Hello, all--

I hesitate to bring up this topic while the Doppler wars are still in an
unsettled state (please, guys, don't bring it here), but I've spent much
of this past week trying to learn about and measure the "transient
intermodulation distortion" and/or "slewing-induced distortion" of some
recording-related equipment, and would like to report in.

This is something I've been wanting to come to terms with for years. I
began by going to the AES Web site and downloading and reading the papers
of Prof. Matti Otala from the 1970s and early 1980s.

More up-to-date analyses of Otala's work have shown that what his tests
show is not limited only to dynamic distortion. And with the equipment I
have available, the best signal source I could create doesn't have nearly
as high a slewing rate as either of Otala's two main recommended test
signals. So my results are probably more like those of ordinary IM tests,
only weighted somewhat more toward revealing dynamic distortions.

I'm virtually certain that better-controlled test signals could provoke
worse behavior from some circuits, and I'd really like to see that, since
it could well reveal even further differences among different circuits.
Still, even with what I have, I was able to find some clear distinctions
among the distortion performance of different pieces of equipment.

That's about all I can say because of the limits of my test signals. OK,
I can say a few things in the more clear-cut cases:

- The ART "Tube MP" is grotesquely bad. By a considerable margin it had
the highest distortion, in its normal operating range, of any equipment
I tested. And the Aphex "Tubessence" preamp (model 107, not the 1100)
had second highest.

- My better preamps (Millenia Media HV-3B, FMR "Real Nice Preamp", Grace
Lunatec V2 and V3, M Audio DMP3, Sonosax SX-M2, Symetrix SX-202, and a
dbx 760X modified by Jim Williams) all had low noise and distortion when
stimulated by this test signal, with small differences in favor of the
Millenia Media preamp. They had distinctly different distortion spectra,
but were more or less similar in the total amount of their distortion.

- Despite its somewhat higher EIN specification, the noise/distortion
floor of the RNP in the presence of the test signal really wasn't
noticeably higher than that of most of the other good preamps.

- The M Audio "Flying Calf" A/D converter, surprisingly, held its own
against far more expensive converters, while the D/A section of the
"Flying Cow" converter was not nearly as good--though my "Cow" is an
older model, and the chips (cow chips?) in their newer models have a
higher specification.

David Satz wrote:
That's about all I can say because of the limits of my test signals. OK,
I can say a few things in the more clear-cut cases:

- The ART "Tube MP" is grotesquely bad. By a considerable margin it had
the highest distortion, in its normal operating range, of any equipment
I tested. And the Aphex "Tubessence" preamp (model 107, not the 1100)
had second highest.

Open up the 107 and you'll find a pot in the middle of the RPA circuit.
That pot is preset at the factory for a particular THD level. You can
turn it up and down if you want more or less distortion.

- My better preamps (Millenia Media HV-3B, FMR "Real Nice Preamp", Grace
Lunatec V2 and V3, M Audio DMP3, Sonosax SX-M2, Symetrix SX-202, and a
dbx 760X modified by Jim Williams) all had low noise and distortion when
stimulated by this test signal, with small differences in favor of the
Millenia Media preamp. They had distinctly different distortion spectra,
but were more or less similar in the total amount of their distortion.

I can believe this. Now, what happens if you put a 1:1 transformer in
front of one of these? The transformer should be much more of a source
of distortion.

- The M Audio "Flying Calf" A/D converter, surprisingly, held its own
against far more expensive converters, while the D/A section of the
"Flying Cow" converter was not nearly as good--though my "Cow" is an
older model, and the chips (cow chips?) in their newer models have a
higher specification.

There are also some board layout changes too, which may make for a change in
stability.
--scott
--
"C'est un Nagra. C'est suisse, et tres, tres precis."

"Scott Dorsey" wrote in message

David Satz wrote:

- My better preamps (Millenia Media HV-3B, FMR "Real Nice Preamp",
Grace Lunatec V2 and V3, M Audio DMP3, Sonosax SX-M2, Symetrix
SX-202, and a dbx 760X modified by Jim Williams) all had low noise
and distortion when stimulated by this test signal, with small
differences in favor of the Millenia Media preamp. They had
distinctly different distortion spectra, but were more or less
similar in the total amount of their distortion.

I can believe this. Now, what happens if you put a 1:1 transformer in
front of one of these? The transformer should be much more of a
source of distortion.

Indeed.

- The M Audio "Flying Calf" A/D converter, surprisingly, held its own
against far more expensive converters, while the D/A section of the
"Flying Cow" converter was not nearly as good--though my "Cow" is an
older model, and the chips (cow chips?) in their newer models have a
higher specification.

