AES Convention Papers Forum

This paper is very cool and hardly anyone has been looking at RF condenser designs recently so it is good to see this sort of work being done.

But... The variable RC network is inherently a phase modulator.  More recent RF condenser microphones have used AM detectors with slope detection of the resulting signal only because of better linearity compared with an FM detector.  So why carry all that analogue detector baggage into the digital world?

If you go back to FM, you eliminate almost all the sources of nonlinearity that you had to deal with before.  The exciting signal can be a square wave.  There is no need for a low-distortion sine waveform anymore, and making low-jitter digital clocks is something we know how to do well.  Likewise, you can eliminate the A/D completely and replace it with a simple FPGA that counts time from one leading edge to another.  All of your amplitude linearity issues just turn into soluble frequency linearity ones.

So it seems to me that although this design is ingenious, there is likely a simpler way to solve things.

Thanks for your comment! 

You're right that there are, of course, simpler methods to build a digital RF microphone like that one you've suggested. There have also been some publications on digital microphones that use an FM approach together with a simple D-Flipflop. In these papers, the SNR was just far from being sufficient for a high quality microphone and this seems to be one major disadvantage. This is, because you'll need an extremely low-jitter oscillator that also generates a high enough frequency deviation on extremely low changes in capacitance to meet SNR and dynamic range requirements. This might be hard to achieve.

With the "AM" approach (yes, inherently it's also phase modulation) one can suppress the oscillator noise using the bridge circuit, which is a big advantage, but comes with a lot more complexity. In another paper from 2020, we have been able to achieve an SNR of ~87 dB(A) using the symmetric condenser capsule from an MKH20. Note that the MKH20 itself has an SNR of ~84 dB(A) (10 dB(A) inherent noise as per datasheet). Additionally, with this approach, the resulting low frequency 1/f noise is reduced (it's the electrical 1/f noise that is bypassed using the digital demodulator).

To conclude, your "simple" way will of course work, but this might come at the cost of SNR degradation. At least I don't know a very low noise RF microphone that uses FM. :)