Bisher habe ich etwas mit den MKR-Boards geliebäugelt, da sie mit dem M0 die I2S-Mikros und die FFT-Lib von Arduino zur Analyse von Sound unterstützen, s. First Steps (digital) Sound Recording . Habe aber leider keine guten News über deren Stromverbrauch gehört und sie daher für den Batterie-Betrieb nicht in Erwägung gezogen. Vielleicht ändert sich das jetzt mit etwas mehr Konzurenz auf dem Markt, von Sparkfun und Moteino gibt es nun neue boards:
SparkFun Pro RF
Interessant hier:
Thanks to the Arduino LoRa library, the RFM95W radio is an easy to use packet radio.
Indeed, I have noticed the same with my Adafruit Feather M0 LoRa. I think it is good for prototyping but might not be good in the long run for small power. I haven’t found any low power libraries for it yet either. At some point I hope to migrate to e.g. Silabs EFM32, as that seems to be well equipped with various low power modes.
For the moment, moving from ICS43432 to ICS43434 will already help if you want to stick with an I2S microphone.
The CMWX1ZZABZ has a LoRaWAN layer embedded (unlike the RMF95). That might give some benefits in being able to put the M0 in sleep mode while not being concerned about how to deal with LMIC in the meanwhile, as you don’t need that anymore.
The ICS43434 has 490 uA / 230 μA and the ICS43432 1.0 mA current consumption. So in case we are not recording continously we should switch power for the I2S mics.
… that means recording and throw away the first part of the recording / FFT or do you have a tip to re-initialize the device or wait for some ms before reading?
The start-up effect takes a number of cycles, which depends on sampling frequency. I just looked over the incoming data (buffered on the MCU and then send over Serial, as the latter takes a lot of time) and concluded how much time needs to be disregarded. At 12500Hz this was 800 ms of data, last I checked.