There are also some board layout changes too, which may make for a
change in stability.

I tested two different revisions of the Flying Cow:

http://www.pcavtech.com/adc-dac/Cow/index.htm

The "Revision E" was pretty impressive, and says a lot for the philosophy of
continual evolutionary upgrades at M-Audio,

"David Satz" wrote in message
om...
More up-to-date analyses of Otala's work have shown that what his tests
show is not limited only to dynamic distortion. And with the equipment I
have available, the best signal source I could create doesn't have nearly
as high a slewing rate as either of Otala's two main recommended test
signals. So my results are probably more like those of ordinary IM tests,
only weighted somewhat more toward revealing dynamic distortions.

I'm virtually certain that better-controlled test signals could provoke
worse behavior from some circuits, and I'd really like to see that, since
it could well reveal even further differences among different circuits.
Still, even with what I have, I was able to find some clear distinctions
among the distortion performance of different pieces of equipment.

That's about all I can say because of the limits of my test signals. OK,
I can say a few things in the more clear-cut cases:

- The ART "Tube MP" is grotesquely bad. By a considerable margin it had
the highest distortion, in its normal operating range, of any equipment
I tested. And the Aphex "Tubessence" preamp (model 107, not the 1100)
had second highest.

[snipped the rest]

David, it's not clear from your post -- were you using Otala's sine/square
test signal? Or were you doing a different sort of IM test?

Peace,
Paul

Paul Stamler wrote:

David, it's not clear from your post -- were you using Otala's
sine/square test signal? Or were you doing a different sort of IM test?

I was trying to come as close as possible to Otala's DIM 30 sine/square
test signal. But I wasn't able to duplicate it with the equipment I
have available, and in fact don't know the actual maximum slew rate of
the test signal that I ended up with--surely less than intended, though.

That's why I'm offering no numbers, and why I suspect that the test could
have been made quite a bit more sensitive.

If you or anyone else here has suggestions for affordable ways to generate
a DIM 30 signal, I'd appreciate them greatly. (DIM 30 = 3.18 kHz square
wave + 15 kHz sine wave in a 4:1 ratio, low-pass filtered at 30 kHz --
see Leinonen, Otala and Curl 1976, "Method for Measuring Transient
Intermodulation Distortion," AES preprint #1185.)

--best regards

In article writes:

If you or anyone else here has suggestions for affordable ways to generate
a DIM 30 signal, I'd appreciate them greatly. (DIM 30 = 3.18 kHz square
wave + 15 kHz sine wave in a 4:1 ratio, low-pass filtered at 30 kHz --

A good analog mixer, perhaps? But I don't know how good (or in what
way) it would have to be to not contribute some of the distortion you
want to measure with the test signal.

Seems like it should be simple with any multitrack DAW program running
at 96 kHz. Must be at least about 50,000 people around here who have
one.



--
I'm really Mike Rivers )
However, until the spam goes away or Hell freezes over,
lots of IP addresses are blocked from this system. If
you e-mail me and it bounces, use your secret decoder ring
and reach me he double-m-eleven-double-zero at yahoo

"David Satz" wrote in message
om...
Paul Stamler wrote:
I was trying to come as close as possible to Otala's DIM 30 sine/square
test signal. But I wasn't able to duplicate it with the equipment I
have available, and in fact don't know the actual maximum slew rate of
the test signal that I ended up with--surely less than intended, though.

That's why I'm offering no numbers, and why I suspect that the test could
have been made quite a bit more sensitive.

If you or anyone else here has suggestions for affordable ways to generate
a DIM 30 signal, I'd appreciate them greatly. (DIM 30 = 3.18 kHz square
wave + 15 kHz sine wave in a 4:1 ratio, low-pass filtered at 30 kHz --
see Leinonen, Otala and Curl 1976, "Method for Measuring Transient
Intermodulation Distortion," AES preprint #1185.)

You can generate an okay square wave using a 555 timer chip, probably
adequate for the task. And you can make quite a good sine wave oscillator
using the circuit on p. 186 of Jung's "Audio IC Op-Amp Applications"; use
5532 chips. Sum the two signals at the appropriate ratio in an active
summing amp made from the spare side of one of the 5532s, or an LM318;
bandlimit the summing amp to 30kHz using a cap in parallel with the feedback
resistor, and Bob's your uncle. It's not a perfect signal source, but it'll
probably have less residual distortion than most of the stuff you're
measuring. For mike pres, use a good-quality passive DI (I've found those
contribute very little extra distortion in HF IM tests, and my guess is that
they'll do the same in a DIM test).

