Repository navigation
Conversation
Signed-off-by: Segran <[email protected]>
Signed-off-by: Segran <[email protected]>
Signed-off-by: Segran <[email protected]>
ef41395 to
a5ded98
Compare
miravoss26
left a comment
There was a problem hiding this comment.
Pixii 2.1.10: reads StVnd (40108) and treats code 116 as calibration, so dispatch is suspended when SunSpec ChaSt is unsupported. Version, index, manifest and sha256 are bumped consistently, and the new test covers 116, 0xffff (keeps the last verdict) and 119 (clears).
- Needs a human decision: this adds
host.set_calibratingtospec/host-api-profile.json. Older Core falls back toset_device_fault, which is fine. But is the Core side that implementsset_calibratingmerged or planned? If not, the new API is spec-only for now. - Nit: the 101-129 range for
service_knownis from the Pixii mapping (table 23). A one-line link in the comment would help whoever extends it. - No secrets, no new hosts or dependencies.
Logic looks right to me. Safe to merge from my read once the Core question is answered.
|
Thanks, Mira. I haven’t confirmed whether |
miravoss26
left a comment
There was a problem hiding this comment.
Treats Pixii StVnd (40108) == 116 as battery calibration as well as ChaSt == 7, so dispatch still pauses when the SunSpec status reads 0xffff. Uses host.set_calibrating when the host has it and falls back to the device-fault gate on older cores.
Read the whole diff, found nothing blocking.
- Security: nothing. No new hosts, deps or secrets.
- Logic: the verdict only updates when
ChaStis recognised orStVndis in 101–129, and holds the last verdict otherwise. That matches the test (0xffff keeps calibrating, 119 clears it). - Low, host support:
set_calibratingis a new entry inhost-api-profile.json. The fallback covers older hosts, but it's worth confirming ftw-core actually implements it. Otherwise the "legacy fault" path stays the only live one. - Low, test gap: the fallback branch (
set_calibrating == nilwith service 116) andChaSt == 7through the new code path aren't covered by the new test. Both look right on read.
Safe to merge from my read.
On a Pixii undergoing scheduled SoC calibration, a read of SunSpec 802 register 40137 (
ChaSt) returned0xffffwhile Pixii register 40108 (StVnd) returned 116, the documented battery calibration service. FTW kept requesting idle and the battery continued charging under its local calibration.Recognize either calibration signal and keep battery and meter telemetry live. Use the Core calibration state when the host provides it. Older hosts continue to suspend control through the existing fault gate.
On the affected Pixii, the local driver derived from this change reported service 116 and stopped dispatching power commands once calibration was detected. The Pi journal shows no Pixii power setpoints between calibration detection at 01:24:41 and the status transition at 03:00:11 CET on 15 November 2026. The same FTW process resumed dispatch automatically and issued a 463 W Modbus charge setpoint at 03:00:12, then 0 W at 03:00:17. A later live read confirmed calibration was cleared and battery telemetry remained fresh.
Companion Core change: srcfl/ftw#2017.