Composed by Naman // Claude
GS. — Observations from 26.85° N, 80.95° E
Register 21 on file

Log. Every capture, success or failure.

The full operational record of this instrument. Failures are kept on file with the same provenance as successful decodes — the register is honest about what didn’t work.

  1. 07 Decoded CADU written, image produced.
  2. 01 Partial Lock intermittent, CADU thin.
  3. 11 No lock Viterbi never synced.
  4. 02 Skipped IQ on file, decode deferred.
2026-05

May 2026

21 entries
No. Captured Satellite Outcome Band Frequency SNR CADU Pipeline / Note
21 2026-05-19 · 02:10Z GEO-KOMPSAT-2A Decoded L 1692.140 12.7 4772 KiB SatDump v2.0.0-alpha · Sanchez v1.0.26.1
20 2026-05-19 · 01:30Z GEO-KOMPSAT-2A Decoded L 1692.140 12.1 4651 KiB SatDump v2.0.0-alpha · Sanchez v1.0.26.1
19 2026-05-18 · 17:19Z GEO-KOMPSAT-2A No lock L 1692.140 0 KiB 5-second test capture from Pi. No CADU output, deframer NOSYNC, viterbi NOSYNC. Short duration likely insufficient for lock.
18 2026-05-18 · 14:39Z GEO-KOMPSAT-2A No lock L 1692.140 0 KiB FREQUENCY ERROR: tuned 1691.000 MHz but GK-2A requires 1692.140 MHz (1.14 MHz low). Good signal quality at wrong freq, Viterbi locked but deframer failed. Pi script needs frequency correction.
17 2026-05-18 · 13:30Z GEO-KOMPSAT-2A Decoded L 1692.140 12.5 4718 KiB SatDump v2.0.0-alpha · Sanchez v1.0.26.1
16 2026-05-18 · 13:20Z GEO-KOMPSAT-2A Decoded L 1692.140 12.4 4714 KiB SatDump v2.0.0-alpha · Sanchez v1.0.26.1 (blue underlay, haze 0.6)
15 2026-05-18 · 13:20Z GEO-KOMPSAT-2A Decoded L 1692.140 12.4 4714 KiB SatDump v2.0.0-alpha (no post)
14 2026-05-18 · 12:10Z GEO-KOMPSAT-2A Decoded L 1692.140 12.0 4689 KiB SatDump v2.0.0-alpha · Sanchez (red-blue gradient)
13 2026-05-18 · 02:41Z GEO-KOMPSAT-2A No lock L 1692.140 0 KiB 15-min capture (timeout 900, ~14.7 min recorded). Canonical decode (no dc_block) Peak SNR 1.53 dB, viterbi BER mean 0.24, deframer 100% NOSYNC, 0 CADU. Cross-check decode with --dc_block=true gave 13 KB CADU / 13 ASMs (all real GK-2A: SCID 195, mix of VCID 0+63). Link margin ~158x worse deframer yield than 2026-05-17 first-light at same flags. Positioning change did not help. Same ~1.5 dB Peak SNR ceiling as 2026-05-17. SD card moved physically to NAS (wifi too slow). IQ retained for re-decode.
12 2026-05-17 · 15:31Z GEO-KOMPSAT-2A Partial L 1692.140 2055 KiB SatDump gk2a_lrit baseband, cu8 @ 2.4 Msps, gain 25.4, dc_block=true. Viterbi SYNCED continuously, deframer never locked. 2.1 MB CADU written but 31 LRIT CRC fails / zero valid LRIT packets / no images. SatDump Peak SNR plateau 0.982 dB. FFT band-power confirms real GK-2A signal centered exactly at 1692.14 MHz (BPSK-shaped ±100 kHz envelope, stable through full 20 min) but only +3.6 dB above edge band-avg / ~+5 dB above noise at peak bin — below lock threshold. Diagnosis: link budget shortfall (likely pointing alignment / polarization sense / coax loss), NOT decoder config. IQ pulled from Pi SD card mounted at /mnt/pi-sd. IQ moved to needs-review.
11 2026-05-17 · 06:39Z ELEKTRO-L3 No lock L 1691.000 0 KiB SD-card transfer (Pi powered off post-capture). 600s @ 1691.0 MHz, 2.4 Msps, gain 40, cu8. SatDump elektro_lrit pipeline (BPSK 294 ksps, conv+concat, dc_block=true): viterbi NOSYNC + BER ~0.43 throughout, 0 CADU. FFT band-power probe (band_power_probe.txt in pass dir): LRIT band vs far-edge gap 0.2-0.4 dB at t=60/300/540s -> no LRIT signal in IQ. L5 retry skipped: elektro_lrit pipeline is identical for L3/L5; no satellite-id distinction at modem layer. Lone -503 kHz narrow spur near end of capture, unrelated. IQ -> needs-review/ for 7-day retention. Suspected causes upstream of decoder (chain/pointing/LNA/no-broadcast-slot); not investigated this session.
10 2026-05-17 · 06:39Z ELEKTRO-L5 No lock L 1691.000 0 KiB SatDump v2.0.0-alpha (elektro_lrit, dc_block=true)
09 2026-05-14 · 22:27Z METEOR-M N2-4 No lock VHF 137.900 0 KiB M2-4 LRPT, --lna --gain 25 (pole, V-dipole). SatDump v2.0.0-alpha-8ce9d5b, --dc_block=true. Peak SNR 2.2 dB, real SNR 0 dB, 0 CADU. FFT verification: LRPT band -54.83 dB, edge -55.3 dB → +0.5 dB above edge (noise only). Tests May 14 plan (--lna --gain 25 to recover from front-end compression); plan did NOT recover signal. IQ in needs-review for re-decode/analysis.
