My single fancoil, a mitsubishi i-life2 slim, is nearly perfect, but not quite. The one problem is that it switches on only when the water temperature exceeds 30ºC (in heating mode) or is less than 20ºC (in cooling mode). This is controlled by a 10K thermistor connected to the management/display board
Unfortunately my ideal WC curve goes down to about 27ºC and of course the fancoil doesnt work then.
I’m trying to come up with a simple circuit (powered if necessary) that I could interpose between the thermistor and the management/display board, to fool the unit into switching on, when in heating mode, at say 25ºC, without changing the behaviour when in cooling mode. I can think of several complex ways of doing it, but not a simple way. Obviously one option is simply to switch in/out manually a parallel resistor, but I want it to be automatic because it seems somewhat naff and confusing to others for it to need to be switched manually,
Does anyone have any ideas/has anyone done something similar before?
You need a different thermistor having the same resistance as the fitted one at 20°C, but at 27°C it must be the same as the fitted one has at 30°C. Or if you can’t find one, a thermistor and series or parallel resistor combination which gives you the required resistances at those temperatures.
Thanks. I am not sure thats actually possible. To do that the change in resistance of the combination from 27-20 needs to be the same as the change from 30-20 of the original, ie the resistance needs to change more quickly. However all common thermistors appear to have the same ‘B value’ which means that, to get a thermistor with a greater slope it needs to be higher resistance. To drop this back down would require a parallel resistor, which reduces the slope back down. I cant prove mathematically that there is no combination which works, but a bit of playing on a spreadsheet suggests that there isn’t.
A simple differential amplifier is probably the easiest way to do it. Set the pivot voltage at the 20ºC output and calculate the gain required to bring the 27ºC output voltage to the 30ºC voltage. Your input voltage below 20ºC will change more quickly, but I don’t think that’s a problem.
I have the same fancoils. I simply unplugged the thermistor from the board and the unit will run. I plugged the mains feed into a smart plug which I can switch using a schedule or a sensor in home assistant.
I was wondering about simply unplugging but was worried that the fan wouldn’t stop during defrost, when the flow temperature goes down quite a lot. According to the manual it only detects presence on switch on so you cant switch it in and out. That said I haven’t actually verified this. They really should have made the limit temperatures configurable!
In practice I’ve not found defrost cycles to be a problem. The Mitsubishi units also retain their settings for fan speed and temperature when the power is turned off so a smart switch works well for additional control.
Another other angle Im going to explore is the solenoid valve output for DLRV versions, which is not used on my version.
The data book for the i-life series suggests that this switches auxiliary fans on and off that are used only in heating mode, to heat the front of the casing (it seems that R stands for radiation). If so then it could potentially form a signal connection distinguishing whether the unit is in heating or cooling mode, and this could then be used to switch a relay controlling whether a series resistor is introduced between the thermistor and the board or not.
Did some checks on mine, the solenoid connector (230V) does indeed seem to follow the cooling vs heating selection however the behaviour is a bit complex:
When you first switch the unit on in heating mode its live. It goes dead if you switch it to cooling mode but doesn’t switch on if you switch it back to heating and the water temperature is <30C. I am guessing it does come back on when the water temp exceeds 30C but haven’t yet been able to verify this.
If you disconnect the thermistor so that the controller ignores the water temperature limits, then the solenoid connector is on if cooling is selected and off if heating is selected.
In summary I think its suitable for the proposal above. Now to find a suitably compact, PCB mounted relay with 230V coil (or solid state equivalent) and work out a way of using the existing connectors non-destructively.
Lots of good information and ideas here @jamespa If you’d like any more information on my suggestion of the differential amplifier, let me know. It would be a grand total of 6 resistors, and an opamp so you’d have change from 20p in its simplest form
Yes, fair point - if the actual voltage that’s sensed is internal, which is quite likely, with the thermistor being one leg of a voltage divider you wouldn’t be able to tap that externally with an amplifier
It could even be the thermistor is driven by a digital output and it’s charging a capacitor and they’re using a digital input to time how long it takes for it to switch! (I’ve heard of the technique, never come across it.)
Yes, its a 10K NTC thermistor, stated in the manual. I havent bothered to measure it but think the resistance curve is pretty much standard, which means that its 8.1K at 30 and 12.1K at 20 and of course 10K at 25.
The problem I see is that all common NTC thermistors appear to have the ratio of resistances vs temp. So a 5K one is 4.1K at 30 and 6.1K at 20 (same ratios). Thus to achieve the same resistance change between (eg) 27 and 20 as a 10K one does between 30 and 20 requires an alternative thermistor with a larger base resistance. Then only way then to bring it passively back into range is to use a parallel, resistance, which reduces the slope back down again.
This seemingly trivial problem isn’t as trivial as it first seems. Digipots are a thing, and of course one could, in principle, wire up a digipot to a microcontroller connected to the existing sensor to fake up the resistance curve, but it seems like a sledgehammer to crack a nut.
Perhaps Im suffering too much from being brought up in the days when analogue was still considered a viable way to build circuits!
I looked briefly through the Amphenol technical guide and I couldn’t see much variation in beta/B value (which is essentially the relative slope), they are all around 3500 to 4000. Maybe Im still missing something.
Im now asking ChatGPT to see what it can come up with. Initial response is to confirm my suspiscions but to offer to search catalogues… after which it comes up with:
Short answer up front: I could not find any off-the-shelf 10 kΩ NTC parts at Mouser / Digi-Key / Farnell with a B (β) ≥ 5 000 K. Distributors and manufacturer datasheets show 10 kΩ parts mostly in the ~3 000–4 300 K range (common values 3435 K, 3950 K, 3977 K, 3988 K, etc.).
Im going to ask about other base values…
Certainly not common, a couple of possible leads, although replacing the thermistor gives another problem - its not obvious how its mounted. That will hopefully turn out to be soluble though. Definitely not simple
