I haven't had enough experience with Homely to be sure but I wasn't impressed. One reason was that it came with a little button that was the sensor but could not keep in wifi contact with the router that came with it.
I thought it was very basic so after a few days I spoke with the member of staff who was monitoring the installation and his words were that it does not give control it takes control. It took a fair slice of electricity too.
It was the last hope for getting a sensible system and I couldn't afford to let it carry on picking my pocket.
However, claims that my heat pump is good for the environment have proved to be true because I switched the whole thing off last April and lived in a cold house since (small oil fill radiator to stay alive) the glaciers are reforming as we speak.
I am sorry to read that you have switched off your heat pump, I hope that you have adequate anti-freeze in your system. From your comments I must presume that your heat has not been correctly commissioned and optimised?
I have honeywell evohome. You can consume the RF bus's data of that honeywell system into a home-assistant PC via a USB-RF module. The protocols are fully reverse engineered by people far cleverer than me and there's a guy sells the RF modules pre-built. To get it working I wrote no code, just installed other peoples code according to the docs...but yes you do need to be some kind of geek...
That gets you this:
which is an "aggregated heat demand for the house" out of 100. Evohome can natively send this as an opentherm percentage demand signal to a suitable boiler, (rather than the simple basic on/off) . I'm trying to figure out if there'll be a way that I can use this to get better control of a heat pump (there ought to be). I don't have a heat pump yet...
Thanks for your input. I note your last remark, which is presumably why you have more of an on/off pattern. I also note the chart is 0 -100% ie it is a relative not absolute plot. Evohome from what I have read is similar in concept to Homely, but more complicated and more expensive, and still someone has to reverse engineer things to get what is after all our data. This is the modern world, you will be a QR code, and you will be happy (and Honeywell and Homely will be even happier, because they have all your gorgeous data).
I haven't had enough experience with Homely to be sure but I wasn't impressed. One reason was that it came with a little button that was the sensor but could not keep in wifi contact with the router that came with it.
I thought it was very basic so after a few days I spoke with the member of staff who was monitoring the installation and his words were that it does not give control it takes control. It took a fair slice of electricity too.
It was the last hope for getting a sensible system and I couldn't afford to let it carry on picking my pocket.
However, claims that my heat pump is good for the environment have proved to be true because I switched the whole thing off last April and lived in a cold house since (small oil fill radiator to stay alive) the glaciers are reforming as we speak.
Very sorry to hear of your experience. None of what you say bodes well. "It does not give control, it takes control" sounds most alarming, that can lead to all sorts of serious problems, with at the most terrible end things like the Boeing 737 Max software caused crashes. The fact that in your case it increased costs is also of great concern, given these 'smart' systems are supposed to reduce running costs.
I wonder if any one else has any experience with Homely that they can report back on? I recall @hughf was upbeat about Homely, but hadn't actually installed and used it himself. Any updated thoughts, I wonder, given the above?
Midea 14kW (for now...) ASHP heating both building and DHW
I think we can agree that users need a system that, in response to the indoor temperature and their preferences, takes control of the flow temperature instead of simply turning the heat pump on/off. What we also need is that system to be easy to use, and completely jargon free. AKA, load compensation.
People simply need a ‘warmer please’ or ‘colder please’ button, and the smarts does the rest. Exactly the same as a standard thermostat on the wall, turn it up to be warmer, turn it down to be cooler.
This is precisely why 90% of ASHP installs use a standard room thermostat. Despite the fact that a standard on/off thermostat is completely unsuitable, it’s simple for the homeowner to use. Yes, a number of manufacturers controls can do this, but it’s far from simple. Take the LG Therma V controller, it has a ‘warmer please’, ‘colder please’ function, and will support load and weather compensation, but the manual is so cryptic that most people couldn’t get it working.
