audiobusio.PDMIn not working on RP2040 since v7.2.0

[caternuson]

CircuitPython version and board name

Adafruit CircuitPython 7.2.0 on 2022-02-24; Adafruit QT Py RP2040 with rp2040

Code/REPL

import time
import array
import math
import board
import audiobusio


# Remove DC bias before computing RMS.
def mean(values):
    return sum(values) / len(values)


def normalized_rms(values):
    minbuf = int(mean(values))
    samples_sum = sum(
        float(sample - minbuf) * (sample - minbuf)
        for sample in values
    )

    return math.sqrt(samples_sum / len(values))


# Main program
# PDMIn(clock, data)
PDM_CLK = board.SCK
PDM_DAT = board.MISO
mic = audiobusio.PDMIn(PDM_CLK, PDM_DAT, sample_rate=16000, bit_depth=16)
samples = array.array('H', [0] * 160)


while True:
    samples_recorded = mic.record(samples, len(samples))
    magnitude = normalized_rms(samples)
    print(samples_recorded, (magnitude,))
    time.sleep(0.1)

Behavior

The values do not change as expected. A loud noise is made after pressing enter on the import and seeing print statements starting to show up:

Adafruit CircuitPython 7.2.0 on 2022-02-24; Adafruit QT Py RP2040 with rp2040
>>> import pdm_test
160 (32594.5,)
160 (31913.7,)
160 (32504.2,)
160 (31898.3,)
160 (32515.3,)
160 (31882.4,)
160 (32516.7,)
160 (31848.7,)
160 (32413.2,)
160 (31906.9,)

For reference, here is expected/working output when running 7.1.1 firmware:

Adafruit CircuitPython 7.1.1 on 2022-01-14; Adafruit QT Py RP2040 with rp2040
>>> import pdm_test
160 (72.2027,)
160 (66.0032,)
160 (66.1652,)
160 (60.7902,)
160 (646.89,)
160 (7786.59,)
160 (1313.55,)
160 (185.272,)
160 (64.3999,)
160 (63.2283,)
160 (61.9871,)

Description

PDM mic not returning expected values.

Additional information

Here is the test setup:

This behavior is also seen running latest (10.0.3) firmware. The above output focusing on 7.2.0 was determined by testing older releases to find when the change was introduced.

This was also tested on a Feather M0 Express but the same behavior was not seen there. It worked as expected for latest (10.0.3) release.

[jepler]

FWW there seem to be two PRs that changed PDMIn from 7.1.1 to 7.2.0:

• Correcting order of sideset parameters in function call #5984
• RP2040 PDMIn frequency based on sample rate #5993

neither is jumping out at me as a likely cause, though maybe something about the preexisting incorrect call order of #5984 was actually required?

If printing the received samples directly, does it show any interesting or obvious pattern? I am guessing that it includes both "very big" and "very small" values, so that the RMS value is around 32000 instead of 0 but I can't guess why.

[jepler]

Please have a look at the build artifacts in #10803 and see whether they resolve the problem for you.

[caternuson]

thanks @jepler !