Besides the obvious reason of filtering Radio Frequency Interference from
external sources, there is another less obvious reason to keep C1 and C2 in the circuit. That is, if the V63m PCB layout is just as messy as that of the MXL 770. If you leave out C1, C2, C14 and C15, you will find the internal bias generator frequency on the outputs. This means the audio circuit is also passing 2.9 Mhz signals, which are not supposed to be there as they could cause intermodulation distortion. But alas, the signal is there and should be filtered, or else your microphone becomes a nice Short-Wave radio. Picture scope_15.png shows the output signals on the XLR with C1, C2, C14 and C15 removed and the JFET gate capacitively shorted to GND. The bias generator signal appears mainly as a common mode signal on the outputs. Note: ignore the small, very high-frequency ripple superimposed on the 2.9 MHz signal. I still have to figure out where it comes from, but it's not from the MXL 770 electronics.
With C14 and C15 back in, much of the 2.9 MHz signal has been filtered out, but not everything. See picture scope_17.png (Mind the different mV/div settings). With C1 and C2 also back in the circuit, the ripple was drowned out by the scope noise. In fact, with all the caps in, we now conceal the existence of the internal EMC issue, but (intermodulation) distortion is still lurking, although that will have improved as well. The caps will mainly prevent our microphone from becoming an RF jammer.
In the case of the MXL 770 PCB, the MXL designers violated at least three very basic EMC design rules in the layout, which caused the injection of the bias generator signal into the audio path :
- The 220uF decoupling cap for the oscillator circuit has been placed as far away as possible from the bias oscillator, whilst it should be placed as close as possible to the oscillator.
- GND and 6.8V power supply tracks to the oscillator should have the smallest possible loop area. Instead, the designer maximized the loop area by running one track along one half of the PCB circumference and the other track along the other half...
- The audio circuit and the oscillator should not have common GND lines, as they do now on the PCB. The oscillator should have its own GND line to its decoupling cap, which should not be part of the audio circuit.
Can this be patched easily? No, I'm afraid not... You'll have to redesign the PCB from scratch. I hope the V63m was better designed.
Btw, the effectiveness of C1 and C2 as RF filter is questionable, at least for UHF frequencies. Put your cell phone next to the mic and I would not be surprised you'll hear some TDMA noise. But cell phones did not exist at the time Schoeps designed this circuit, and the chinese copy-paste culture seems to perpetuate this design forever. For really good RF suppression in the UHF range, you'll need a common mode filter and possibly a ferrite bead between pin 1 and the PCBA ground. Among others, Takstar has done this with their CM60 and 63 designs. Well done for such a cheap microphone.