I still stand by my previous statement that I think the homely concept is precisely what the industry needs to address the above issues but I do think that the hardware needs some work. The main hub/controller needs a better power supply connection (Micro USB is completely unsuitable), and the unit should be made compatible with wired, opentherm, thermostats. Simply to provide a nice, standard, alternative to the current app only control. App only control is a terrible idea - not everyone wants to use an app.
Yes, it takes control of the flow temperature, because that’s precisely what a modulating control is supposed to do.
I also still stand by my statement (as an embedded software engineer by trade) that doing the smarts in the cloud (on AWS in their case) is a better solution that running it on the in-premises microcontroller/hub. The Homely cloud system actually calculates the flow temperatures required to meet your target heating schedule for the next 24Hr block, 24hrs in advance (knowing electricity TOU prices, the weather forcast, and the building response curves) and pushes those to the in-home hub/controller. Without an internet connection, it still works, it just reverts to standard weather compensation (I think, need to clarify that).
If they added support for my specific heat pump and had a physical wall mounted stat/controller/interface, I’d buy one tomorrow. As it stands, I’m going to have to figure out my WC curves/flow temperatures by trial and error.
Off grid on the isle of purbeck
2.4kW solar, 15kWh Seplos Mason, Outback power systems 3kW inverter/charger, solid fuel heating with air/air for shoulder months, 10 acres of heathland/woods.
My wife’s house: 1946 3 bed end of terrace in Somerset, ASHP with rads + UFH, triple glazed, retrofit IWI in troublesome rooms, small rear extension.
As it stands, I’m going to have to figure out my WC curves/flow temperatures by trial and error.
And such is reality. Please don't think I was trying to put you on the spot. It is more that I see you as an unbiased advocate for Homely (or more accurately Homely type systems, but as Homely is pretty much the main one out there, the two are currently effectively synonymous), and I wanted to hear the upbeat side of things. Reading between the lines, it hasn't arrived yet. The idea of load compensation (which I am going to define as adjusting heat output by changing flow temp and/or rate in response to any current difference between room set temp and actual temp, in much the same way that weather compensation adjusts heat output in response to any difference between the current ambient temp, and the lower right hand set point on the weather compensation curve) is simple, straightforward and obvious in theory, but has yet to become a practical reality. Micro USB port for main power supply??? What on earth is wrong with a simple robust standard plug and socket?
I still don't get the obsession with using forecasts and ToU tariffs. Forecasts are just that, and flakey too. The forecast is totally unnecessary when you can measure the current temps. If the room temp is below where it should be given the current weather comp curve setting, then something is cooling the building, but in fact it doesn't actually matter what has caused the cooling, all that matters is the rooms are below temps, and need a boost. Ditto if the building is warmer than expected. I don't have a smart meter, and don't have any plans to get one any time soon despite a constant barrage of pestilential emails from EDF demanding I BOOK MY ESSENTIAL UPGRADE NOW (the emails really are that shouty) so ToU tariffs aren't going to happen here any time soon.
The sort of control system I have in mind is actually very simple. It should both maintain the room temps at the right level and minimise energy use by having the heat pump run as a Steady Eddy rather than a Jumpin' Jack Flash. In practice, there are only three states, and only three responses needed, using very simple feedback:
Steady state: rooms temps correct, building heat loss balanced by heat pump output as determined by weather compensation curve. No action needed.
Room temp(s) are below what they should be, given current ambient and flow temp. It doesn't matter why, they just are. Action: boost ouput a bit.
Room temp(s) are above what they should be, given current ambient and flow temp. It doesn't matter why, they just are. Action: lower ouput a bit.
The first state is 100% weather comp curve controlled, the second and third state are weather comp tweaked by load compensation controlled.
It is a given the system in its entirety is correctly designed and sized, if the heat pump output and/or the emitters are too small, then no amount of compensation will ever work in demanding circumstances.
