This is a topic of some interest to me and first of all I'll say that the accuracy of the HR92 is
much better than any mechanical wax pellet TRV operating on the same radiator in the same room. Period. Part of this is measurement accuracy, part of it is the ability to more precisely and intelligently control the valve pin position on the radiator - an expanding wax pellet in a conventional TRV is pretty "dumb", and in fact is downright crude.
The HR92 uses a small geared down motor to move the pin and can make very small precise adjustments to the flow and adapt to the characteristics of the radiator and room thanks to clever heuristics. For those of a technical bent it is a full "PID" controller.
The temperature sensor itself is very accurate - there are ways of getting the true reading its sending to a couple of decimal places (the reading displayed on the evotouch is rounded to 0.5 degrees and also biased towards the set point) and it agrees with another thermometer I have to within a small fraction of a degree when the radiator is off and the other thermometer is placed beside it.
You can see where I'm going with this though can't you - if placed beside it with the radiator off... Like any other TRV it is influenced by radiator heat and generally this makes it read a bit higher than the room when the radiator is hot, I find about 1 degree is typical. This means it would regulate the room about 1 degree lower than what you've asked for when the radiator is running. This can be compensated for to some degree with the calibrate function which can apply a fixed offset from -3 to +3. In this case you would use -1. However when the radiator is cold it reads closer to room temperature and may actually read slightly below the true room temperature due to close proximity to the floor as its generally colder at the floor than higher up when there's no convection in the room, especially if you have a crawl space under your floor with no insulation as we do.
To get a good reading from the HR92's built in sensor (or any TRV for that matter) you really need the following conditions:
1) Good strong convection from the radiator. This means plenty of air circulation around the room due to heat rising at the radiator, travelling along the ceiling and looping back down the far wall and across the floor. This convection pulls the cool return air across the floor past the HR92 and gives it a lot more accurate impression of what the room temperature is and largely offsets the direct heating from the side of the radiator. (Especially if you have modern side panels on your radiator, which work as a heat shield to block the direct radiant heat to the HR92)
Anything that reduces the effectiveness of that convection will tend to cause the HR92 to read too hot due to direct radiator heat which will consequently cause the room temperature to be regulated too low. That can include:
* Old fashioned radiators without convection fins (we have a few of those) which don't convect nearly as much and rely more on direct radiation than modern convection panels. They still work as there is still some convection but not as well.
* Curtains over the top part of the radiators or covering the HR92. (big no no)
* Clothes/towels etc over the top of the radiator blocking the convection vents.
* Decorative radiator boxing.
* Lots of furniture in the immediate vicinity of the HR92 preventing a good flow of air across the floor to the HR92 end of the radiator - the floor within a metre of so of the HR92 should really be clear of obstacles.
* Odd shaped rooms that are not typical shoe box shape could cause peculiar convection patterns that won't work as well.
* Radiator boxed in between kitchen cabinets under a work top. (we have this problem)
2) No window open above the radiator. This is mainly an issue with bedrooms. If you schedule your bedroom for say 16 degrees at night but sleep with the window above the radiator open a bit, cold air will tend to come in through the window and pass the HR92 as it pools onto the floor, causing it to read lower than actual room temperature and bring the heating on when the bulk of the room isn't actually below 16 degrees. This can cause the room to heat up quite a bit past your target as the radiator will come on until the radiator gets hot enough to reverse the air flow so that hot air is rising instead of cold air falling, then it will start reading the room temperature again. Unfortunately this will typically cause the radiator to cycle on and off as the air flow reverses.
The good news is that unlike a manual TRV where you're just stuck with poor results in certain room situation you always have the option to install a remote wall mounted sensor with Evohome, and the position of the sensor can be chosen to optimise performance. (There are a certain set of rules you should follow for locating a wall sensor for best results)
In our house I use the Evotouch wall mounted in the hallway as the sensor for the hallway - not that I really need to, the hallway works pretty well on its own with the HR92 as sensor, but I figure if the Evotouch is spending 99% of the time left on the wall mount I might as well make use of it!
