"Hidden" Distortion

[jcx]

IanHarvy asked (more or less, I’m paraphrasing here) whether amplifier distortion shouldn’t always show up on continuous single tone distortion testing and if I could give an example where distortion "hides" from the single tone test

The circuit below gives ~ -90 dB 2nd harmonic distortion at V(out) when probed with a single tone sine but as can be seen, the 2nd order difference freq distortion products are at ~ -9 dB each for the 10+20KHz with the worst case two tone excitation frequencies

You might object that multipliers aren’t common audio amplifier subcircuits until someone points out the near identity of the diff input stage+tail current source and the Gilbert cell multiplier

I think the point of the example is that low single tone THD doesn’t guarantee well-behaved IMD

[Pjotr]

You might object that multipliers aren’t common audio amplifier subcircuits until someone points out the near identity of the diff input stage+tail current source and the Gilbert cell multiplier

Well, any non-linearity in the transfer function will show up as a “multiplier” someway in the very end.

Cheers

[hermanv]

It's a little long winded but there is a point in here somewhere.

Various kinds of instruments sound different because they have different harmonics as part of their accoustical output. Violins and trumpets making the same note sound quite different they both have fairly large amounts of harmonic distortion, but both sound good in the audiophile sense.

As far as I know, accoustical instruments can not create intermodulation distortion products because there aren't any non-linear elements - only electronic amplifiers can have this illness.

This may speak to why a group heard live, even if it is a group with electronic amplifiers, speakers, etc. usually sounds quite distinctively "live" as compared to a recording of the same event.

I have long suspected that this is because most groups play with each musician having their own amplifier and speaker thereby preventing the intermodulation products. I am guessing that these products are one of the recording anomalies that typically allow us to easily distinguish between live and recorded sounds.

Over the years I have read various reviews with hidden speakers and musicians where the listener couldn't tell the difference between a recording and a live source. All of these I can recall used a single instrument at a time for the demonstration.

If I am right, testing with multiple signals and measuring the intermodulation products is an important way to evaluate electronic designs. Not to be taken to mean that listening tests aren't the most important of all.

[5th element]

This may be not significant and surely is not a techicaly minded thing. But my dad was listening to my speakers because i'd changed them to open baffle and I wanted to hear his thoughts on it.

He said (not in comparison to the other speakers) but when a single intrument played, it was placed very well in the sound stage and probably sounded more reaslistic. However when the whole band came in that same feel was lost slightly, even thought it still all sounded very good and there was an impressive spread of sound.

[sreten]

A thorough grounding in IM distortion would
explain observations expounded so far.

And I don't disagree .

sreten.

[Bill Fitzpatrick]

Originally posted by hermanv


As far as I know, accoustical instruments can not create intermodulation distortion products because there aren't any non-linear elements - only electronic amplifiers can have this illness.

This may speak to why a group heard live, even if it is a group with electronic amplifiers, speakers, etc. usually sounds quite distinctively "live" as compared to a recording of the same event.

I have long suspected that this is because most groups play with each musician having their own amplifier and speaker thereby preventing the intermodulation products. I am guessing that these products are one of the recording anomalies that typically allow us to easily distinguish between live and recorded sounds.

I totally disagree. As far as the factors that distinguish live from recorded, I would think that any distortion products would be so much swamped out by other clues as to become a non-issue.

[hermanv]

Must have been at your end - I don't usually keep my chickens in the house.

[SY]

Violins and trumpets making the same note sound quite different they both have fairly large amounts of harmonic distortion, but both sound good in the audiophile sense.

They have different harmonics but by definition they have no harmonic distortion.

As far as I know, accoustical instruments can not create intermodulation distortion products because there aren't any non-linear elements - only electronic amplifiers can have this illness.

There are LOTS of nonlinear elements in acoustical instruments. But again, there's a big difference between intermodulation and intermodulation distortion.

[hermanv]

Obviously you are correct. Five seconds of discussion with a friend reminded me that there is a reason we make trumpets out of brass instead of lead and violins out of hardwood instead of leather. The overtones created by these materials that flex as sound is made create the character and interest for the instrument. Duh.

That goes to show what single issue thinking and myopia can do. In my own defense, I was focused on the model of feeding accoustic energy into an infintely rigid cavity where I think my hypotesis is correct. No matter how complex the shape, certain harmonics may be emphasized or reduced but no mixing or products will be created.

The other problem may have to do with language. Clipping a sinewave to make it a square wave adds a lot of distortion products. Alternatively, if we feed a square wave to a narrow filter it comes out as a sine wave, is this called distorion?

If certain harmonics are intentionally emphasized in an instrument is that distortion? Reeds and bowed strings produce waveforms with a predictable series of harmonics. Shaping the accoustic path can intentionally enhance some and reduce others, distortion? SET amplifiers have high amounts of even order distortion products, most of us think that sounds good, so we think they sound good. So the origional question, is intentional harmonic enhancement distortion?

My original post was meant along a related line: The debate that intrigues me is "why does electronic reproduction tend to fall quite short of a live event"?

Smarter people than me have worked on this and some things that might seem to be the answer have been mostly ruled out.

Dynamic range? More powerful amplifiers/systems do not automatically sound more life like.

Distortion? There was a period when amplifiers could be bought that had distorion products that were tiny and almost unmeasurable, most sounded perfectly awful.

Frequency response? Again largely ruled out.

I do not know. The debate is confused by accurate vs pleasant. I spend more time on my system reducing "listener fatige" than striving for accuracy but it is often that an improvment in one will cause an improvement in the other. Unfortunately this does not seem hold true every time. Whee.

[SY]

The debate that intrigues me is "why does electronic reproduction tend to fall quite short of a live event"?

Ahhh, now that's a whole different issue, and one that, I think, has very little to do with distortion, assuming the distortion isn't gross.