Peace,
Paul

I wrote:

- The M Audio "Flying Calf" A/D converter, surprisingly, held its own
against far more expensive converters, while the D/A section of the
"Flying Cow" converter was not nearly as good--though my "Cow" is an
older model, and the chips (cow chips?) in their newer models have a
higher specification.


whereupon Arny Krueger wrote:

There are also some board layout changes too, which may make for a
change in stability.

I tested two different revisions of the Flying Cow:

http://www.pcavtech.com/adc-dac/Cow/index.htm

The "Revision E" was pretty impressive, and says a lot for the philosophy
of continual evolutionary upgrades at M-Audio,

The Cow that I tested has a Rev. E board also. Its D/A wasn't nearly as
clean as, for example, the line output of a ~$100, 10-year-old Sony D-131
"Discman" portable CD player(!). I didn't test the Cow's A/D section.

The product that did test well from this series was a "Flying Calf" A/D,
labeled on the outside as the 20-bit version of this product and having
a "Rev. A" circuit board, also (c) 1997. It actually held its own fairly
well against a Lucid AD 9624 and a somewhat older Mytek (model AD 2018).

Perhaps the A/D section of the Flying Cow would have done equally well.

--best regards

P.S.: The equipment that I tested was all my own stuff, which I bought
at various times for various amounts, under various suppositions as to
what level of performance each piece of equipment would offer. I have
to admit that I found I had a real "vested interest" in the outcome of
these tests--I really wanted to see each piece of equipment perform as
well or as badly I had thought it should do.

This turned out not always to be the case. And after I had made graphs
of the noise and distortion for seven or eight preamps, for example, at
some moments I found myself staring again and again at certain graphs,
looking for the evidence that I felt _had_ to be there establishing the
superiority or inferiority of one or another item. I really wanted the
(x) preamp to be better than the (y) preamp, say, in part because I was
prepared to feel cool using preamp (x) at recording jobs, etc.

It was disturbing to have to accept that my mental picture of a product's
level of quality can be quite mistaken and even inauthentic, i.e. formed
on the basis of a product's "public image" (e.g. in this newsgroup) and,
perhaps, the "back story" of how I came to own it, for example. It
really took me by surprise how much all that cr@p still persists in my
way of thinking. Maybe I'm human after all ...

"David Satz" wrote in message
om
I wrote:

- The M Audio "Flying Calf" A/D converter, surprisingly, held its own
against far more expensive converters, while the D/A section of the
"Flying Cow" converter was not nearly as good--though my "Cow" is an
older model, and the chips (cow chips?) in their newer models have a
higher specification.


whereupon Arny Krueger wrote:

There are also some board layout changes too, which may make for a
change in stability.

I tested two different revisions of the Flying Cow:

http://www.pcavtech.com/adc-dac/Cow/index.htm

The "Revision E" was pretty impressive, and says a lot for the
philosophy of continual evolutionary upgrades at M-Audio,

The Cow that I tested has a Rev. E board also. Its D/A wasn't nearly
as clean as, for example, the line output of a ~$100, 10-year-old
Sony D-131 "Discman" portable CD player(!).

I tested a number of Sony Discman players and posted the results at

http://www.pcavtech.com/play-rec/summary/index.htm

None of them came close to a Rev E Flying Cow, nor was it a fair comparison
given that the Cow was capable of handling 24 bit data, and any CD player
is limited to less than 16 bits worth of actual performance.


I didn't test the Cow's A/D section.

My tests are for loop-back, so the noise and distortion of both are
represented by one test and one set of performance numbers.

David,

Dynamic distortion - something I've been thinking about recently too.
Too bad this specification doesn't make the A-list in most
manufacturer literature.

I could be way off here, why couldn't you just create the individual
square and sine waveforms in something like Wavelab, bandwidth limit,
and then mix the waves together at the desired ratio? Then just
perform a loop test with them?

Harry



On 20 Aug 2004 08:32:00 -0700, (David Satz) wrote:

Hello, all--

I hesitate to bring up this topic while the Doppler wars are still in an
unsettled state (please, guys, don't bring it here), but I've spent much
of this past week trying to learn about and measure the "transient
intermodulation distortion" and/or "slewing-induced distortion" of some
recording-related equipment, and would like to report in.

This is something I've been wanting to come to terms with for years. I
began by going to the AES Web site and downloading and reading the papers
of Prof. Matti Otala from the 1970s and early 1980s.

