I have send you email to speed up the process, you are not asking a lot, after all you did for this forum that is at least I can do
**Update: SAI/EVAP/N249/O2-delete write (mostly) successful, one stuck checksum sector, and testing a newer version for read stability**
Quick update on the 06A906032HN saga.
**The write (v1.9.4.3, battery charger disconnected):**
Did a Full Write Flash of my SAI/EVAP/N249/rear-O2-CAT delete file (patched onto my own ORI, then checksum-corrected with ME7Sum — verified 0 errors before flashing). Connection held for the entire ~12 minute write with zero TesterPresent disconnects.
However, 4 sectors near the top of the chip failed to erase and were skipped via the tool's "Sector Erase Failed → No" prompt:
- 0x8F0000-0x8F7FFF
- 0x8F8000-0x8F9FFF
- 0x8FA000-0x8FBFFF
- 0x8FC000-0x8FFFFF
Result: "Wrote 15 of 19 sectors." I checked with ME7Sum and confirmed these 4 sectors contain none of my intended codeword changes except the very last one, which holds the final ROM checksum at $FFFE0. So all 14 SAI/EVAP/N249/O2 codewords are correctly written, but the top-level ROM checksum is still the stock value — ME7Sum flags it as "1/1 uncorrected error."
**Real-world result:** Car starts and idles completely normally. Logged ~10 minutes total across two sessions (cold start through full warm-up to 105°C, plus some throttle blips up to ~2800rpm) — zero misfires, zero knock-related ignition retard, clean AFR control throughout. No P0605-type internal ECU error DTC either. So the one unwritten checksum block hasn't caused any noticeable runtime issue so far, but I'd still like to get it written correctly.
**Testing v1.9.6.0 for connection stability (read-only, no write attempted with it):**
Since erase kept failing specifically on those boot-block sectors, I tried v1.9.6.0 to see how it handles connections. Results so far:
- Connected instantly and reliably via KWP1281 → KWP2000, even with my battery charger still connected (which seemed to cause instability on 1.9.4.3 earlier the same day)
- Successfully read and cleared DTCs multiple times, no issues
- Ran a Full Read Flash: completed in 6:07 with zero disconnects, and the resulting bin matches byte-for-byte (verified with ME7Sum) the actual known state of my ECU (15/19 sectors patched, only that one checksum still stock)
So the read path on v1.9.6.0 seems fully reliable here, for both short and long operations. Haven't tried a write with it yet.
**Question for BlackT specifically:** You'd mentioned diff/full write hanging (ECU staying in programming mode) on recent NefMoto versions for the same 06A...HN ECU family, with 1.9.4.3 being fine. Do you remember which exact version(s) you saw that on? I'm currently sitting on v1.9.6.0 and trying to decide whether it's safe to attempt a write with it to retry that one stuck sector, or whether I should stick to 1.9.4.3 for any write and only use the newer build for reads.