08 2026-05-14 · 20:47Z METEOR-M N2-4 No lock VHF 137.900 0 KiB M2-4 LRPT, --lna --gain 25 (pole, V-dipole). SatDump v2.0.0-alpha-8ce9d5b, --dc_block=true. Peak SNR 3.6 dB, real SNR 0 dB, 0 CADU. FFT verification: LRPT band -51.44 dB, edge -55.7 dB → +4.3 dB above edge (CLAUDE.md threshold ≥5 dB). Borderline noise, no decode. Tests May 14 plan (--lna --gain 25 to recover from front-end compression); plan did NOT recover signal. IQ in needs-review for re-decode/analysis.
07 2026-05-14 · 15:41Z METEOR-M N2-3 No lock VHF 137.900 0 KiB Pole-mounted V-dipole + --lna --gain 25 (noise-floor experiment from 2026-05-14 plan). SatDump --dc_block=true. 0 viterbi locks, BER stuck ~0.41. IQ moved to needs-review for 7-day retention. dataset.json mis-IDed as METEOR-M1 (expected when no frames decode). Pi REPORT.md claimed 44.2 dB SNR / polarization mismatch — DC-spike artifact per CLAUDE.md memory.
06 2026-05-14 · 06:18Z METEOR-M N2-4 Skipped VHF 137.900 0 KiB DIAGNOSTIC capture (not a pass): 60s baseline noise at 137.900 MHz, gain=25 (no LNA), V-dipole E-W at POLE position (~3.6m). Part of noise-floor experiment 2026-05-14. Noise floor (edge 300-500 kHz): 46.16 dB — 7.5 dB QUIETER than stool at same gain. Overturns prior hypothesis that pole-height move raised the floor. See /satellite/_diagnostics/2026-05-14_noise-floor-experiment/REPORT.md
05 2026-05-14 · 06:14Z METEOR-M N2-4 Skipped VHF 137.900 0 KiB DIAGNOSTIC capture (not a pass): 60s baseline noise at 137.900 MHz, gain=25 (no LNA), V-dipole E-W at ORIGINAL STOOL position (~1.2m). Part of noise-floor experiment 2026-05-14 to disentangle gain/height/environment. Noise floor (edge 300-500 kHz): 53.63 dB (matches May 12 baseline within 0.4 dB). See /satellite/_diagnostics/2026-05-14_noise-floor-experiment/REPORT.md
04 2026-05-13 · 21:09Z METEOR-M N2-4 No lock VHF 137.900 0 KiB 17.9 min capture, no Viterbi lock (0/34), CADU 0 B. CORRECTED DIAGNOSIS: NOT polarization mismatch. FFT shows noise floor at ~76 dB across whole 1 MHz baseband; May 12 IQ at same SDR/freq had 54 dB floor (delta +21 dB). Pi REPORT 'avg SNR 19.7 dB' is DC-spike artifact (cu8 zero-bias, fixed 88-dB single-bin at f=0) mistaken for signal — same misread documented for 2026-05-13 entry. Probable primary cause: tuner gain 44.5 dB drives R828D front-end into compression on horizon-level interferers newly visible to 12-ft antenna; lowering gain to ~25 dB is the first thing to try. Re-decode of preserved May 12 IQ through current SatDump yielded 2.2 MB CADU (vs 179 KB original) — processing is verified fine. Image data at /satellite/meteor-m2-4/2026-05-12_1435/_redecode_2026-05-14/MSU-MR/.
03 2026-05-13 · 08:44Z METEOR-M N2-4 No lock VHF 137.900 0 KiB V-dipole N-S orientation (single-variable change from 2026-05-12_1435 E-W partial baseline). Pass cut short at 14:30 IST (≈16 min real signal + 2 min noise tail) due to IMD lightning warning; antenna retrieved early. Peak SNR 4.3 dB, Viterbi BER ~0.41 throughout (no lock), deframer NOSYNC entire pass. Pi-side REPORT.md monitor SNR (26.7 dB avg) is DC-spike-dominated and misleading; lock%=0 is truer. Pi setup.txt was unedited template (known template-leakthrough issue) — N-S orientation note exists only in operator message. SatDump auto-ID says METEOR-M1 in dataset.json which is a fallback on no decoded channels; pipeline run was meteor_m2-x_lrpt (no _72k variant exists in this build). 2026-05-14: re-decoded with 4 SatDump flag combos (--dc_block on/off × --iq_swap on/off, plus --rs_usecheck + --fill_missing). Best result was --dc_block=false --iq_swap=true: 8 brief Viterbi locks with BER ~0.03 but 0 deframer locks → 0 CADU. No images possible. Retry artefacts at /satellite/meteor-m2-4/2026-05-13_1414/_retry_2026-05-14_no_dc_block_iq_swap/. IQ moved to needs-review for 7d retention.
02 2026-05-12 · 14:54Z METEOR-M N2-3 No lock VHF 137.900 0 KiB polarization mismatch per REPORT.md (V-dipole vs RHCP); IQ never transferred to NAS; AOS from REPORT.md
01 2026-05-12 · 09:05Z METEOR-M N2-4 Decoded VHF 137.900 7.5 2176 KiB SatDump v2.0.0-alpha (--dc_block=true)