Such a control system could have manual 'warmer please' and 'cooler please' override buttons, although in practice the room stat/control (temp setting) can take care if that: if I want warmer please, then I just turn up the control by one or two degrees, ditto for cooler please. The buttons could be timed, ie boost/lower output for one hour, if using the using the set controller, you would have to remember to turn it down again once the boost had done its thing. Optionally, you could add in a so called 'smart' function, the system 'learns' the heat characteristics of the building, ie how fast each degree of output boost warms the building, and then adjust the amount of boost to get the warm up to happen in a reasonable time frame, but such an addition is not essential. The other key thing is such a control system works by adjusting level of output gradually over a range, rather than by turning the heat pump on and off, ie it has the heat pump running in such a way as to minimise energy use.
What's more, I can't believe we are that far from such a system, at least on modbus RTU/RS-485 based systems (Midea, Samsung and ? others). All of my reading so far points to the fact the connection is as simple as it can get: two wires. Here's the Midea wired controller end of things:
H2/A+ and H1/B- are standard mobdus RTU/RS-485 terminals, physically they are simple screw down connections in the controller unit. These two terminal are then connected to a standard RS-485 to USB converter (image from first hit in ebay search for "USB To RS485 Converter", cost £5.99):
Note the use of the same A/B terminal labelling: A on the Midea unit connects to A on the converter, ditto for the B terminals. Plug the USB end of the converter into a PC (or a raspberry pi or whatever) and that's it, unless I have got this all completely wrong, you are now connected to your Midea heat pump. The modbus register addresses are listed in the wired controller manual, eg 104(PLC: 40105) Water inlet temperature (read RWT), 105(PLC: 40106) Water outlet temperature (read LWT) and (I think this is right, typos notwithstanding) 2(PLC: 40003) Setting water water (sic) temperature T1S (set LWT). I imagine writing a python or similar program to take the right measurements and then adjust the LWT at regular (maybe six minute) intervals should not be that difficult. The PC/raspberry pi (the former I do know about, the latter I currently have no experience of) would also need to know the room set and actual temp, another very simple wired connection. All very simple, and very robust - at least in theory. No need to bother with all that wifi/cloud nonsense, just ultra simple ultra reliable local wires.
Midea 14kW (for now...) ASHP heating both building and DHW
Thanks for your input. I note your last remark, which is presumably why you have more of an on/off pattern. I also note the chart is 0 -100% ie it is a relative not absolute plot. Evohome from what I have read is similar in concept to Homely, but more complicated and more expensive, and still someone has to reverse engineer things to get what is after all our data. This is the modern world, you will be a QR code, and you will be happy (and Honeywell and Homely will be even happier, because they have all your gorgeous data).
yes, no heat pump for me yet, so the zero's are "boiler off" during the night. evohome is indeed expensive but I'm already invested - sunk cost - so I may as well make use of it. actually all data stays local - one of the reasons a lot of evohomers use it , being averse to being cloud reliant for something as basic as heating (which includes me). it will run at full functionality with no internet connection. all the evohome geeks are on automatedhome if you wanted to know more. most of those experimenting with domiticz platform heat control were on there using the ramses honeywell RF protocols and then it moved to home assistant (better underlying platform) and is a lot easier to setup now. you can build a system now that runs entirely your own control algorithm if you want.
@cathoderay I have that converter for the battery cable I built to read data directly from my Seplos
12kW Midea ASHP - 8.4kw solar - 29kWh batteries
262m2 house in Hampshire
Current weather compensation: 47@-2 and 31@17
My current performance can be found - HERE
Heat pump calculator spreadsheet - HERE
you can build a system now that runs entirely your own control algorithm if you want
That is very encouraging, and thanks for the details on where to look for more info. I also unsurprisingly like the fact it is all local, no wifi/internet/cloud based crap (I have no need to control my UK heat pump while travelling on a Japanese bullet train). Meanwhile, here's a link to a post showing someone has actually done something similar (this is from the Home Assistant forum), even if the details are a bit scanty, it's only reading data, and the wired controller isn't identical (the buttons are the same, but the display is similar but not identical, I think the post author may have just used a stock photo, as the date on the display is rather old, 2015). The thread is about Midea air conditioners, but this post is about connecting to a Midea heat pump:
@batalto - that also sounds encouraging, the actual hardware in actual use, albeit a different scenario. Have you considered testing whether it can also connect to the Midea controller (on a temporary basis, for proof of concept?). I'm always happy for someone else to be the guinea pig! There a number of open source programs that claim to be able to read modbus RTU data, you may already have one if you are already using the converter.
