wes/lippert-onecontrol
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

canbus: decode the water-heater DSI fault + add fault binary_sensors

Browse Source 
Forced a real lockout (captures/dsi-fault-*.log) and diffed vs the healthy
baseline:
- Water-heater DSI fault = node 95 page-3 b0 bit5 (0x20). Healthy 0x80/0x81,
  fault 0xA0. (Earlier suspects b1=FF and node AE were both wrong.)
- Bus-wide "system fault present" = page-0 b0 bit0 (every node flips 02->03).

Both wired into esphome/onecontrol-canbus.yaml as binary_sensor (device_class:
problem). README + HANDOFF updated; DSI item closed.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
This commit is contained in:
wes
2026-06-12 11:29:05 -04:00
co-authored by Claude Opus 4.8
parent f3073f180f
commit d57c2b21d2
3 changed files with 51 additions and 27 deletions
Download Patch File Download Diff File Expand all files Collapse all files
canbus/README.md
+15 -10
View File
@@ -187,18 +187,23 @@ the app uses to build its UI.
`6A`/`7F`/`9C` (movement — slide?), and find the battery SoC / "4 green lights"
source.
### Planned: capture the DSI fault (2026-06-12)
### DSI fault — decoded (2026-06-12, `captures/dsi-fault-*.log`)
The water-heater DSI fault is almost certainly on the bus, but every capture so
far is of a *healthy* heater, so the fault encoding is unknown. **Plan:** close
the propane tank valve, run the water heater on gas until it locks out (DSI fault
light on the panel), capture ~20 s, then diff against a healthy baseline. Prime
suspects (both sit at a constant "all-clear" value in current captures):
- **node `95` (heater) page-3 `b1`** — always `0xFF`; expect it to drop a bit on fault.
- **node `AE` (type 0x27, ?LP-gas/diagnostics) page-3** — always `0x00`; expect non-zero on fault.
Forced a real lockout (propane valve closed, heater run on gas until it gave up)
and diffed against the healthy baseline. Two signals:
Whichever flips → becomes a `binary_sensor` in the ESPHome node (the DSI fault
the Bluetooth app never exposed). Reset = reopen valve, re-light.
- **Water-heater DSI fault = node `95` page-3 `b0` bit5 (`0x20`).** Healthy heater
reads `0x80` (off) or `0x81` (running); during the lockout it read **`0xA0`**
(bit0 cleared, bit5 set) for every sample. Bit5 never appears healthy → it's
the gas-ignition lockout flag. (`b1` stays `0xFF` and `node AE` stays `0x00` —
the two earlier suspects were both wrong.)
- **Bus-wide "system fault present" = page-0 `b0` bit0 (`0x01`).** *Every* node's
page-0 `b0` flipped `0x02`→`0x03` during the fault, so any node carries a
generic "a fault exists somewhere" flag.
Both are wired into the ESPHome node as `binary_sensor`s (`device_class:
problem`) — the DSI fault the Bluetooth app never exposed. Reset = reopen valve,
re-light.
---