tell if the orchestra sounds good or bad. But the human ear
"can". So, the human ear remains the best *overall* test instrument.

Bob Stanton

[Tony Pearce]

"Bob-Stanton" > wrote in message
om...
> "Tony Pearce" > wrote in message
>...
> > "Bob-Stanton" > wrote in message
> > om...
> >
> > > The best test instrument is the human ear.
> >
> > A quaint notion, but not able to be substantiated these days.
> >
>
> Yes, a test instrument, such as a spectrum analyzer, can measure
> things that are beyond the threshold of human ear. It can measure
> noise , distortion, flatness, and transient response, beter than any
> human ear.
>
> Test instruments are more sensitive than human ears, but they don't
> understand what they "hear". Can test instruments tell us the best
> place to put a microphone, for the most muscically balanced sound?

Yes. But first you have to define "most musically balanced sound".

> Test instruments can "hear" the reverb of concerts halls. Can they
> tell the difference between a great concert hall, and a less than
> great concert hall? I don't think so. I don't think we have developed
> our understanding of reverb data to that point. If we did, we could
> make every hall concert hall, a "great" concert hall, with just a few
> measurements and a few tweeks.

Even using ears you will not get complete agreement on that, but test
equipment will show you what the differences are.

> A spectrum analyzer can record all the data defining how something
> sounds. But with a complex signal, such as an orchestra playing in a
> concert hall, the data is overwelming. We simply can't look at the
> data and tell if the orchestra sounds good or bad. But the human ear
> "can". So, the human ear remains the best *overall* test instrument.

But YOU specifically use the term "TEST instrument" which the ear is NOT.

For determining a persons SUBJECTIVE audibility, their ears/auditory system
cannot be beat. Of course the result only applies to that particular person,
which is why OBJECTIVE test measurements are so much better for many
purposes, but not all I agree.

TonyP.

[Arny Krueger]

"Tony Pearce" > wrote in message


> For determining a persons SUBJECTIVE audibility, their ears/auditory
> system cannot be beat. Of course the result only applies to that
> particular person, which is why OBJECTIVE test measurements are so
> much better for many purposes, but not all I agree.

I see other strong advantages to test equipment-based evaluation:

(1) Fast.

(2) Useful for testing equipment that is used repetitively or in cascade,
and therefore must have performance that is integer multiples better than
merely sonically transparent.


The test suite at my PCAVTech web site was designed to detect both *normal*
equipment deficiencies and also a wide range of pathological deficiencies
like relatively narrow-band performance problems.

The "Audio Rightmark" test suite cuts the PCAVTech test suite down somewhat
further, and in the process leaves a few more doors open to more
pathologies, but remains complete enough to be useful and interesting.

[Bob-Stanton]

((null)) wrote in message news:<1060711324.886040@smirk>...


> Of course. The power distribution of music is mostly below 2kHz,
> so if you remove everything below 2kHz before it hits the amplifier,
> you'll have quite a bit more power above 2kHz to send to the tweeter
> before clipping; I'd guess at least 10 dB more than in the case of
> a wideband signal sent to a passive crossover.
>
>

That's right.

Also if the tweeter is powered by it's own (lower power) amplifier, it
can be safe from burnout. When a single (high power) amplifier is
driven into overload, a lot of high frequency energy (splatter) is
sent into the tweeter, sometimes enough to burn it out. With an
electronic crossover, even if the amplifiers are both driven into
overload, the tweeter is safe, as long as the high frequency amplifier
is of a low enough power.

I think the best crossover is a 24 dB/octave (Linkwitz), consisting of
a 12dB/ octave high level (electronic crossover), and 12 dB/octave
passive crossover on the drivers.

If the crossover is a combination of electronic and passive, there is
an advantage: When the low frequency amplifier is driven in to
overload, the high frequency splatter (caused by the overload) is
largely prevented from getting into the woofer. This gives the system
better overload characteristics that a single amplifier system could
have.

Bob Stanton