I think I said somewhere that I'd post the circuit when I settled on my pre-amp design, so here it is if anyone is interested. It's a combined 3-wire battery box (for the AT943 caps) and mic pre-amp.
As you can see, it's very similar to Richard's. R1, R2 and C3 are shared between the channels to save space. Some of the values have been chosen because it's what I have lying around. I've kept it as small as possible, to fit in a small box. I'm sure it could be improved at the cost of space and power consumption (although I've not worried too much about the latter). Unfortunately, C2 needs to be an electrolytic since my iRiver iHP-120 has a very low 2K2 input impedance.
The input impedance is high (about 100K), so it wouldn't be too good with longer cables. The software I used to draw it doesn't do switches (!), but it should be clear what they do. SW1 is switchable bass cut with a DPDT, centre off switch (gives rolloffs at 3, 94 and 233Hz). SW2 is for power and SW3 is a -20dB setting - in one position the gain is 0-20dB and in the other 0-40dB.
All I have to do now is build it 
Hey Graeme, Teddy told me that people were over here talking circuits. I'm glad to see circuit discussions going on.
What's the purpose of C6? Is that just to make sure you don't get high frequency instability by cutting the gain at high frequencies?
Also, I notice that you are using a virtual ground on the positive input side. Why don't you connect your virtual ground on the negative input side by connecting the common terminal of R4 and R14 to the common terminal of R1 and R2, then deleting C1, C13, R3 and R15? If it's from worrying about signal loading on your virtual ground, you could put an emitter follower on your virtual ground and adjust R1 and R2 so that the output of the emitter follower is still halfway between the power supply rails.
Just curious...
OK, switch off now if you're not interested in the tecchie stuff:)
C6 is there to help stability when driving capacitive loads. The op-amp data sheet gives some equations to calculate the value, but it's difficult to use when the gain is variable and you don't know what it's driving into! 22pF is as high as I wanted to go without much effect on the HF response at full gain.
As to the so-called virtual ground...
I don't like to think of it in those terms - trying to emulate a dual-rail supply. The mic output is relative to true ground, so if you make the op-amp work relative to the virtual ground then you are amplifying the difference between the two. This gives poor supply rejection (PSRR). The purpose of R1/R2 is simply to bias the op-amp mid-rail with as little supply noise at that point as possible. In other words, think in terms of a single rail supply rather than trying to fake a dual one.
Think of it this way - with your connection, any supply noise on the R1/R2 junction is seen at the op-amp negative input in full. The positive input, however, sees virtually none of it as the noise is reduced via R3 and the output impedance of the mic/FET. Therefore, the op-amp will amplify this difference by the full gain of the circuit (up to 40dB). I simulated the 2 circuits (assumed the op-amp has perfect PSRR for simplicity). I set R12 to be 5K to emulate the output impedance of the mic/FET - if the figure was lower, the difference between the circuits would be greater. This is done at max gain (i.e. worst case).
The following are the supply rejection figures (in dB) at 3 frequencies. In each case, the first figure is my circuit and the second is your suggested change.
20Hz: -65 -2
1kHz: -143 -36
20kHz: -195 -62
As you can see, the circuit as it stands will beat the specs of the op-amp at anything above about 100Hz - i.e. the op-amp is the limiting factor for PSRR. With the suggested changes, this is nowhere near true even at 20kHz.
Another thing is that your suggested change will amplify any input offset voltage (either from the op-amp or due to bias currents) by the full circuit gain. Not too much of an issue with the LT1469, but worth bearing in mind.
Finally, the suggested change would make the crosstalk between channels worse, unless R1/R2/C3 were duplicated for each channel.
HTH
Topic: DIY: three-wire batt. box + preamp (Read 10841 times)