Feedback in audio esp wrt op-amps.

[William Sommerwerck]

Pop the cover on an SSL 4000 some time...

When I worked at Rupert Neve, I learned that SSL stood for "sure sounds
lousy".

 

[Paul Stamler]

Yes, the article was written by Peter Baxandall. I haven't been able to find
the article itself online, but Stereophile published a short summary here:

http://stereophile.com/news/10065/

I've read Baxandall's article, and it's written with his usual thoroughness.
He explicitly stated that the absolute level of higher harmonics rose when
he introduced feedback, and noted that with 40dB of feedback the level of
5th harmonic was higher, on an absolute basis, than it was with no feedback.
At higher levels of feedback (on the order of 60dB) the levels of high
harmonics began to decrease again.

Thank you for pointers that got me to the right place, or at least its
vicinity!

Peace,
Paul

 

[William Sommerwerck]

If you start with a hypothetical perfect 2nd order device, I MIGHT be

ready to concede that neg feedback might produce some small level of
4th order that wasn't there before.

That isn't what we're talking about.

This may be an interesting mental exersize, but it has very little
connection to actual practice. In practice using any REAL amplifer,
neg feedback REDUCES all the harmonics.

True. The claim (which I made) was that the spectrum of the harmonics
changes.

 

[William Sommerwerck]

Small amounts of feedback created the most distortion.

As I increased the closed loop gain, as expected all the
distortion levels were reduced.

Okay, but what about the spectral distribution?

 

[Bob Myers]

How do you know that?

The question was whether or not negative feedback
NECESSARILY resulted in more distortion than operating
a comparable set-up in an "open loop" fashion. While
we have seen various mathematical treatments that show
how additional harmonics/distortions may be generated
via non-linearities, none have shown that the total distortion
is necesseraly greater in the negative-feedback case. On
the contrary, the math regarding the feedback case shows
how such distortion will actually be reduced in total.

Bob M.

 

[Kevin Aylward]

And I am saying NFB CANNOT create new higher harmonics.

Ho humm.... no.

Look, it is this, a 1st order approximation is:

Vo = aVipsin(wt) + b(VipSin(wt) )^2 ++...

This means Vo will have some 2nd after you expand the sin squared term. Now,
that 2nd harmonic of Vo added to the input as in when feedback is applied,
means that the net Vin to the amp is say, (Vp qSin(wt) + kSin(2wt))^2. Now
expand this and you will get a Sin(3wt) term.

 

[Mark]

Ho humm.... no.

Look, it is this, a 1st order approximation is:

Vo = aVipsin(wt) + b(VipSin(wt) )^2 ++...

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

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

Mark

 

[Eeyore]

OK I ran the sim...yes you are correct adding neg feedback to a
perfect 2nd order device creates higher order harmonics 3rd 4th etc
that were not there before.

Some PSPICE code for those that want to play....

Neg Feedback Amp does neg feedback create high order distortion

.TRAN 1uS 10ms

*transient analysis sine wave
Vin 1 0 Sin(0 1 1KHz)

Eamp 2 0 poly(1) (1,2) 0 100 -10 ;with 100% neg feedback
*Eamp 2 0 poly(1) (1,0) 0 100 -10 ;with NO neg feedback

Rloadin 1 0 600
Rloadout 2 0 600

.probe

.end

Small amounts of feedback created the most distortion. As I increased
the closed loop gain, as expceted all the distortion levels were
reduced.

In most any real amplifier, there will be high order non-linearities
in the device

Why ?

Graham

 

[Eeyore]

It absolutely is. However, in the 1970s it was regarded as a cure-all that
could fix all ills, and it's not. The resulting sonic issues were severe,
and the current backlash you see in the community against the use of feedback
is mostly a reaction to that. This is a shame, since feedback is a useful
tool.

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.

Graham

 

[Eeyore]

When I worked at Rupert Neve

When and where was this ?

I was at Neve Melbourn myself for 3 years (1985-1988).

Graham

 

[Eeyore]

That isn't what we're talking about.


True. The claim (which I made) was that the spectrum of the harmonics
changes.

And the spectrum is important.

Graham

 

[Eeyore]

The question was whether or not negative feedback
NECESSARILY resulted in more distortion than operating
a comparable set-up in an "open loop" fashion.

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 ?

While we have seen various mathematical treatments that show
how additional harmonics/distortions may be generated
via non-linearities, none have shown that the total distortion
is necesseraly greater in the negative-feedback case.

