Capture IQ to a file

tools/sigmf-capture.py records IQ into a file that describes itself and says whether samples were lost.

# run from: the repo root
tools/sigmf-capture.py record out --rate 3e6 --freq 900e6 --seconds 5 \
    --channels both --verify --annotate

That writes out.sigmf-data (the samples, untouched) and out.sigmf-meta (JSON describing them, including a verdict on whether the capture is intact). MATLAB reads the same files with fishball.readSigMF.

You should see: the --verify verdict in the output and in out.sigmf-meta. This is a real run, both receivers at 3 MSPS, with the chip's built-in test tone injected and without --annotate:

# run from: the repo root - output of the command above, captured from the board
recording 15000000 instants (5.00 s) x 2 channels at 3.000000 MSPS = 24.0 MB/s, 120.0 MB total
+ iio_readdev -u ip:fishball.local -b 1048576 -s 15000000 cf-ad9361-lpc voltage0 voltage1 voltage2 voltage3 > out.bin
captured in 5.61 s wall (21.4 MB/s sustained)
verify RX1: continuous
verify RX2: continuous
out.sigmf-data  120.0 MB, 15000000 sample instants x 2 channels, 5.000 s
out.sigmf-meta  3.000000 MSPS, 900000000 Hz centre, RX1 + RX2
ok   integrity RX1: continuous
ok   integrity RX2: continuous

Good to know
lost samples	the capture still completes; only --verify tells you
receive alone sustains	about 40 MB/s: 1 channel at 10 MSPS clean, 2 channels at 10 MSPS drop samples; 2 channels at 3 MSPS clean
full scale	±2047 (12-bit): dividing by 32768 reads every level 24 dB low
why not iio_readdev	a bare capture has no sample rate, frequency or gain in it, and returns the byte count you asked for whether or not the hardware kept up

How do I find my own rate limit?

# run from: the repo root
for r in 3e6 5e6 10e6; do
  tools/sigmf-capture.py record /tmp/t --channels both --rate $r --seconds 2 --verify
done

Reading a capture back, --verify and --annotate in detail: capturing IQ.

Next: receive on both channels.