Feedback in audio esp wrt op-amps.

[Arny Krueger]

Why ?

Electronic devices seem to tend to have exponential characteristics.

 

[William Sommerwerck]

It seems to make a lot more sense to apply NFB

in rather more moderate amounts to
a gain stage that's already quite linear.

I first read this rule in the '60s.

 

[William Sommerwerck]

When and where was this ?

Neve Connecticut, USA. 1983, I think.

 

[Arny Krueger]

And the spectrum is important.

The spectrum is important, in fact it is critical. In many cases, 2nd order
distortion is hard to hear, particularly when listening to solo instruments.
This is due to concurrent spectral masking in the human ear.

Higher order harmonics are not so well masked as 2nd, so the ear can pick
them out, even when they are relatively small.

Complex music may be more critical of lower order distortion, because
audible IM products are more likely to be generated by complex sounds.

One of the worst cases for IM is music with lots of energy at high
frequencies, and less energy around 4 KHz. Any IM products that are
generated by the energetic highs fall in the midrange where the ear is most
sensitive, and there is very little masking due to the lack of midrange.
Thus, recordings of keychains jangling are better for judging nonlinear
distortion than say a full big band.

When I was at HE2005, I found that many SET demos were based on simple
music - solos and the like. I did hear some SETs being demoed with more
complex musical sounds, and frankly it could get ugly.

It is pretty atypical for feedback to fail to reduce even high harmonics,
perhaps with less improvement than lower harmonics. As long as you keep
feedback factors >= 10 (20 dB) across the audio band, inverse feedback is
very much your friend.

Of course there are exceptions, such as mic preamps based on cheap op-amps
that don't have enough bandwidth. A mic preamp with a 40 dB gain stage
needs to be based on an op amp with a GBW of at least 20 MHz, for example.
Thus, there is 60 dB open loop gain at 20 KHz, and 20 dB feedback factor
with a closed loop stage gain of 40 dB.

Even the hot new op amps like the LM4562 have only 55 MHz GBW, so 60 dB gain
in one stage remains elusive. The old Neve favorite 5532/4 had only 10 MHz
GBW. But, Neve wasn't shy about cascading them to reduce stage gain
requirements.

I've seen cheap mic preamps built with op amps whose GBW was well below 10
MHz. This is no doubt one reason why cheap mic preamps tend to slack off on
maximum available gain.

 

[MooseFET]

Yes, as abused in the 70s, vast quantities of NFB were used in attempts to
correct significant non-linearities.

It seems to make a lot more sense to apply NFB in rather more moderate amounts to
a gain stage that's already quite linear.

No, that is the wrong way to go. You want to apply buckets of NFB to
a stage that is already perfectly linear.

You want to push the gain crossover up to high frequencies and have
enough NFB to ensure that none of the harmonics made by any
nonlinearities can have significant amplitudes. "Moderate" may not
mean enough for this. One mans "moderate" is another mans "extreme"
so perhaps you should have used a number.

 

[Scott Dorsey]

Yes, but do you understand what I said, or are you just choosing to ignore
the difference between parallel circuits and series circuits?

I understand what you said, but it isn't really relevant except in the
case of the equalizer mentioned by another person in this thread, or if
you want to handwave about the number of individual op-amps in all the
channel strips on that SSL.

Pop the cover on an SSL 4000 and count the number of individual gain stages
from the front to the back, then into the tape mchine and back again. There
are lots.
--scott

 

[William Sommerwerck]

The spectrum is important, in fact it is critical. In many cases, 2nd
order

distortion is hard to hear, particularly when listening to solo instruments.
This is due to concurrent spectral masking in the human ear.

Higher order harmonics are not so well masked as 2nd, so the ear can
pick them out, even when they are relatively small.

The audibility of harmonic distortion might also depend on tuning. In a
system using whole-number ratios, harmonic distortion of all orders might be
less audible than in a tempered system.

 

[Arny Krueger]

The audibility of harmonic distortion might also depend on tuning. In a
system using whole-number ratios, harmonic distortion of all orders might
be
less audible than in a tempered system.

Why might that be?

 

[Mark]

That's not what I asked.

I aked if overall NFB can create 'new' harmonics, and it's now clear from
responses here that it can. I'm curious about the effect of local NFB in this
respect too. Does linearising a single gain stage with e.g. emitter degeneration
do the same ?

Yes, the sim I posted is local feedback around a single stage.

