TOPIC:

I have a hardware module with 5 PDM MEMS Microphones coming into the MCHStreamer, using data lines (1+2), (3+4) and (5+6). One mic in each pair is tied to GND, the other High.  They all share a clock line, total wire length is around 30cm and scoped that clock looks good (no ringing, overdamping, etc), and data lines look good as far as I can tell.  Everything is working except line 3 is has a noise floor significantly higher than any other input line, on the order of -32dBFS while other lines are -80dBFS in a quiet room. It still receives signal if that signal is high enough above the noise floor. As far as I can tell this is a general MCHStreamer issue, as here's what I've tried for debugging:

- This has been the same on two MCHStreamer boards. 
- I can take my same mic array and plug it into an NXP DSP Eval board and there is no noise issue on any channel. 
- I can take one my mic pairs (3+4) and swap one wire to put it into line (7+8) and then there is no noise issue on (7+8). 
- I can take a different pair (1+2) and swap it into (3+4) and it has the noise issue on line 3. 


 

I'm having the same issue, has anyone replicated this or found a solution?

Most likely a hardware/layout issue.. You can see the UMA16 (i.e. 16ch MEMs all working in unison). No noise issue and the MCHStreamer PDM is basically the same as UMA16  
So don't think that it's a MCHstreamer generic issue. 

Now over the years that we've done PDM mics on many boards, we did see some issue in one layout with that. Basically PDM is going high speed, you do have to be careful with layout when running these lines. We do have ground isolation between the tracks on the UMA16. Worth for you to investigate. 

DevTeam