Bela mini: audio input suddenly died

minnowahaw

Dear Giulio et alii,
For a wearable instrument I have been using a Bela mini for some realtime processing (in SuperCollider) of the audio that is coming in through the audio input and this setup has worked great until now.
After not having worked on the project for a couple of months I needed to continue, but I was not able to get the project to work. All of the sudden the audio input does not give any signal. I have checked the scope, I have updated the image, but not even the tiniest sign of life comes from the audio input. I tested my input signals on a normal Bela that I had laying around and there the audio flows in through the inputs as expected. I have checked all the settings, but I really suspect it to be a hardware issue somehow. But hopefully I am missing something. Is this something that happens more often? Any known solution or is the Bela mini now broken? It's strange, because it was just stored in a box in my studio.
Thanks for your reply!

Sjoerd

giuliomoro

What revision is your BelaMini cape? The revision number should be printed on the pcb close to the headers.

minnowahaw

giuliomoro the BelaMini cape is rev. B3 and the PocketBeagle is rev. A2

giuliomoro

rev B3 didn't have protection against overvoltage, these were added to rev B4 which started shipping in early 2021. Until then you had to make sure you respected the specified 1.8Vpp input range. In fact, it will probably tolerate just fine a 3.3Vpp input or slightly more, but anything higher than say 4Vpp could cause trouble. A piezo disc it too hard could be the cuplrit, or if you plugged something into it that has a loud power-on transient. Or simply driving it with a strong signal while Bela is powered down.

minnowahaw

giuliomoro inside my setup I was using the Bela mini with a very nice Schertler dyn uni p48 contact microphone in combination with an external phantom power unit. I have been using this setup for a few years for prototyping and performances with great satisfaction. It actually surprises me, because I made a LOT of harsh sounds through different piezo discs and the current fancy one over the years. So hearing your story - I would have expected it to die earlier then.

But then the newer Bela mini's have this protection against overvoltage, right? So I could better use a newer one for my project.

giuliomoro

yes from Rev B4 they have added protection besides what's on the codec, in the form of schottky diodes pointing towards GND and and the 3.3V rail, see:

minnowahaw ery nice Schertler dyn uni p48 contact microphone in combination with an external phantom power unit.

I am wondering if the issue here is about the preamplifier output providing a large voltage swing on startup. If that's large enough it could even damage the input capacitors , which would fail before the protection diodes could do anything about it ...

What preamplifier are you using and can you describe the wiring?

minnowahaw

giuliomoro I am actually using a Schertler DYN-UNI-P48 active contact microphone, which has a built-in pre-amp. I provide the Schertler with 48VDC with this external phantom power unit: XVive P1 Portable Phantom Power Supply
and from there I run the signal into the audio input of the Bela.

These are the specs of active Schertler contact mic:
DYN P48
Transducer type: Electrodynamic moving coil active contact microphone
Mechanical decoupling: Butterworth 2nd order, Q = 0.6
Frequency response (@ ±3 dB) - filtered: 20 Hz to 16 kHz
Sensitivity: ca. -30 dBu
Dynamic range: 139 dB
Output impedance (@ 1 kHz): 280 Ω
Operating voltage: 22 to 48 VDC

giuliomoro

-30dBu was likely measured at 94dB SPL and I take the 139dB dynamic range as meaning that it will not distort "significantly" until 139dB SPL (even though the SPL reference is missing there). So you could reasonably get a signal that is (139 - 94) = 45 dB larger than the sensitivity, so that gives you (- 30 + 45) = 15 dBu, or approximately 12V peak to peak (Vppp). That's about 20*log10(12/1.8) = 16 dB more than the specified 1.8Vpp input voltage. That means that even a (139 - 16) = 123 dB SPL signal hitting this microphone would be capable of producing dangerous (for Bela) voltage swings. That's still pretty loud, but I guess you can get that if you attach it to the soundboard of a piano or a drum head.

All of this before considering any potential spurious power-on voltage transient from the mic's preamp and/or the phantom power adapter or any failure thereof, and assuming I got the numbers right. A meagre consolation I guess, but Bela's from early 2021 onwards should all be protected against these issues.