Heat Pump Wireless Room Thermostats – A Cautionary Note

This note is aimed at prospective heat pump owners, and is intended to bring together the relevant information from a number of discussion threads. My apologies to seasoned heat pump owners for the repetition.

You might be considering a wireless room thermostat (aka roomstat) if the wired roomstat installed with your heat pump will not be in the ideal location for temperature control of critical rooms (e.g. living room) and cabling of a wired thermostat is not practical (or if you want to be able to easily relocate the roomstat from time to time).

Wireless roomstats come with different levels of smartness (for example the more sophisticated – thus expensive – ones can automatically vary programme timing and temperature setpoints according to “learned” user preferences, and some even accommodate varying tariffs), but in most cases they apply the same basic control principle – TPI (time proportional/integral).

Unlike electro-mechanical thermostats which have a built-in hysteresis (i.e. the temperature range between “off” and “on”), these control by changing the duration times of “off” and “on”. (In other words, wired thermostats offer amplitude modulation, whereas wireless thermostats offer frequency modulation.)

In wireless roomstats, a cycle is a fixed period of time that includes an “on” period and an “off” period, but where the relative durations of “on” and “off” vary. These proportions are calculated and continuously updated by the thermostat software. It does this by sampling the rate of room temperature change when “off” (due to heat loss from the room) and “on” (due to the surplus of heat from the emitters over room heat loss), and, applying in-built TPI factors for setpoint departure and rate-of-approach, calculates the new ratio of “on” to “off” time needed for the next cycle. A time plot illustrating this is included below.

In any wireless roomstat, the user can typically set the desired cycle rate (may be in factory settings) depending upon the application. For example the Honeywell T3R (a fairly basic wireless roomstat) allows a cycle rate selection of 1,3,6,9 or 12 per hour. Gas boilers would operate without problems at the higher cycle rates, but heat pumps would definitely not (heat pump stop/starts should not exceed 2-3 per hour, to minimise mechanical wear in the compressor).

In addition to cycle rate selection, wireless roomstats may allow the setting of minimum “on” or “off” times to give time for the controlled device to react to changes. Gas boilers will tolerate rapid switching, but heat pumps do not, and any minimum “on” or “off” times available should be maximised.

Wireless roomstats provide exceptionally (arguably unnecessarily) stable room temperatures even at just 1 cycle per hour, but at the cost of inducing more frequent heat pump cycling than a wired thermostat would. (Several wired thermostats offer adjustable hysteresis. This allows users to optimise comfort – small room temperature swings – against compressor longevity – minimised cycling.)

To illustrate the performance of a wireless roomstat and its effect on the heat pump over a 4-hour period, here is a recent plot which is worth a brief study. In this example, the cycle rate set in the roomstat (a Honeywell T3R) is 1 per hour.

On this plot, roomstat “on” and “off” is best indicated by the circulating pump flow (the black line). As you would expect, each cycle starts (about) 1 hour later than the previous one. For the cycle starting at 13:15, the T3R had calculated a mark-space ratio of about 0.6, and so switched “off” about 0.6 x 1 hour = 40 minutes later. For the next cycle, the T3R had recalculated the required mark-space ratio to be about 0.5, so shortened the “on” period appropriately. This same ratio was maintained – approximately – for the third cycle.

Unfortunately the T3R has a display precision of only 0.5degC so it was not possible to detect room temperature changes less than this, but throughout this whole period the T3R was displaying a room temperature of 21.5degC, the same as the setpoint (i.e. the room temperature did not vary by more than 0.25degC).

The conclusion is that cycle rates well below 1/hour would still provide an acceptably constant room temperature, but unfortunately such rates are not widely available in commercial wireless roomstats. If heat pump cycling at this frequency is unacceptable to you, you should avoid them in favour of wired thermostats, preferably ones with an adjustable hysteresis.