Midea 14kW (for now...) ASHP heating both building and DHW
@cathoderay I don't see why not - Its only a few wires from an ethernet. However Modbus isn't my area of expertise so I wouldn't really be sure of what I was reading out.
12kW Midea ASHP - 8.4kw solar - 29kWh batteries
262m2 house in Hampshire
Current weather compensation: 47@-2 and 31@17
My current performance can be found - HERE
Heat pump calculator spreadsheet - HERE
As it stands, I’m going to have to figure out my WC curves/flow temperatures by trial and error.
And such is reality. Please don't think I was trying to put you on the spot. It is more that I see you as an unbiased advocate for Homely (or more accurately Homely type systems, but as Homely is pretty much the main one out there, the two are currently effectively synonymous), and I wanted to hear the upbeat side of things. Reading between the lines, it hasn't arrived yet. The idea of load compensation (which I am going to define as adjusting heat output by changing flow temp and/or rate in response to any current difference between room set temp and actual temp, in much the same way that weather compensation adjusts heat output in response to any difference between the current ambient temp, and the lower right hand set point on the weather compensation curve) is simple, straightforward and obvious in theory, but has yet to become a practical reality. Micro USB port for main power supply??? What on earth is wrong with a simple robust standard plug and socket?
I still don't get the obsession with using forecasts and ToU tariffs. Forecasts are just that, and flakey too. The forecast is totally unnecessary when you can measure the current temps. If the room temp is below where it should be given the current weather comp curve setting, then something is cooling the building, but in fact it doesn't actually matter what has caused the cooling, all that matters is the rooms are below temps, and need a boost. Ditto if the building is warmer than expected. I don't have a smart meter, and don't have any plans to get one any time soon despite a constant barrage of pestilential emails from EDF demanding I BOOK MY ESSENTIAL UPGRADE NOW (the emails really are that shouty) so ToU tariffs aren't going to happen here any time soon.
The sort of control system I have in mind is actually very simple. It should both maintain the room temps at the right level and minimise energy use by having the heat pump run as a Steady Eddy rather than a Jumpin' Jack Flash. In practice, there are only three states, and only three responses needed, using very simple feedback:
Steady state: rooms temps correct, building heat loss balanced by heat pump output as determined by weather compensation curve. No action needed.
Room temp(s) are below what they should be, given current ambient and flow temp. It doesn't matter why, they just are. Action: boost ouput a bit.
Room temp(s) are above what they should be, given current ambient and flow temp. It doesn't matter why, they just are. Action: lower ouput a bit.
The first state is 100% weather comp curve controlled, the second and third state are weather comp tweaked by load compensation controlled.
It is a given the system in its entirety is correctly designed and sized, if the heat pump output and/or the emitters are too small, then no amount of compensation will ever work in demanding circumstances.
Such a control system could have manual 'warmer please' and 'cooler please' override buttons, although in practice the room stat/control (temp setting) can take care if that: if I want warmer please, then I just turn up the control by one or two degrees, ditto for cooler please. The buttons could be timed, ie boost/lower output for one hour, if using the using the set controller, you would have to remember to turn it down again once the boost had done its thing. Optionally, you could add in a so called 'smart' function, the system 'learns' the heat characteristics of the building, ie how fast each degree of output boost warms the building, and then adjust the amount of boost to get the warm up to happen in a reasonable time frame, but such an addition is not essential. The other key thing is such a control system works by adjusting level of output gradually over a range, rather than by turning the heat pump on and off, ie it has the heat pump running in such a way as to minimise energy use.
