Here’s an example of two runs, without much change in outdoor or indoor temperature, the first one is pretty good but the second one is dismal with quite a few drops in output and what looks like unnecessarily high ramping up during the run.
Yes, seen this sort of thing before. What size is this model? Nicely configured graphs. All going nicely for 1st period. Then it goes into oscillation/hunting where it is slow to respond, then overreacts. My guess is that it’s the discharge sensor and EEV control that needs tweaking. I have failed to get a dialogue with Mitsubishi about it. Hard to believe they have not noticed it
One of my reasons for using HomeAssistant for my thermostats, was the ability to set a step of 0.1°C on my UFH. I simply found steps of 0.5° (effectively a 1°C tolerance) too great for my comfort level. Using HA I have steps of 0.1°C which is much better.
I was disappointed to find that Mitsubishi’s controls will only allow 0.5 C increments and readings, as my previous homemade integration was controlling the temperature within ±0.05 C. However, it seems to be [mostly] okay for comfort.
Mostly seen it on the 6, which is almost identical to the 8.5 Only seen it on the Ecodan. When running well, COP is excellent. maybe its a bit like a performance car… great when going well, but coughts and splutters when out of its comfort zone… thing is, you should be in the comfort zone
Will be interesting to see if it changes over time as the AA adapts and as outdoor conditions change.
Also I have a large radiator still to install and might get some primary pipework consolidated and made larger so will see if it still does it after or if I just happen to be in some set of conditions that makes it happen.
What hardware are you using to control your pump(s) @dave_walker ? I’m considering replacing my Grundfos UPS2 with MAGNA3’s and am holding off until I’ve figured out the control method as my Ecodan is dumb from a pump speed perspective. Interested to hear how you have approached and your experience with implementation if you are happy to share. My system is flow limited to around 16-18l/min with the existing pumps and I’m planning to increase pump head capacity plus reduce resistance with enlarged flow/supply pipework (but thats a bigger job).
I use a Raspberry Pi with node red using the Melcloud node to determine the Ecodan operating state- heating, hot water or Antifreeze. I also use node red to extract the flow temperature from EmonCMS. I then send this data to a Siemens S7-1200 PLC (Full disclosure I work for Siemens as a control systems programmer). The S7-1200 runs the control. I use the S7-1200 as it is very cost effective, less than £200, has built in PWM, is designed to run 24/7 at between -25 and +60 degC and has an inbuilt web server to allow setup and adjustment. Pump is a Grundfos
I did this with the full support of my installer so it has not affected my warranty. In fact it is working that well that I am currently setting up a second set of equipment for him as he runs a Ecodan and says it looks bad for a customer system to be out performing his.
I am happy to supply code and wiring if you want it. Anything to help team heatpump.
I am currently working on an expansion so I can take control of the Ecodan flow temperature as I am not impressed with how Mitsubishi have programmed the weather compensation or autoadapt. This should allow me to tune the system to my house and allow modes such as holiday/house full of guest etc
Thanks for the tip re Siemens PLC. I have a use for one and will look it up. Re your dt. have you thought of varying the dt for DHW? though maybe a small affect, but with a stratefied tank, arguably the dt should be greater than 5… at least at the start. Also for heating, arguably the dt could be partly proportional to difference between mean radiator temp and room temp. advantage probably very marginal though
I’ve since realised that auto-adapt doesn’t pay any attention to the weather compensation curve at all, and lets the flow temperatures rise to a stable temperature. So, this is great for accurately sized heat pumps that can hum along nicely, matching the heat loss, but not great when you need to cycle to keep the flow temps down.
I was coming to a different conclusion with our system where there currently is inadequate radiator output (mixture of old rads with low output and rads in well insulated bedrooms being closed down by TRVs to stop them getting to warm).
My COPs appear better under AA than pure WC, I think because it compensates for the rads by taking the temperature higher so their output is increased by a larger radiator-room dT. Also by putting a bit more heat into the house at the end of a run it can keep a longer gap between runs. And runs are longer as it allows for more overshoot and for flow to drop further before it starts the next one.
I suppose it can do this by balancing its behaviour around the actual effect on room temperature.
I also suspect that if we had the right rad output pure WC might work better (or there would at least be a smaller difference between AA and WC) and maybe in winter when all the rads are in play and heat losses are higher so there is no issue with minimum modulation if the HP.
Just to clarify, I agree that it doesn’t look like it pays any attention to the set weather compensation curve but takes outside temperature as one of its variables. Read somewhere that it is learning how these room temperature reacts at different temperatures.
Yes, Mitsi were not the first to do this. i think Hitachi and certainly Neura. Always liked Auto Adapt. I was surprised that Samsung (seemingly) take no account of room temperature when using their room control.
The HP frequency is 26 fixed … but the power consumption is not stable while all the other parameters (apart outside temperature … with a change of 1 degree). Any clue of such instability on the power consumption ?