I have an Econet-based heatpump water heater, and I’ve been successfully controlling it with the esphome-econet integration. It only has two operating modes, “Off” and “Auto”. With the 2025.10 update, in “Auto” mode, I can’t set the temperature:
I have to set the mode to “Off”, before the temperature slider comes back and I can then set the temperature. Am I doing something wrong, is the integration doing something wrong, or is there a bug in the thermostat card?
@kbx81 seems to be here as well…perhaps he can explain the thought process behind removing this mode. It used to be described as something about AI and such in the older docs, now it just says automation and explicitly removes the set point. My guess he didnt know we were using it this way.
now
auto_mode (Optional, [Action]): The action to call when the climate device is placed into “auto” mode (it may both cool and heat as required). This mode is different from heat_cool_mode (above) in that it takes control of the set points away from the user; it’s generally intended for exclusive use by automations.
vs
used to be
auto_mode (Optional, [Action]): The action to call when the climate device is placed into “auto” mode (it may both cool and heat as required).
so looks like we can no longer use this in the same way…which is kind of confusing. I am not sure why this “fixed” but ideally we can revert or have some idea on how to move forward from @kbx81
As you’ve noted in the documentation, “auto” mode is for on-device control/automation of the set point(s); this behavior is now consistent with other integrations in Home Assistant.
I gather that, in the past, heat_cool and auto modes have often been conflated; we identify them as two unique modes of operation for the climate component.
To have two user-controllable set points in the UI, a component should use heat_cool mode, not auto mode; see the documentation. If the ESPHome component was (incorrectly) using “auto” mode for “heat/cool” mode, the component should be patched to use heat_cool mode which will permit user control of both set points in the UI.
If there is a specific ESPHome component that needs to be patched, please create an issue, identify the specific component which needs patching and tag me on it. I’m happy to do a patch. For external components, the external component author will need to update their component accordingly. If it’s just a matter of swapping in heat_cool mode for auto and the component isn’t doing something particularly unusual under the hood, the change is quite trivial (it’s just a flag).
So the problem is that the millions and millions of Mitsubishi pumps that use the cn105 protocol do not support dual set point - we have to fake it. They have modes that are heat/cool/fan/off and auto. They have built in logic that makes auto take 1 set point and for a +/-2C around it. This allows them to negotiate with multi head systems and make some heat up/cool down choices.
This is why we long ago (and I mean years now, I started working on this 7 years ago) implemented Auto as we did to match the way the pumps worked. For about 3 years now this Auto mode has been working exactly as we needed it to.
You have made a change in the framework that has broken the way we use this in real life because the documentation stated clearly the auto mode in esphome/ha was for… you guessed it auto mode; that is why I pasted the documentation above. None of us are understanding why you did this…
Can you explain why you removed this capability? Did you consult with any of us using it per the documentation? What is the process to have this rolled back to be useable? Who wanted this reduction in capability?
We have implemented dual setpoint heat/cool but now we have no way to use the primary working mode that Mitsubishi implemented in these devices. This seems really strange… and it was exactly per the documentation which existed for years.
We would like this rolled back/fixed or a carry forward plan because as I have written there, heat/cool/auto are very purposefully distinct things in the heat pump world.
Just to close the loop here, I discovered that the esphome-econet folks have fixed this problem in their component. I also do not agree with this reduction in features in the “Auto” mode, but at least for the water heater issue for me, it’s been solved.
I want to be sure I understand the issue correctly:
The Mitsubishi pumps only natively support a single set point.
The external component is designed to support a dual set point mode–in which the two set points are “faked”–which you access by setting the component to use “auto” mode.
Is this correct?
We have implemented dual setpoint heat/cool
What I do not understand is the difference between the Mitsubishi “auto” mode and the heat/cool mode you mention here? Is this anything more than cosmetic? (By “cosmetic” I mean you see/select “auto” mode in the UI but it’s really just doing what we define as “heat/cool” mode behind the scenes.)
Home Assistant defines “heat/cool” mode and “auto” mode as different things. The change I made simply brought the thermostat component to parity with that definition.