Lab 11.9 - AD9361 RF discovery sweep for the gpreg BPSK overlay

Lab 11.9 - AD9361 RF discovery sweep for the gpreg BPSK overlay

Goal

Prepare the reproducible RF search procedure that should be rerun once the sample-domain burst path is restored around the validated stock boot shell.

Engineering question

If the control plane is already alive, which small set of RF and timing parameters should be swept first to recover the first burst bits?

Current gate

Do not interpret all-zero RECEIVED_BITS from the current checked-in gpreg_only overlay as an RF result.

Before using this sweep as a meaningful hardware experiment, all of the following must be true:

1. the custom boot-time PL image passes validate_clean_boot_overlay.py with AD9361 initialized and all expected IIO devices alive;
2. the lab_11_8 helper sees a live sample-domain status path again, not only the reduced safety baseline;
3. RX capture still works after that clean boot.

Script

File	Purpose
blocks/block_11_integrated_sdr_project/python/lab_11_9_rf_discovery_sweep.py	sweeps START_OFFSET, RX manual gain and TX attenuation over SSH, runs the gpreg burst helper, and writes one JSON report

Current sweep strategy

Once the bridge is back, the first over-the-air search should vary only the parameters that are most likely to block deterministic recovery while still keeping the experiment safe:

1. keep AGC disabled;
2. keep TX LO enabled and TX attenuation conservative;
3. sweep RX gain in manual mode;
4. sweep START_OFFSET around the current synthetic reference value;
5. record the best tuple by RECEIVED_BITS first, then by timeout/error behavior.

That keeps the sweep explainable and avoids turning the first search into a blind optimization problem.

Example run

python blocks/block_11_integrated_sdr_project/python/lab_11_9_rf_discovery_sweep.py \
  --ssh-host 192.168.40.1 \
  --ssh-user root \
  --ssh-password analog \
  --start-offsets 48,54,58,62,66,70,74 \
  --rx-gains-db 10,20,30 \
  --tx-attenuations-db -80,-70,-60 \
  --json-out docs/assets/lab119_rf_discovery_sweep_live.json

Run that sweep only after the three preconditions above are satisfied.

Report contract

The JSON report should capture:

• board RF state before the sweep;
• board RF state after the sweep;
• every attempted parameter tuple;
• busy/done/timeout behavior for each attempt;
• RECEIVED_BITS, TOTAL_ERRORS and PAYLOAD_ERRORS;
• the best attempt chosen by the script.

Interpretation

The first good outcome is not necessarily zero BER. The first success criterion is smaller:

• RECEIVED_BITS > 0, even if partial;
• a repeatable best tuple across repeated runs;
• no obvious overload or unsafe gain settings.

Only after that should the course move to BER-oriented tuning and longer campaigns.

Engineering conclusion template

The discovery sweep tested ______ parameter combinations.
The best tuple was START_OFFSET = ______, RX gain = ______ dB, TX attenuation = ______ dB.
The burst recovered ______ bits, and the next step is ______.