opamp inverting input sounds better?

[jan.didden]

Originally posted by JohnG
If there is benefit to be gained from throwing away some of the open loop gain of the op-amp, can you throw it away at the output, by a resistive divider, for example? How does this affect the noise of the circuit, compared to throwing away gain at the input?

John

Suppose you want to throw away open loop gain be having the opamp amplify more then required. You would get too much output signal than required.
If you attenuate the input signal, the disadvantage is that the signal-to-noise ratio (SNR) deteriorates: the amount of opamp noise in the output signal increases with respect to the signal. The advantage is improved overload margin because you attenuate the input signal, which may or may not be of some value.
If you attenuate the output signal, you also attenuate the opamp noise, so the SNR is preserved. The disadvantage mentioned by Nelson isn't really there, because if you would *not* throw away gain you would have the same output signal as when increasing gain and then attenuating at the output.
As discussed above, throwing open loop gain away by manipulating the open loop gain directly also worsens the SNR.
So, IMO, if you *have* to do it, I would favor output signal attenuation.
But it is smarter not to get in this situation by chosing a better (for the purpose) opamp.

Jan Didden

[sonnya]

Something about opamp's and capacitive load, noise gain and other things....

http://www.analog.com/library/analog...ing/index.html

The one above was wrong but still helpfull :

Try this one :

http://www.analog.com/library/analog...2/appleng.html

[jan.didden]

Yes, Sonnya,

The section on noise gain is in fact almost the exact wording of my earlier post.

Jan Didden

[Christer]

Thanks Sonnya, that seems to be a very good reference. I
suppose it also answers my question earlier in this thread,
so unless Jan and Jonathan meant something else, not covered
here, there is no need to answer it now.

[Nelson Pass]

We are not talking about throwing away input signal here,
but open loop gain.

If you attenuate the output of a typical op amp by
20 dB, it will only swing about 1 v rms. This is probably
not going to satisfy most users.

There are very few applications where the noise figure
of a state-of-the-art op amp can't afford to be degraded
a bit in exchange for better sound.

Realistically, this sort of thing is only an issue when the
gain of the op amp is set at less than 10 (20 db), usually
in line level stages and followers and the like. Under these
low gain conditions, the noise of the input stage of the
op amp is an infinitesimal. Imagine a 1 nanovolt/square-root-
Hertz degraded to 10 nanovolts. You still won't hear it, but
there's a good chance that the stability difference will be
quite audible.

[jcarr]

The second link that Sonnya posted is quite good, and I feel that it makes many excellent points and maps out the general direction admirably, but still doesn't explain the entire story.

In the end, you will personally have to build your circuit of interest, and listen to it with varying amounts of global feedback. There will be a point that you can go up to without problem, above that you will encounter an area where the precise value is a matter of personal preference, and then you will likely hit a region where most listeners agree that the sound has taken a turn for the worse. And this is all with adequate phase margins (with capacitive loading applied).

My findings are that the optimal amount of global NFB depends on the physical structure of the circuit as well as the topology and component choice. And that is why I suggest that you do the experimentation for yourself, rather than asking someone else.

I will add that _how_ you deal with the excess gain will also affect the sonic outcome.

regards, jonathan carr

[GRollins]

I seem to be shooting myself in the foot recently, all in the name of brevity and/or haste.
Something I was going to say in my last post, but left out, was that you can remove noise at pretty much any point in the amplification chain that you choose...at a price.
Either analog or digital methods are available. Filtering, dynamic gating, statistical analysis for patterns in the signal, you name it. The price is that you're getting into some serious complexity; the noise-removal part of the circuit can easily have ten times the parts count of the circuit itself. Commercial examples were all the rage for a while: the Phase Linear Auto-Correlator (dynamic gating) and the Burwen TNE-700/A (triggered off signal rise times) come to mind. I seem to recall seeing the schematic for the Burwen unit. Ugh. <i>Not</i> minimalist.
All strategies I'm aware of work best if you're willing to accept decreased bandwidth, or are able to work with an uncomplicated/predicable signal--characteristics that don't generally apply to music...
waitaminnit! Sounds like just the ticket for the boom-boom car stereo guys...

Grey

[rljones]

Dorkus originally asked how does the sound of inverted versus non-inverted op amps compare. The thread has covered various ground, but I've not read any comments regarding listening tests per se. When Thorsten was asked this question on the Amp Chips DIY forum, he said he no longer listens to solid state and that his comparisons of inverted vs non-inverted came from basically a decade ago. Certainly opamps have improved in that time period.

Therefore, to re-address Dorkus' question: has anyone, for example, used an AD8610 in a circuit and actually listened and compared inverted vs non-inverted topologies?

One thing intriguing about this particular chip is that the spec sheets indicate its slew rate is the same at unity gain both in inverted and non-inverted mode (unlike the AD627, where the AD627 has a markedly slower slew rate in the non-inverted mode).

Robert

[planet10]

Originally posted by jcarr
My findings are that the optimal amount of global NFB depends on the physical structure of the circuit as well as the topology and component choice. And that is why I suggest that you do the experimentation for yourself, rather than asking someone else.

A very apt statement. Everything matters. The idea behind this statement applies to almost anything hifi.

I had a good laugh today reading a letter to the editor of Audio Critic magazine. I paraphrase: "could you do more speaker reviews, the only components that actually sound different..."

dave

[Variac]

Actually, he might have not been too far from the mags philosophy.

Here's a quote from a review of a QSC Audio pro audio amp in the same issue #28 :

"Regular readers of The Audio Critic know that we don't go into the specifics about the sound of a well-designed amplifier, since it is the same as that of any other well designed amplifier...."
-Peter Aczel

At the end, the techical reviewer in the sidebar is forced to say:

"Good for sound reinforcement applications, but ultimately other choices are better for consumer aplications" He can't say WHY , because there are no differences....

Of course the out is saying "well-designed" but in fact the only knock Mr. Aczel has on the amp is the noisy fan..

I picked it up on the news stand. Probably will be awhile before it do it again.....