A few other points and observations:

  • With a wireless roomstat, the heat pump controller does not have actual room temperature information – only a switched signal (demand or no demand). Most modern heat pumps can accept this, but some older ones may require a modulated signal directly from an instrument.
  • The heat pump controller may be primarily designed for a hysteresis-based (wired) roomstat (which would normally provide a very low cycle frequency due to high room thermal inertia), and contain an algorithm that applies 2- or 3-term control to any switched (“on”/”off”) roomstat signal to manage room temperature overshoot and hysteresis. If the switched signal is more frequent than that designed for, the controller can behave unpredictably (e.g. switch off the heat pump before the roomstat target has been reached). This may be another reason to avoid wireless roomstats (or to ensure that if one is installed its cycle rate can be set to or below 2-3 per hour).
  • On the above plot, it is the roomstat that is switching off the heat pump, not the weather compensator acting to limit LWT – the latter (red line) does not significantly exceed the target (yellow line), and the compressor speed (mauve line) remains above the minimum - 20Hz in this case. (If LWT exceeds target noticeably, the controller normally reduces compressor speed to minimum before switching it off.)
  • If you do use a wireless roomstat to control space heating, you will need to set your heat pump controller to ignore any in-built wired roomstat.
  • (Relevant to my particular arrangement only, and nothing to do with roomstats) Notice how the measured ambient temperature (the brown line) dips slightly every time the heat pump evaporator fan starts. This is due to cold air recirculation flowing over the ambient temperature sensor, and results in a discernable increase in weather compensator target temperature (yellow line). This may be out of your control, but a badly-located instrument can adversely affect the heat pump operation.
3 Likes

Thanks @SarahH

This may well be one of the leading factors that have resulted in poor performance in heat pump installations in the UK to date. Weather compensations set too high coupled with roomstat based on/off switching of the heat pump. An example from the Electrification of Heat dataset of how bad this can be: Analysis of Electrification of Heat trial data — OpenEnergyMonitor 0.0.1 documentation

Recommendations: Make sure to tune the weather compensation first so that your heat pump is running the lowest flow temperature possible to the extent that the room thermostat would not be needed and then adjust any room thermostats more as over temperature limits - ideally use the heat pumps own room influence based control rather than 3rd party thermostats of course..

and for heat pumps where you need to set a minimum flow temperature on the weather comp adjust this carefully to match minimum modulation of the heat pump and emitter spec. (I created a little tool here to work out suitable values Weather Compensation )

1 Like

I’m not sure this is a function of the room stat being wireless, more a function of the overall controller. There is nothing to stop a wireless ‘room’ stat from transmitting temperature information any more than a wired traditional stat that is simply an on/off switch.

Perhaps the learning is that the controller employed (wireless or wired) needs to do more than tell the HP to turn on or off.

Perhaps smart TRVs are more suitable for a HP & radiator installation?

2 Likes

Thanks @TrystanLea and @borpin.

In hindsight I should have added a paragraph or two on OpenTherm. As I’m sure you already know, this standard, fairly common on gas boilers and increasingly so on heat pump controllers, provides for modulating - as opposed to toggled (on/off) - heat source control, including temperature target offset and rate-of-approach management, but (correct me if I’m out of date) still requires a wired connection. This is not always practical (e.g. post-recabling decoration costs), hence the attraction of wireless roomstats. My message was supposed to be “be aware that wireless roomstats may involve some compromise between comfort and energy efficiency” but perhaps I should have added “…at least until modulating wireless roomstats - or smart emitter TRVs - are widely available at a more competetive cost”.

1 Like

Hi @SarahH, great write up.

I will chip in with my own experience of using a wireless room thermostat with my heat pump, if that’s OK.

As @TrystanLea says, under normal operation it’s best to use weather compensation first to control room/house temperature, and to then use any room thermostat to prevent any overheating. This is how we operate for the most part in winter, but we do use it for a set back overnight (off at midnight, and back on at 4am) where turning down the flow temp to minimum still results in too much heat.

However, there are times when a decent 3rd party controller can be beneficial, and we have used ours with ToU tariffs (e.g, Cosy) to set room temps higher or lower in cheap/expensive periods (our heat pump is oversized so for us this is not always possible by adjusting flow temps as we are already running on the lowest flow temp possible). We have also used our room thermostat when we go away during winter and do not want to maintain the house at 20C

We have an EPH CP4 wireless room thermostat:

This wireless thermostat works extremely well with heat pumps. It has 3 modes of operation, Normal, Optimum Start and TPI modes, as discussed by @SarahH above. We use the Normal mode which operates as a simple on/off switch (+/-0.5C adjustments, with a 0.1C display resolution) with an adjustable hysteresis of 0.2C-1.0C (in 0.1C steps).

The programmable timer has 6 slots where the room temp can be set (there is no on/off, so a low temp is off and a high temp is on) allowing the room temp to be adjusted up or down 3 times per day which works perfectly for ToU tariffs such as Cosy where you may want to increase the room temps during the cheap slots and set back in the more expensive slots. It also works well for us where we like 18C overnight, then 19C when we get up rising to 20C in the evening. The timer function has a daily mode or a weekday/weekend mode.

