Capacitive feedback, anyone tried this?

[scott wurcer]

I wired up one of these and it seems to work as expected. It uses capacitive feedback to set the gain in a microphone interface. The FET on top is an AC only current source. The beauty of this circuit is that the input is a virtual ground, and several sources of distortion are eliminated. I also expect that back to back diodes could be used instead of the 1G Ohm resistor for bias. I will confirm this when i get some BAV199's (the diodes need to be ultra low leakage). The resistance at the origin for a diode is related to Vt/Is.

[moamps]

Hi Scott!

First off, thank you for sharing. It's a very neat design and I'd appreciate it if you could answer a few question for me:

1. Have you measured the pulse response of the mike-amp?
2. The quality of the Cf cap might be critical in this design. Do you have any preferences regarding the cap type?
3. In my experience, FETs may be effectively used as low-leakage diodes. What are you thoughts on this? Have you tried that?

Thanks,
Milan

[unclejed613]

i have seen 2n3904 and 2n2222 transistors used as low leakage diodes. NAD used to use them in the memory capacitor circuits of their digital receivers and fm tuners. just cut off the emitter lead, and use the b-c junction as your diode.

[mikeks]

Originally posted by scott wurcer
The beauty of this circuit is that the input is a virtual ground, and several sources of distortion are eliminated.

Excellent design!

[scott wurcer]

Yes, I removed the comment that a spare FET makes a good low leakage diode to get the picture to fit into the 1000x1000 pixel limit. I made some measurements on an "air" wired BB that showed beauitful square wave response but was oscillating at 400MHz. The PC board and the 100 Ohm snubber cured that. I hope to characterize this more soon. I did look at using standard BAV99 clamp diodes instead of the 1G resistor and could see the increase in low frequency noise so I assume the theory is right about equivalent diode resistance at 0 V. I need to try known low Is diodes and confirm this.

The capacitor quality must matter but for low values there are lots of options, not the least of which is a sealed vacuum cap.

I still wonder about other applications of this since your feedback impedances have no real part they contribute no noise.

One caveat is that the value of the capacitor that AC couples into the top FET current source is somewhat critical for low frequency flatness. You can get a nice flat response down to 2 or 3 Hz with the right capsule (at least in simulation).

[scott wurcer]

Originally posted by mikeks


Excellent design!

Thanks, I started this line of thought when National and others introduced mike interface IC's with MOSFETs and back to back diodes on the input. Open loop the diodes severely limit dynamic range. Mike designers are still skeptical about dumping those huge bias resistors. But without them you can get everything at Digikey and cheaply to boot.

[SY]

I still wonder about other applications of this since your feedback impedances have no real part they contribute no noise.

Are you thinking capacitive phono cartridge?

[scott wurcer]

Originally posted by SY


Are you thinking capacitive phono cartridge?

Not exactly, those MEMS capacitances are just SO small that other things get difficult. We are are working on some exposed die packages so I suppose you could glue a stylus right on a chip, but the electrostatic restoring force for a couple of grams tracking would probably be thousands of volts. There were those electret cartriges, do you know if their internal losses were low enough to benefit? The microphones are actually limited by the internal Q and air damping in the end. You really can't beat thermodynamics and physics.

[EC8010]

Tim de Paravicini mentioned this technique being used in Neumann microphones at ETF06. And that was with valves.

[SY]

I wasn't thinking so much of MEMS as I was about something much more conventional, more like the old Weathers cartridges. The Weathers used a conventional suspension and cantilever, but instead of coils and magnets, it had a plate at the end of the cantilever which wiggled around a plate fixed in the cartridge body. The variable capacitance frequency modulated an RF oscillator.