ECH2O: the hot water figures were wrong since beta.23. The heat is computed from the sensor after the backup heater. On the ECH2O the heater sits in the tank and that sensor is on the tank coil's return while the unit makes hot water, so it reads close to the return water: 0.3 kW shown for about 8 kW made, a hot water COP of 0.1 to 0.2 (@leonidnf, #659). The ECH2O models now take the heat exchanger's outlet in every mode, and the tank heater's power counts as heat while it runs. It is what the unit's own software does for its counters: one sensor for every mode, the exchanger's outlet, plus the electric heaters that are on. Heating figures barely move (the two sensors agree within a few tenths of a degree in heating); every other unit is unchanged.
Hydro board software 20007902 is supported (the 11-16 kW EBB/EBV/EBS boards on their earlier software): leaving water and hot water targets, power limit, the unit's meter and counters. There is no image of this software: the addresses come from a memory dump by @griezla (#658), who then checked them on his EBVH16S23DJ9W: hours, energy and meter equal to the interface's, and the leaving water target written both ways (held in the working memory, saved in the settings) with the curve restored each time. The power limit is not tried there yet, and the limit the unit applies is not shown on this software.
A memory dump from the web page. Diagnostics → Memory read → From / To → Dump, then Download: a whole range of the hydro board's memory as a text file, read-only, while the values keep being read (16 KB take about two and a half minutes). It is how a software that is "not available" for the targets gets mapped: 0x0800 to 0x0FFF and 0x3000 to 0x5FFF, attached to a discussion with what the interface shows at that moment (setpoint modes, the points of the curves, the power limitation settings).
Fixes from @ppuetsch's review of the code (#657).
- A target or a limit held in the unit's working memory no longer saves the board's configuration at every change: it is kept in the board's RTC memory at once, and written to its flash after ten minutes without a change (an hour at most). An automation that moves a target every few minutes was wearing the flash.
- The event log takes one line per ten minutes for the target and limit changes, with a count of the others.
- With a long base topic, some MQTT topics and the discovery of the "Restore the unit's curve" buttons and of the firmware update entity were cut. They are sized for the longest base now.
- "Restore the unit's curve" acts on an explicit request only (a
falsein the request used to restore as well), asks for a confirmation on the page and writes who asked in the event log. - A restart command left retained on the broker restarted the board at every connection; one that arrives as the board subscribes is now ignored and logged. Only R, RESET, REBOOT and RESTART restart the board, no longer any text starting with R.
- The thermostat relay gets its kept state before its pin becomes an output: a low-triggered relay no longer closes for a moment at each restart.
- Smart Grid counts as configured with both pins set; the cap on a heat sample follows the model when it changes, not only at the start.
Smaller things. The Wi-Fi signal strength is read every 5 s instead of on every pass of the main loop (@ZakKemble measured the current it cost on a C3). An empty payload on the thermostat topic (a retained message being cleared) could switch the relay to what the previous message said: ignored now.
Update: System → Firmware update → Check for an update, or the web installer. Card: 1.0.36 on HACS (the Altherma HT's tank, 3-way valve and pump found by themselves, and its 0 / 1 sensors count as on; since 1.0.35 the plain target texts open their entity too).