Any update here? The calibration loss is really annoying. The smartmode is not working at all. There is always a big overshoot, and maintaining the temp is poor.
What we need is:
A software calibration trigger so that we can recalibrate the valve by need.
A force close mode so that the valve can close with more force to not loose its calibration.
A possibility to tweak PID parameters so that we can finetune the smart mode.
A true “valve position” expose, so that we can see the real position.
I disasembled a TRVZB and had a look on its hardware. I just say, there is good hardware built in. Especially the stepper motor. With this one, we could archive the best results in precise control. We just have to enable more current to the motor.
Are there any experience about valve deviations with the smart mode? I guess so, because the normal mode if you set opening and close percent it is affecting it also.
The main problem is, if the valve does really small steps, it will loose its calibration. For a precise control, you have to do really small steps. like 4-8%. if your radiator will go from like 35-45°C all the time, the temperature in the room will still fluctuate alot.
I tried out the smart mode, and it was much more worse then my own PI controller controlling opening and closing percent.
I know such a heatingsystem can be very different.
Gasboilers turning on and off, central heating systems with high forwarding temperature, or low temperatures, big sluggish radiators or small agile radiators.
Sonoff had to find the golden middle.
BUT you should have an option to tweak the parameters for experienced users.
Give them a “smart agressive, slow, normal” mode, or an advanced mode where you can directly set PID params
From my testing, it isn’t the small movements that cause the valve to lose calibration. It’s the valve closing repeatedly from a small opening position to fully closed (eg from 5% to 0%). I have had valves running with 1% steps for several weeks now without them going out of calibration in Home Assistant. I use an automation which fully cycles the valve whenever the controlling thermostat sets the valve to zero. You can read about that here if you are interested.
My assumption is that this is due to how the TRVZB detects it has reached the closed position - I don’t know if it can actually sense the motor position directly or if it relies on sensing voltage/current change when it meets resistance from the valve body.
The other problem with controlling the valve externally is that most valves have the majority of their opening effect in a very small range of motion (the standard for radiator valves is something like 0.45mm with a 2º temperature error). A single 5% adjustment step could take you from no flow at all to more than half, or it could make almost no difference at all if the valve is already past its linear response zone.
I’ve read through the testing you have done, from the link you provided.
I was considering some basic testing myself, to find how the opening movements of both lockshield and TRV affected flow.
That info is really useful and you have already done the testing.
So thanks for that.