We also use the thermostat when we are away over winter, setting the desired room/house temp to 16C with a 1C hysteresis (on at 15C, off at 16C). When we were away over Christmas, looking at my Powerwall usage data the heat pump ran in cycles for around 45-50mins (to heat 15C to 16C) and was then off for 1h:30min to 1h:45min. Obviously lowering the hysteresis would have shortened the run times and gaps between run times.

So if you want or need the convenience of a wireless programmable room thermostat that works well with heat pumps (i.e, has a ‘normal’ mode of operation, not TPI) and has adjustable hysteresis, then I can recommend the EPH CP4 unit. I’m not sure what the range is, but works fine in our house. If the plant room were further away, outside in the garage, maybe check the range/signal strength is adequate first.

2 Likes

The other factor in all this is your overall system design. If you have a large buffer tank, that will impact how you might control it over the heating circuit being fed directly.

This (time proportional integral control being default rather than hysteresis control) is a relatively recent development.

You get EU points (energy efficiency points) for the thermostat offering more precise room temperature control when used with a “digital” heat source.

The merchants stock products suited to direct electric heating and the artificially dumbed down gas boilers designed especially for the UK market.

The previous standard used to be hysteresis control. That doesn’t get as many energy efficiency points as the time proportional integral control that’s set to the limit of what the heat source can work with in trends of cycles per hour. Hence not stocked widely.

Many are configurable between the two modes (hysteresis Vs time proportional) in software.

2 Likes

A further, very basic point, is considering where you place your wireless thermostat/controller. I have just found that my system has struggled to meet demand when placed in an area of the living room that was subject to draughts from under skirting (me not doing the final seal after installing under suspended floor insulation) and from vent installed to avoid hazardous wood burning stove (rarely used) carbon monoxide levels.

Fixing these two problems and resiting the controller has made a huge difference as system was struggling to overcome cold on near zero days when rest of the house was fine.

As I say, a basic point but heat pumps need to suit non-technical householders and only they (not the engineer) know the foibles of their own home.

1 Like

I wonder how many points I’d get for my UFH, fed from a tank (mixed down - Gas Boiler), controlled by a virtual ‘thermostat’ in HomeAssistant based on the (large) room temperature -0.1°C hysterisis. I also ‘step’ the times and rising target temp to allow for a lower starting point on cold days.

Seems to work well enough.

However, I would like to account for potential sunshine as the Solar Gain can be significant. The temperature of the floor is a factor as well but I don’t have a sensor for that.

Many thanks @Old_Scientist and @Marko_Cosic for your updates – it’s a while since I was last in the market for a roomstat and clearly technology has moved on since.

The EPH CH4 manual referenced by @Old_Scientist reminds me that I omitted one aspect of TPI controllers from my original post – that of control temperature range (EPH call this “bandwidth”).

In my original post, the room temperature was already close to its target temperature at the start of the time plot I used. This did not address the effects of TPI start-up from cold conditions (e.g. after a night-time setback).

This is better illustrated in this recent time plot, which starts a few seconds after setback ended at 06:50:

I had been advised previously by a Resideo (Honeywell UK agent) engineer that for the T3R the first “off” signal should occur “about 1.5degC from setpoint”, so unsurprisingly, the indicated room temperature when the first cycle started at about 10:05 was 20.0degC (1.5degC below target).

The indicated room temperatures (at 0.5degC precision) when the subsequent cycles started were 20.5deg at 11:05, and 21.5degC at 12:05, remaining there for the rest of this plot.

The mark-space ratios for the specific conditions prevailing on this day were thus:

>1,5degC from target always “on”

1.5degC from target 52min in 60min = 0.87

1.0degC from target 45min in 60min = 0.75

0.5degC from target 40min in 60min = 0.66

0.0degC from target 37min in 60min = 0.62

Presumably these ratios would continue to rise above setpoint, so that if the room exceeded 1.5degC above it (e.g. due to solar gain) the demand signal would be contiuously “off”.

So at steady state when the room temperature was constant (12:00 onwards), my emitters – running at an average temperature of about 44degC, thus an emitter-to-room deltaT of about 23degC – were able to supply 1/0.62 = 161% of the room heat loss to an average ambient of about 5.5degC. (I can use this information – along with knowledge of heat pump output from Q=m.Cp.(LWT-RWT) – to estimate the room heat loss at any room temperature and ambient temperature, and the effective thermal inertia of the room.)

1 Like