hil: LNA A/B experiment scripts, refine health railed check

This commit is contained in:
Lars Nolden
2026-09-01 08:35:08 +02:00
parent a422305059
commit 78a8635ac8
5 changed files with 267 additions and 11 deletions
+37 -9
View File
@@ -45,17 +45,45 @@
## LIKELY
- Fixed 255-byte RX capture adds up to ~20 ms artificial dead time per
frame (by construction; to be measured).
frame (by construction; measured airtimes are consistent with it).
- The remaining CRC failures on long Techem frames are bit errors
accumulating over frame length, not a hard sensitivity floor: the same
meters decode cleanly when a few dB stronger (docs/RF_FRONTEND_TESTS.md).
## DISPROVEN
## CONFIRMED (measurement methodology)
- "rtl_sdr pipeline works out of the box": the PLL wedge cost ~20 minutes
of empty captures. Always energy-probe the SDR before trusting a null
result.
- Heltec RX is competitive with the Nooelec observer on identical traffic:
124 of 127 SDR-classified originals matched byte-exactly in a 30 min run
(97 % across 32 meters).
- Repeat payloads are byte-exact: 63/63 logged `tx_payload` values equalled
mode bytes + the received raw frame. No corruption in the RX->TX path.
- Our repeats DO radiate. IQ capture shows burst pairs at 54-55 ms spacing
with 9-10 ms duration (matches 100 kbps x ~100 bytes) - the repeat is
real RF, independently observed, even when rtl-wmbus cannot decode it.
- The observer, not the repeater, limits repeat decoding: our TX arrives
~36 dB stronger than meter traffic and sits at RTL-SDR full scale.
Raising the repeat delay 50 -> 200 ms improved SDR repeat decoding from
3 % to 30 % (demodulator level-estimate recovery).
- Synthetic C1 bursts from the Heltec (L = 19/40/60/98) are decoded by
rtl-wmbus as valid C1 with correct CRCs, including two airings of the
same frame 50/100/200 ms apart. TX framing is therefore correct and
standards-shaped independently of meter traffic.
- rtl_wmbus runs two demodulators, so one airing yields 1-2 output lines;
and rtl_sdr delivers ~81.92 ms buffers, so decode timestamps are
quantized - SDR timestamps cannot be trusted below ~100 ms resolution.
Correlation must use byte identity plus occurrence counting, not
timestamps alone.
- The Nooelec intermittently enters a deaf state (R820T I2C init failure /
PLL wedge). `hil/sdr_health.py` detects it by checking that noise tracks
tuner gain and the spectrum has structure; the harness aborts rather
than reporting an unattributable empty capture.
- "[R82XX] PLL not locked!" printed during tuning is BENIGN (emitted while
rtl_sdr probes and disables direct-sampling mode); it is not a wedge
indicator. The real symptom is railed, gain-independent noise.
## NEEDS TESTING
- Heltec vs SDR RX match rate on real meters (first HIL run in progress).
- LNA on/off comparison (KCT8103L CSD low in RX = bypass?).
- Repeat delay accuracy, RX deaf time distribution over many events.
- Long-run stability (>=100, >=500 repeats).
- LNA on/off comparison (`set fem_rx_csd 0|1` hook exists, not yet measured).
- Narrower RX bandwidth (156 kHz) vs current 234.3 kHz on weak meters.
- Long-run stability at >=100 and >=500 verified repeat events.
- Early-stop RX (poll GetRxBufferStatus) to cut the ~20 ms deaf tail.