run 1

2026-05-14 01:55:36 +10:00
parent d169cc4c39
commit afdd93bd07
8 changed files with 868 additions and 0 deletions
--- a/docs/discovery-notes.md
+++ b/docs/discovery-notes.md
@@ -9449,3 +9449,151 @@ Current read after including the paused runs:
  across parallel surfaces
 - so the "semi-awake" question is probably about which of these surfaces is the
  right maintained background class, not about whether the map exists at all
+
+### HE39: Narrow In On The `0x50 24 DD 74` Band
+
+Working idea: the `0x40-0x5F @ 20 D0` region may be a stronger maintained
+background class than the earlier low-band `0x40 24 DD 64` region.
+
+Known lead so far:
+
+- `0x41 @ 20 D0` -> `07 80 50 24 DD 74`
+- paused run also showed:
+  - `0x45` -> `07 80 51 24 DD 75`
+  - `0x49` -> `07 80 52 24 DD 76`
+  - `0x4D` -> `07 80 53 24 DD 77`
+  - `0x59` -> `07 80 56 24 DD 72`
+  - `0x5D` -> `07 80 57 24 DD 73`
+
+Goal: see whether this band:
+
+- holds the panel clear longer,
+- behaves better as a maintained subset,
+- or opens anything richer than the earlier `0x40 24 DD 64` surface.
+
+#### HE39a: repeat only `0x41 @ 20 D0`
+
+Target frame:
+
+- `00 00 41 20 D0 EB`
+
+```powershell
+python scripts/serial_probe_response.py --port COM5 --tx-frame "00 00 41 20 D0 EB" --repeat 30 --interval 0.60 --delay 3 --after 3 --frame-size 0 --log captures/he39-repeat-cmd41-20d0.txt
+```
+
+#### HE39b: repeat only `0x45 @ 20 D0`
+
+Target frame:
+
+- `00 00 45 20 D0 EF`
+
+```powershell
+python scripts/serial_probe_response.py --port COM5 --tx-frame "00 00 45 20 D0 EF" --repeat 30 --interval 0.60 --delay 3 --after 3 --frame-size 0 --log captures/he39-repeat-cmd45-20d0.txt
+```
+
+#### HE39c: repeat only `0x49 @ 20 D0`
+
+Target frame:
+
+- `00 00 49 20 D0 E3`
+
+```powershell
+python scripts/serial_probe_response.py --port COM5 --tx-frame "00 00 49 20 D0 E3" --repeat 30 --interval 0.60 --delay 3 --after 3 --frame-size 0 --log captures/he39-repeat-cmd49-20d0.txt
+```
+
+#### HE39d: small `0x50`-band mixer
+
+```powershell
+python scripts/serial_sequence_probe.py --port COM5 --prompt --prompt-screen --pre-read 2.0 --frame "00 00 41 20 D0 EB" --frame "00 00 45 20 D0 EF" --frame "00 00 49 20 D0 E3" --repeat 10 --frame-interval 0.60 --read-after-frame 0.20 --read-after-group 0.80 --log captures/he39-mixer-41-45-49-20d0.txt
+```
+
+#### HE39e: wide `0x50`-band mixer
+
+```powershell
+python scripts/serial_sequence_probe.py --port COM5 --prompt --prompt-screen --pre-read 2.0 --frame "00 00 41 20 D0 EB" --frame "00 00 45 20 D0 EF" --frame "00 00 49 20 D0 E3" --frame "00 00 4D 20 D0 E7" --frame "00 00 59 20 D0 F3" --frame "00 00 5D 20 D0 F7" --repeat 6 --frame-interval 0.60 --read-after-frame 0.20 --read-after-group 1.0 --log captures/he39-mixer-50band-wide-20d0.txt
+```
+
+#### HE39f: control repeat from the older low-band lead
+
+Target frame:
+
+- `00 00 01 20 D0 AB`
+
+```powershell
+python scripts/serial_probe_response.py --port COM5 --tx-frame "00 00 01 20 D0 AB" --repeat 30 --interval 0.60 --delay 3 --after 3 --frame-size 0 --log captures/he39-repeat-cmd01-20d0-control.txt
+```
+
+#### Recommended order
+
+1. `HE39a` repeat `0x41`
+2. `HE39f` control repeat `0x01`
+3. `HE39b` repeat `0x45`
+4. `HE39c` repeat `0x49`
+5. `HE39d` small `41/45/49` mixer
+6. `HE39e` wide `0x50`-band mixer
+
+Interpretation guide:
+
