Either way, the capsule FET is still by far the dominant source of distortion in the circuit, probably 10x the distortion of even a ceramic capacitor.
Hey there, can you answer why a constant current source (a FET with gate wired to source) is used instead of a source resistor? I've seen this in AKG CK1x capsules, for example. Most circuits seem to be just a single FET with a source resistor, but maybe they aren't perfect? You piqued (sp?) my interest when you said you're working on a "cascode" circuit. Not quite the same thing as a constant current load, but I read a nice explanation of this circuit.
OPA2134 is 8nV/rtHz. That's the equivalent of a 5K source resistor, and given that your source impedance is much less than that, the opamp is the dominant source of noise in the circuit (and as a semiconductor, it has shot noise that dominates the <100Hz region, see its datasheet on pg. 4--you can see illustrations of both the opamp's shot noise as well as the relation between the thermal noise of the source resistance and the opamp's noise). The good news is that nobody really cares about <100Hz noise in audio, because the source is nearly always much louder, and even if not the ear is not sensitive at all to such noise.
I'm suspicious of opamps, when we run them on a 5-10V swing (5V=usb power only, 10V is what we might get in a battery circuit). Shouldn't these amps run on +/-12V or +/-15V to get their full performance? Note that there are special low voltage opamps (a lot of JRC stuff, for example), but a lot of people are using "standard" NE5532 based designs at lower voltages. In a cheap mixer 5532 are fine running at +/-15V. But I would not trust those in a battery pre or a USB interface (Edirol UA5, etc). I've started putting LM4562 in USB or battery stuff. No measurments, but they did seem to sound more detailed. I keep meaning to actually try one of these battery preamps and measnure distortion (eg., using RightMark or something on a PC). To see if high-end opamps are really needed.
If you like, I'll share a circuit that is balanced from capsule to preamp, that will give you a very nice CMRR figure that will reject what can otherwise be very audible interference (you often don't even need to rely on shielding if balanced), a much more useful concern than worrying about flicker noise of carbon resistors in a circuit that has <10mA current (much less through any of the resistors). I will also add that Linkwitz-modded capsules are slightly more susceptible to interference above 10kHz, even when balanced and shielded. But for drums or loud concerts, that's probably a good trade.
Yes, please share any circuits. A few of us may experiment...
Richard