The master bedroom has a DTS92 on the wall above the head of the bed, and the radiator is below a loft window on another wall. Without the remote sensor in the bedroom we had the issue that sleeping with the window ajar in cold weather caused the radiator to come on when it shouldn't - we have the night time set point set to 16 as we have a 6 month old sleeping in the room with us, but using the HR92's own sensor it would come on well before it dropped to 16 degrees and heat the room up above the set point, sometimes uncomfortably so by a couple of degrees or more due to that cold window air fooling the sensor. Conversely if the window was closed and the door was open (or the radiator partially covered in clothes) the room would actually be colder than the set point by a degree or two.
The DTS92 solved all that - it now doesn't matter whether window or door are slightly open or radiator partially covered - the room temperature near the head of the bed is kept very close to the desired set point with no unnecessary activation of the radiator. Not only is it away from the window it is up away from the cold air at the floor as well and gives a much more representative measurement of the temperature we feel in the room in bed.
The living room I used with only the HR92 for about a year then added a DTS92. It worked quite well with the HR92 by itself and I was fairly happy, but I did find that I had to turn the set point up a bit manually when the weather got really cold or the door was left open a bit, and also that in mild weather it would come on a little bit when not really needed. After trying three different locations in the room for the DTS92 I've found one where the reading corresponds very closely with perceived comfort - now the same set point of 21 feels pretty much spot on regardless of weather conditions - if the weather is mild it doesn't come on unnecessarily and if we go from cool weather to very cold weather it compensates so perfectly that we are not even aware that its extra cold outside as we don't notice any change in the room that needs a manual adjustment.
So whilst I was quite happy with the HR92 on its own (it was a massive improvement on the constant too hot/too cold cycling of the manual TRV) the DTS92 gave that extra bit more precision and ability to adapt to widely varying weather conditions without needing manual adjustments.
I figure that the living room and bedroom are the two rooms you spend most of your time in and deserve special treatment. I rely on the HR92 sensor in all other rooms aside from the hallway where I use the evotouch sensor "for free". The only room that gives us a bit of "trouble" is the kitchen where the radiator is tightly boxed in between two kitchen cabinets with a work top over it as well, therefore the HR92 is crammed right into the resulting corner - a very bad location to measure the room temperature.
I can see on my temperature graphs that no matter how hard the HR92 tries to adapt, whenever that radiator comes on it always overshoots 2-3 degrees in a big spike and then drops down again - the room itself probably isn't overshooting but the small boxed in area under the work top that the radiator is located in will be. Also I notice that cooking heat from the oven and stove top doesn't tend to be picked up by the HR92 where it is located, meaning it will keep trying to heat the radiator even though the rest of the room is now too hot from the oven/stove. (As measured by a wall mounted CO alarm that doubles as a temperature sensor)
I know if I installed a DTS92 on a suitable location on the kitchen wall that both problems would be solved, but I'm not sure the expense is worth it when the precise temperature in the kitchen just isn't that important. I schedule it to 20 degrees in the morning for breakfast time so its nice and warm then, but for evening cooking time I only schedule it to 18 so that the room is cool but not cold, then the oven will warm the room the rest of the way to 20-22 and the radiator will go off. So one day I will fit a DTS92 but it is not a priority at all.
Likewise the bathroom I don't really care what the precise temperature is - it gets scheduled to 16 during the day when its not in use so that it's cool but not freezing, and 22 in the morning for showers and that's about it.
So I would say start with just HR92's, see how things go and if you have any problematic rooms you can always add remote wall sensors if required. An exception to this would be if you have curtains over an HR92 I would put a remote sensor in that room right away. In an ideal world every room would have a remote sensor but the cost really isn't justified when not all rooms require super precise temperature control.
. For this zone, I will need a separate temperature sensor to report back to the Evohome.