+- if `0x41/0x45/0x49` hold the panel clear longer than `0x01`, the `0x50` band
+  is probably a better maintained-background candidate
+- if the mixer behaves better than any one frame, that supports a small
+  recurring in-family scan model
+- if everything behaves the same as `0x01`, then this is probably just another
+  parallel mapped surface, not a privileged wake-adjacent one
+
+#### HE39 results
+
+Observed panel behavior across all HE39 runs:
+
+- all runs kept the panel in the clear / semi-awake state for the duration of
+  the script
+- none produced a fuller wake-up beyond that state
+
+Single-frame repeat results:
+
+- `HE39a` repeating `00 00 41 20 D0 EB`:
+  - first send: heartbeat only
+  - second send: opens `07 80 50 24 DD 74`
+  - several copies of that family then drain out
+  - later repeats stay heartbeat-only
+- `HE39b` repeating `00 00 45 20 D0 EF`:
+  - first send: heartbeat only
+  - second send: opens `07 80 51 24 DD 75`
+  - then returns to heartbeat-only behavior
+- `HE39c` repeating `00 00 49 20 D0 E3`:
+  - first send: heartbeat only
+  - second send: opens `07 80 52 24 DD 76`
+  - several copies drain out, then heartbeat only
+- `HE39f` control repeating `00 00 01 20 D0 AB`:
+  - same overall pattern as before
+  - second send opens `07 80 40 24 DD 64`
+  - later repeats are heartbeat-only
+
+Mixer results:
+
+- `HE39d` small `41/45/49` mixer:
+  - group 1 only:
+    - `0x41`: no RX
+    - `0x45`: opens `07 80 51 24 DD 75`
+    - `0x49`: drains more `07 80 51 24 DD 75`
+  - later groups are heartbeat-compatible only
+- `HE39e` wide `41/45/49/4D/59/5D` mixer:
+  - same practical pattern as the small mixer
+  - group 1 opens on `0x45` into `07 80 51 24 DD 75`
+  - later `0x49/4D/59/5D` frames do not open additional stages
+  - later groups are heartbeat-compatible only
+
+Takeaways:
+
+- the `0x50`-band is real and coherent:
+  - `0x41` -> `07 80 50 24 DD 74`
+  - `0x45` -> `07 80 51 24 DD 75`
+  - `0x49` -> `07 80 52 24 DD 76`
+- but it behaves serially just like the older low-band `0x40 24 DD 64` lead:
+  - an early one-shot family burst
+  - then flat heartbeat-compatible maintenance
+- so the "keeps the panel alive" effect is broader than one specific command
+  slot inside the `20 D0` surface
+- the mixers did not outperform the simpler repeats
+- current best read: this is another maintained-background surface, not yet a
+  privileged wake-up band
--- a/docs/pt2-protocol-summary.md
+++ b/docs/pt2-protocol-summary.md
@@ -494,6 +494,17 @@ Current caution:
 - So the best current model is a parallel set of response surfaces selected in
  part by the host payload pair, with the remaining question being which surface
  behaves like the true maintained background layer for an active panel.
+- A tighter HE39 pass on the `0x50`-band showed:
+  - `0x41 @ 20 D0` -> `07 80 50 24 DD 74`
+  - `0x45 @ 20 D0` -> `07 80 51 24 DD 75`
+  - `0x49 @ 20 D0` -> `07 80 52 24 DD 76`
+  - all of those can keep the panel in the clear / semi-awake state while the
+    script is active
+  - but serially they still behave as one-shot branch openings followed by
+    heartbeat-compatible maintenance, not as a richer multi-turn wake sequence
+- So the `0x50`-band currently looks like another valid maintained-background
+  surface inside the broader `20 D0` class, not yet a uniquely privileged
+  "real wake" surface.

 ## What We Do Not Know