This is an interesting math excersize but of little practical value.
It tells us that in the case of an ideal second order device that
otherwise would only create 2nd order distrotion and NO 3rd 4th etc.
adding neg feedback will reduce the 2nd as desired but will also
create a small amount of third and forth etc that otherwise would not
be there.

Any practical non ideal real amplifier stage will create its own 3rd
4th etc and the neg feedback will reduce them as well.

So neg feedback (done correctly) is always a good thing.

 

[Kevin Aylward]

Read the later posts.. I have conceeded the point.

Well, there are too many of them...

It is an interesting observation but not very relevant to real world
audio amplifier circuits that are not ideal square law devices.

Ho hummm...it illustrates the principle mate, so is very relevant, I just
cant be bothered to include all the terms. If you want to see some more
terms of distorion thingies, try
http://www.kevinaylward.co.uk/ee/bipolardesign2/bipolardesign2.html

or even http://www.kevinaylward.co.uk/ee/tapebias/tapebias.html might be
interesting as well.

 

[William Sommerwerck]

The audibility of harmonic distortion might also depend on tuning.

Why might that be?

Masking effects -- or lack thereof. For example, in such a system, the third
harmonic would fall exactly on a scale note, whereas under our current
system, the third harmonic comes _very_ close, but not exactly.

This is speculation, of course. It's unlikely anyone has researched this.

 

[Arny Krueger]

Masking effects -- or lack thereof. For example, in such a system, the
third
harmonic would fall exactly on a scale note, whereas under our current
system, the third harmonic comes _very_ close, but not exactly.

Masking is not *that* frequency-sensitive.

This is speculation, of course. It's unlikely anyone has researched this.

A lot of masking curves have been developed, based on measurements. They are
relatively broad affairs. So there has been a lot of research into masking,
and how it works is fairly well-known.

Besides, lots of musical instruments generate overtones that are not exact
integer multiples of the fundamental.

 

[Eeyore]

I first read this rule in the '60s.

Where did you originally come across it ?

Graham

 

[Eeyore]

Neve Connecticut, USA. 1983, I think.

Bethel presumably ? Did you ever meet Robin Porter. I was told he'd spent some
time at RN Inc.

Graham

 

[Eeyore]

The audibility of harmonic distortion might also depend on tuning. In a
system using whole-number ratios, harmonic distortion of all orders might be
less audible than in a tempered system.

But who's still using those old tunings ?

I've heard the difference it makes to how the music sounds btw (at college 30
yrs ago) and it's quite remarkable.

Graham

 

[William Sommerwerck]

William Sommerwerck wrote:

Bethel presumably? Did you ever meet Robin Porter.
I was told he'd spent some time at RN Inc.

Bethel, yes, right across from Mallory/Duracell.

The name doesn't ring a bell. I've forgotten the names of the people I
worked with. Hope none of them is reading this!

 

[William Sommerwerck]

William Sommerwerck wrote:

Where did you originally come across it?

In an "Electronics World" article about the new JBL solid-state amp with a
"T" output stage. The author explicitly stated the principle -- get the
circuit as linear and wideband as possible before applying feedback.

 

[Eeyore]

Bethel, yes, right across from Mallory/Duracell.

The name doesn't ring a bell.

Just thought I'd ask. He was UK based but I'd just heard he spent some time over
there. We worked in the same dept (operations/custom engineering) and we had
even been on the same course at uni but not in the same year.

What was your role there btw ?

Graham

 

[Eeyore]

In an "Electronics World" article about the new JBL solid-state amp with a
"T" output stage. The author explicitly stated the principle -- get the
circuit as linear and wideband as possible before applying feedback.

Right. I think I've come across the article online. I also read Wireless World
back then too. I may have saved it somewhere. I do recall smiling slightly when
I read it recently on account of a certain element of 'old thinking' it
embodied. Makes you realise how far we've advanced over the years.

Graham

 

[isw]

The spectrum is important, in fact it is critical. In many cases, 2nd order
distortion is hard to hear, particularly when listening to solo instruments.
This is due to concurrent spectral masking in the human ear.

Higher order harmonics are not so well masked as 2nd, so the ear can
pick them out, even when they are relatively small.

The audibility of harmonic distortion might also depend on tuning. In a
system using whole-number ratios, harmonic distortion of all orders might be
less audible than in a tempered system.[/QUOTE]

Most instruments produce overtones as well as -- or instead of --
harmonics. They don't fall directly on multiples of the fundamental no
matter what tuning is used.

Isaac