No, I wouldn't expect it to be *numerically* greater. That would make no sense.
Hiowever the ear responds not only to the quantity of distortion but it's
spectrum, with higher order harmonics sounding more unpleasant.

It's therefore entirely possible to have an amplifier with a numerically smaller
THD figure that actually sounds worse than an amplifier with higher THD.

On the contrary, the math regarding the feedback case shows
how such distortion will actually be reduced in total.

The headline x % THD is not actually especially helpful when being very
critical.

Graham

 

[Eeyore]

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

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

That's a simple proof of the effect. It's not as if gain stages with other
non-linearities won't be similarly affected.

Graham

 

[Karl Uppiano]

There was part of a thread a while back about how adding negative feedback
can
create higher order harmonic distortion products than exist open-loop in
an
amplifier stage.

This made me think about the application of op-amps in audio generally.
Negative
feedback is used primarily to linearise the transfer function and is used
in
huge quantites as much as 80dB @ 1 kHz for example.

Since this amount of NFB is not required to provide an accurate gain
setting, it
struck me that it's somewhat counter productive. If instead the open-loop
transfer characteritic was made more linear by degeneration of the
open-loop
gain for example, when NFB is applied, the overall result should be
largely
similar (i.e. no worse) but would presumably also suffer less from the
creation
of these new distortion products .

There is a good analysis of a "blameless" amplifier here:
http://www.dself.dsl.pipex.com/ampins/dipa/dipa.htm

Note that DC amplifiers of the type described here are remarkably similar to
the topology used in a typical op-amp.

This amplifier is optimized for a specific amount of feedback. Op-amps are
typically general-purpose devices that, while usually well-designed, trade
optimal performance for convenience (you program their operational transfer
function with feedback - hence the name "op-amp"). When used not too close
to their design limits, I think high quality op-amps can be quite acceptable
even for high-fidelity applications, but probably not for the "bleeding
edge" audiophile.

On the other hand, some of the esoteric circuits that attract audiophiles
are measurably inferior to an equivalent circuit that uses op-amps, so in
many cases there is more to it than mere audio fidelity.

 

[Scott Dorsey]

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

Some of which (like crossover distortion) are not really solvable with
feedback.

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.

Yes, but then you need a gain stage that has plenty of extra gain, good
margins, and good linearity. That's quite a bit to ask for.
--scott

 

[Paul Stamler]

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.

It does, intuitively. But Baxandall's results suggest otherwise; if one
wishes to avoid high-order distortion components, one should either use a
lot of feedback or none at all.

I'd be very interested to see Baxandall's experiment (which I believe was
done using a JFET) repeated on other active devices such as MOSFETS, vacuum
tubes and of course bipolar transistors.

Peace,
Paul

 

[Eeyore]

Some of which (like crossover distortion) are not really solvable with
feedback.


Yes, but then you need a gain stage that has plenty of extra gain, good
margins, and good linearity. That's quite a bit to ask for.

It's not actually *that* difficult as I keep saying.

Going back to the 60s/70s when transistors were actually quite expensive (I recall the
BC109 cost 6s/6d from hobby stores - that's 32p in decimal plus adding the inflation
makes it well over £1 or $2 in today's money) designers used all manner of tricks to
screw the last dB of gain from them, potentially at the cost of linearity.

With a typical general purpose low noise transistor costing around 2 cents in quantity
these days, there is absolutely no need to have to do that any more and linearity
should be the goal.

Graham

 

[isw]

["Followup-To:" header set to sci.electronics.design.]

The idea that you can 'get away' with sloppy circuitry for replay because
the
source was in some way 'impaired' is totally false.

I don't think anybody proposed "sloppy" circuitry for replay. The point is
that studio audio gear is just solid, reliable, conventional good audio
stuff (none of that high-end low-oxygen power cord crap). Plenty of opamps,
plenty of NFB, plenty of digital processing, plenty of all the things that
high-enders loathe.

Since the recording studio already did 90% of the work of completely
destroying the audio signal beyond repair, it doesn't matter how much your
home audio gear adds to that.

Sometimes when I hear the golden earers talk I'm surprised that I can make
out any music at all when listening with my Cantons fed from an old Sony amp
through particularly oxygen-rich cables.

If they really were, they'd break when you tried to bend them 8^}

Isaac

 

[Eeyore]

If they really were, they'd break when you tried to bend them 8^}

Indeed. 'Ordinary' copper is pretty oxygen free too. It's just a load of nonsense
about how many 9s you want after the decimal point.

Graham

 

[Mr.T]

Pop the cover on an SSL 4000 some time...

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

MrT.