OpAmp FB resistor ratios and noise

[seedlings]

Sorry for the vague title...

In a TS-type opamp distortion, the ratio of the resistance in the feedback loop to the resistance to ground determines the gain (within the bandwidth limits of the opamp and cap-to-ground selection).  This means a 100k fb resistor and 1k to ground would be the same gain as a 1k fb resistor and a 10R to ground, and the same as a 1M fb paired with a 100k to ground.

In a given opamp, will the 1k and 10R combo always have the least noise (white noise)?

Thx,
CHAD

[Mark Hammer]

page 8.  read, starting from there.  http://hammer.ampage.org/files/Device1-8.PDF

[seedlings]

^ Good article

Either 1) that article didn't answer my question, or 2) it answered my question by not saying anything at all about the resistors in the feedback circuit 

Basically, the resistor input-to-ground was the only one tested for noise in this  article.  In the case of that resistor, lower was better.  Perhaps it's safe, then, to assume lower resistance is less noisy wherever practical.

CHAD

[Mark Hammer]

Some chips are more comfortable with lower resistances at their inputs and in their feedback loop.  That's why you'll regularly see high-value resistors around BiFet op-amps, and fairly low ones around devices like the NE5532 and LM833.

[Bill Mountain]

^ Good article

Either 1) that article didn't answer my question, or 2) it answered my question by not saying anything at all about the resistors in the feedback circuit  

Basically, the resistor input-to-ground was the only one tested for noise in this  article.  In the case of that resistor, lower was better.  Perhaps it's safe, then, to assume lower resistance is less noisy wherever practical.

CHAD

I guess my only concern would be the effect of tolerance is less in higher values but I guess 5% is 5% no matter what scale.

[seedlings]

Some chips are more comfortable with lower resistances at their inputs and in their feedback loop.  That's why you'll regularly see high-value resistors around BiFet op-amps, and fairly low ones around devices like the NE5532 and LM833.

Interesting because I tend to favor the LM833 and lower resistors like 10s of ohms and a 50k fb pot.

CHAD

[PRR]

> same gain as a 1k fb resistor and a 10R to ground

Some chips won't drive the 1.010K load well. And you could go futher, 100r and 1r, or 1r and 0.01r, and really overload the chip.

Often there is a cap in series with the to-ground resistor. For hi-fi 20Hz bass, this is 10uFd for 1K, 1,000uFd for 10r, 10,000uFd(!) for 1r. Cap-costs (and size) matter. True, guitar needs less bass, maybe lots less in a distortion circuit.

There is a diminishing-returns issue. A TL072 has around 2uV of audio hiss. 20K resistance has about 2uV of audio hiss. So 20K is rather high (3dB noise-figure). But 2K is 0.6uV audio hiss, which adds to TL072's 2uV hiss in RMS style to make just 2.1uV hiss, which is indistinguishable from a TL072 with a noiseless NFB network. So for TL072 there is little reason to go below a few-K for the to-ground leg. '5532/833 have lower hiss and ~~500r may be a wiser choice.

[R.G.]

"Audio Ic Op-Amp Applications" Third Edition, Walt Jung, section 2.5 Noise

The answer is... (drum roll please)

It varies.

It depends on the source resistance, the opamp's internal noise generators and the equivalent resistance on the inverting input, which is Rin paralleled with Rf.

Says Walt:
"Thus, a real amplifier has five potential sources of noise that have to be considered for minimization. First there is the thermal noise of (*possibly) two source resistors seen by the inputs, which is an irreducible minimum, existing even with an ideal noiseless) amplifier.   ...
there are the noise-current and noise-voltage generators.  [he means inside the amplifier itself here] For low values of source resistance, the effect of [the internal current noise generators] is often at a minimum. Under such conditions, which are usually typical for audio applications, [... but not necessarily guitar pedal applications...] [the voltage noise generators] will dominate as the source of amplifier noise. As the source resistance is increased, the effect of the [current noise generators] becomes larger until, at high source resistances, [internal noise current times the + input bias resistor] and [internal noise current times the parallel combination of Ri and Rf] are the dominant components of amplifier input noise. Thus, in specifications, these two parameters are detailed separately, with [input voltage noise] specified at a low source resistance and [input current noise] specified at a high source resistance. Both [current and voltage noise sources] are given in terms of spectral density..."

on and on and on. Henry Ott's "Noise Reduction Techniques in Electronic Systems" is more informative, less brief and more dense.

Bottom line: use opamps with low internal voltage noise for low source resistances (generally, under 10K) and opamps with low internal current noise for high source impedances. This leaves you kind of hung where your pedal may plug into a guitar with an inductive impedance peaking at about 100k-200K at 7kHz, or into the hundred ohms of another pedal output. And make the parallel combination of Ri and Rf as low as you can. And the bias resistor on the + input as low as you can.

Where "as you can" means to make them still do their jobs in the larger circuit. The plots of output noise per frequency given tend to be about 6-10db higher than source resistance above about 1k source resistance, leveling out below that. This just shows that the resistor thermal noise dominates at high source resistances. Flicker noise tends to dominate below 100-200Hz, depending on the amplifier.

So make them as small as you can; if you have to plug a raw guitar into it, this may not be all that small.

There is a lot more to this. I picked out the best short synopsis I could on short notice.

[seedlings]

I just fixed a Crate V15 and had it pretty cranked to make sure the feedback issue was fixed.  So, I can't be for sure if the explanation above is ringing my head or the amp check.

I will take small doses at regular intervals, and in the meantime I'll carry on.  I try to use the lowest resistors possible, but of course, need to keep input impedance high.

Thanks!

CHAD