What's more, I can't believe we are that far from such a system, at least on modbus RTU/RS-485 based systems (Midea, Samsung and ? others). All of my reading so far points to the fact the connection is as simple as it can get: two wires. Here's the Midea wired controller end of things:
H2/A+ and H1/B- are standard mobdus RTU/RS-485 terminals, physically they are simple screw down connections in the controller unit. These two terminal are then connected to a standard RS-485 to USB converter (image from first hit in ebay search for "USB To RS485 Converter", cost £5.99):
Note the use of the same A/B terminal labelling: A on the Midea unit connects to A on the converter, ditto for the B terminals. Plug the USB end of the converter into a PC (or a raspberry pi or whatever) and that's it, unless I have got this all completely wrong, you are now connected to your Midea heat pump. The modbus register addresses are listed in the wired controller manual, eg 104(PLC: 40105) Water inlet temperature (read RWT), 105(PLC: 40106) Water outlet temperature (read LWT) and (I think this is right, typos notwithstanding) 2(PLC: 40003) Setting water water (sic) temperature T1S (set LWT). I imagine writing a python or similar program to take the right measurements and then adjust the LWT at regular (maybe six minute) intervals should not be that difficult. The PC/raspberry pi (the former I do know about, the latter I currently have no experience of) would also need to know the room set and actual temp, another very simple wired connection. All very simple, and very robust - at least in theory. No need to bother with all that wifi/cloud nonsense, just ultra simple ultra reliable local wires.
It's the ASHP manufacturers that should be supplying all this. There shouldn't be a need for 3rd party apps (and costs thereof), cloud-based services, modbus registers, strawberry pies or anything else 99% of people have no ability nor interest in. A lot of them (the manufacturers) do cover this already; Mitsubishi has auto adaptation and I think Vaillant does something too.
It's the ASHP manufacturers that should be supplying all this. There shouldn't be a need for 3rd party apps (and costs thereof), cloud-based services, modbus registers, strawberry pies or anything else 99% of people have no ability nor interest in.
I couldn't agree more, but we are where we are, strawberry fields forever. Unfortunately, I am one of that pesky minority who can't resist the challenge of blowing obfuscation to smithereens, especially if there is a useful outcome.
Midea 14kW (for now...) ASHP heating both building and DHW
That is very encouraging, and thanks for the details on where to look for more info. I also unsurprisingly like the fact it is all local, no wifi/internet/cloud based crap (I have no need to control my UK heat pump while travelling on a Japanese bullet train). Meanwhile, here's a link to a post showing someone has actually done something similar (this is from the Home Assistant forum), even if the details are a bit scanty, it's only reading data, and the wired controller isn't identical (the buttons are the same, but the display is similar but not identical, I think the post author may have just used a stock photo, as the date on the display is rather old, 2015). The thread is about Midea air conditioners, but this post is about connecting to a Midea heat pump:
that heating control front end is out of the box in home asisstant. I have something very similar (pm me if you want a direct look). the only things I don't have are the data they are showing that has obviously come directly from the heat pump via the bus (LWT, RWT, compressor Amps, etc). My versions of these are a temp sensor on the F and R primary pipes (which is very reliable), and a 30 minute sampler of the smart gas meter (which is not)
for minimal outlay experimental get started: any PC-class device running home assistant, plus an ESP32-RS485 module ( https://esphome.io/components/modbus_controller.html ) for the physical connection to the heat pumps modbus. integration of ESP modules into home assistant is out of the box via the esphome system. wifi between the PC and the ESP32. ESP32's are cheap as chips.
The 'make it slightly warmer' or 'make it slightly cooler' facility that you request is already available on most controllers when operating in weather compensation mode, it is the LWT temperature 'offset' which is often displayed.
Modbus is a data communication standard that has been utilised within industrial control systems for many years. Reading the data is not really the problem, it is knowing what to do with the data once it has been obtained.