More up-to-date analyses of Otala's work have shown that what his tests
show is not limited only to dynamic distortion. And with the equipment I
have available, the best signal source I could create doesn't have nearly
as high a slewing rate as either of Otala's two main recommended test
signals. So my results are probably more like those of ordinary IM tests,
only weighted somewhat more toward revealing dynamic distortions.

I'm virtually certain that better-controlled test signals could provoke
worse behavior from some circuits, and I'd really like to see that, since
it could well reveal even further differences among different circuits.
Still, even with what I have, I was able to find some clear distinctions
among the distortion performance of different pieces of equipment.

That's about all I can say because of the limits of my test signals. OK,
I can say a few things in the more clear-cut cases:

- The ART "Tube MP" is grotesquely bad. By a considerable margin it had
the highest distortion, in its normal operating range, of any equipment
I tested. And the Aphex "Tubessence" preamp (model 107, not the 1100)
had second highest.

- My better preamps (Millenia Media HV-3B, FMR "Real Nice Preamp", Grace
Lunatec V2 and V3, M Audio DMP3, Sonosax SX-M2, Symetrix SX-202, and a
dbx 760X modified by Jim Williams) all had low noise and distortion when
stimulated by this test signal, with small differences in favor of the
Millenia Media preamp. They had distinctly different distortion spectra,
but were more or less similar in the total amount of their distortion.

- Despite its somewhat higher EIN specification, the noise/distortion
floor of the RNP in the presence of the test signal really wasn't
noticeably higher than that of most of the other good preamps.

- The M Audio "Flying Calf" A/D converter, surprisingly, held its own
against far more expensive converters, while the D/A section of the
"Flying Cow" converter was not nearly as good--though my "Cow" is an
older model, and the chips (cow chips?) in their newer models have a
higher specification.

Dave,

the DIM 30 test signal as you described it passes through a 30 kHz low
pass filter. Therefore almost any square wave signal generator will
have a suffienctly fast rise and fall time i.e. slew rate for this test
since it will be limited by the 30 kHz LPF anyway. What makes you
think the slew rate of your test signal was not good enough? How did
you generate the square wave? Even if you had a "perfect"square wave,
after passing through a 30 kHz LPF it will have only a modest slew
rate.



Mark

On 14 Apr 2005 19:40:17 -0700, "Mark" wrote:

the DIM 30 test signal as you described it passes through a 30 kHz low
pass filter. Therefore almost any square wave signal generator will
have a suffienctly fast rise and fall time i.e. slew rate for this test
since it will be limited by the 30 kHz LPF anyway.

It's certainly scary that David's tests found so many mic preamps
that performed so poorly. It's just these (un)bandwidth-limited
locations that are at greatest risk of out-of-band overloads.

Mic preamps, D/A summing junctions and phono equalizers are the
unsung Davids in the world of out-of-band Goliaths. Most modern
electronics lives and works behind a wall of low-pass filtering,
so let's all give thanks and appreciation to the front line
soldiers who don't.

And there's another category of "things that get bumped in the
night", the digital LPF. There's lots to be learned about 'em,
but it won't be from me, sadly. They may even exhibit their own
"TIM". And I don't mean Burton. Chomps cigar.

Chris Hornbeck
6x9=42 April 29

Harry Houdini wrote:
David,

Dynamic distortion - something I've been thinking about recently
too.
Too bad this specification doesn't make the A-list in most
manufacturer literature.

AFAIK, dynamic distortion has no formal definition.

TIM & SID have been studied to death and found to be the same as high
frequency nonlinear distortion.

Probably the best way to measure high frequency nonlinear distortion
is a high frequency twin tone test, such as equal parts of 18 & 20
KHz sine waves. Analyze the output with a spectrum analyzer. Look at
the sidebands that are generated, if any.

Chris Hornbeck wrote:
On 14 Apr 2005 19:40:17 -0700, "Mark" wrote:

the DIM 30 test signal as you described it passes through a 30 kHz low
pass filter. Therefore almost any square wave signal generator will
have a suffienctly fast rise and fall time i.e. slew rate for this test
since it will be limited by the 30 kHz LPF anyway.

It's certainly scary that David's tests found so many mic preamps
that performed so poorly. It's just these (un)bandwidth-limited
locations that are at greatest risk of out-of-band overloads.

I'm not surprised a bit. An excellent rough test is just to run a
1 KHz square wave through a given piece of equipment and eyeball it on
a scope. The square wave calibrator on a Tek scope works well enough
for the job.

You would not _believe_ how nasty things look going through some of the
inexpensive fake tube preamps. Not to mention some of the cheap
microphone transformers out there.
--scott
--
"C'est un Nagra. C'est suisse, et tres, tres precis."