Vbias components in Maestro MFZ-1 circuit.

[brett]

Hi
I've got a Maestro MFZ-1 clone that is quite noisey.  Probably hard to totally avoid in a circuit with two 1M feedback resistors for the op-amps, and the high gain (about 500).

But I wonder if Vbias network is also making noise?
In standard form, it consists of 2 x 2.2M resistors and a 0.1uF cap. 
The arrangement is the usual voltage divider, with the cap connecting the Vbias junction to ground.

My questions are:
Won't the 2.2M resistors make a lot of thermal noise? (Most Vbias resistors seem to be 10k, maybe 100k)
Is the 0.1uF cap big enough to effectively conduct all of the noise to ground? (Most caps seem to be 10uF or 22uF)

Is there some reason why such an arrangement might be useful?  Or did Maestro just have some surplus 2.2M resistors?

Maybe I'm barking up the wrong tree and I should focus on those feedback resistors.  (I'm currently using a TL072, but might swap to an NE5532 to cut some noise)
thanks

[R.G.]

The cap to ground from the center of the two bias resistors shunts noise to ground above frequencies greater than a few Hz. It's impedance is under 1M at 1.59Hz.

However, high resistance networks are not particularly low noise. That would better be done as two 10K's from +9-gnd bypassed by a 10uF, and a 1M from the bias voltage to the (+) input. Keeping 1M impedance on the (+) input minimizes the DC offsets for the first opamp, which is part of what they were trying to do.

Frankly, it's a poor circuit from a noise standpoint. The input is a high-input-impedance inverting buffer done with high value resistors and followed by a very high AC gain. That's practically a prescription for hiss.

The circuit would be much better if it
(a) used the first opamp as a non-inverting buffer with a 1M input bias resistor to the 10K-10K stack I mentioned
(b) Made the first stage a non-inverting gain of perhaps 10 to get some gain for the hiss it's inevitably going to add
(c) used a 10K drive pot, not a 100K one
(d) used metal film for all the resistors on the first stage and the 1M in the second stage
(e) used a noninverting buffer first stage so as to not invert the signal overall; I wonder what they were thinking to go to the trouble to do a 1M input impedance input buffer the simple - but low performance - way.

The NE5532 may not improve noise in this circuit without changes because of its low input impedance.

I don't know what if anything they had in mind. Not noise performance, certainly.