Accurately controlling a heat pump to maintain the temperature within one's home, is little different from controlling a power station boiler to meet the energy demand imposed by the turbine and generator, the primary difference is the scale and complexity. Your home heating system may have 10 sensors providing data, whereas the power station boiler may have 1000 or more sensors supplying data.
To accurately control the temperature within a room it only requires two temperature measurements, along with a third input which is the desired temperature setpoint. What appears to be missing with many controllers is the necessary control algorithms, and the ability for them to be fine tuned.
Control algorithms are already resident within heat pump controllers, there is one taking the outside air temperature reading and setting the desired LWT with reference to the weather compensation curve, there is another one varying the compressor speed to achieve the desired LWT, and there could be a further one varying the water pump speed to maintain the desired DeltaT between LWT and RWT. All of these are fairly fast acting systems and will have been optimised by the heat pump manufacturer, since their response is fairly predictable.
The problem for the manufacturer in creating a system that will accurately control the room temperature, is that home's can differ greatly, and hence the system would need to be fine tuned to the individual property. A further problem is the operating conditions will vary quite considerably from Summer through to Winter, so it is not such and easy task, but far from impossible.
If I was designing such a system it would operate in the following manner:-
The primary input would be the actual room temperature, which it is quite easy to obtain using a thermistor or Resistance Temperature Detector (RTD). This reading would be compared against the desired temperature setpoint within the Master controller, and in so doing produce a deviation signal. The Master controller takes the deviation and programmed response times and varies its output.
The output from the Master controller acts as the setpoint for the Slave controller.
The second temperature input would ideally be the temperature of a radiator or UFH pipework, but could also be the LWT measurement, since this is what is being controlled. This temperature reading is compared in the Slave controller with the output from the Master controller which is acting as the Slave controller setpoint. The deviation signal along with the programmed response times is used to vary the compressor speed and hence the LWT.
So how does it work?
If the room temperature is in a steady state at say 20C, then the temperature setpoint is increased from 20C to 21C. The Master controller calculates the deviation and starts to very slowly increase its output signal. The Slave controller receives this increasing signal as an increasing setpoint, so in turn calculates the deviation, and increase its output slightly, which in turn increases the speed of the compressor slightly. The compressor works slightly harder and puts more heat energy into the LWT, which causes the LWT to get slightly warmer. The increase in LWT is measured either within the heat pump or at the heat emitters, and in so doing reduces the deviation within the Slave controller. This process continues until the measured LWT is equal to the setpoint within the Slave controller, the deviation becomes zero and the Slave controller is balanced.
But that is not the end of the story, since the room temperature is still not at the desired 21C. Because a deviation still exist within the Master controller, it will continue to gradually increase its output, which will in turn once more unbalance the Slave controller and slightly increase the LWT. As the actual room temperature gets closer to the desired setpoint, the rate at which the LWT is increased will also reduce, so that when the room temperature is equal to the desired setpoint, both the Master and Slave controllers will have zero deviation and the system is balanced. Obviously in the real World this is a continuous process of unbalance and re-balance, so there will be slight variations in the actual room temperature, but these should be quite small in a correctly optimised system.
As I stated previously, the problem is with the fine tuning of the system to match the response times of different homes. A home with a large thermal mass will heat up much slower than one with a small thermal mass, so the response of the Master controller will also need to be much slower otherwise the room temperature will swing above and below the desired setpoint.
There are self tuning controllers available, which normally require the tuning process to be initiated so that the controllers can calculate the response times of the system.
When I have time I will have a look at the relatively cheap industrial controllers that are now available, which it may be possible to integrate into a heat pump control system.
Edit.
I forgot to mention that this is not just theory but actually works in practice. I have an industrial type controller which controls the indoor temperature by simply measuring the temperature in the hallway and the temperature of the radiator in my 'Man Cave' and achieves a temperature of 20.5C +/- 0.1C for most of the time with a deviation of +/- 0.2C on the odd occasion when the outdoor air temperature changes quite rapidly. It achieves this without the need for any wi-fi, cloud or even outdoor temperature measurement.
