The boiler relay was just wired up as a "zero volt" thermostat to the FTC6 - closed circuit for heat, open circuit for no heat.

With the switch closed, Evohome will just be controlling a few TRVs as limiters, while the Mitsubishi remote is doing the "call for heat" and controlling the heat pump. Basically the same as if I was just using a few wax TRVs, but programmable.

Now that I'm back in pure WC mode, I don't think the wireless remote will be having any influence, apart from "on or off"?

Sounds like moving the wireless thermostat is the correct option for you.  Hope it works out for you.

The Evohome controller starts "modulating" the BDR91 boiler relay as all the TRVs get to within 1 degree of target temperature. Since it can only modulate by switching off, it starts cycling every 20 minutes (the minimum number of cycles, set in the system configuration). 

After a few minutes, the Evohome boiler relay switches on again, the pumps start running, flow & return temps start to drop, and the compressor switches on.

This wasn't happening before, because it was cold outside, and a couple of rooms weren't getting close to target temperature. Evohome was continuously calling for heat.

Main culprit was the bathroom, losing a lot of heat through the extractor fan ducting. The electrician had removed the baffle flaps, because they fouled the flexible ducting. I've refitted the flaps, modified the ducting, and they now work. Much less heat loss, bathroom gets up to temperature, so now the Evohome is cycling.

Not sure what to do about this?

I guess I could disconnect the boiler relay, and just use the Mitsubishi wireless remote as a the controlling thermostat, and the Evohome as adjustable TRVs (without any "call for heat" function)?

I'd have to find a suitable location for the Mitsubushi wireless remote though - somewhere that it actually has some control over the heatpump, but doesn't switch it off too soon.

Graphs from today, 8 degrees OAT, WC curve target is 40 degrees

Graph shows flow temp and power consumption for the complete setup over a couple of 20 minute cycles.

The first part of each cycle makes sense - compressor starts up, flow temp increases. Compressor runs harder, flow temp increases further. Reaches target temp, compressor backs off a bit.

Flow temp stabilises for a bit, then everything shuts off - including both pumps. Flow temp is steady, because the water isn't moving. 

After 2 minutes, the pumps start up, flow temp drops rapidly. 

3 minutes later, the compressor starts up again.

I don't understand why everything switches off when the flow temp is stable?

I don't understand why both pumps switch off the instant that the compressor stops?

Difficult to say as I was still in the process of adjusting my WC settings at the time I made the change.  What I know is that the house temperature is significantly more stable having turned off room influence.  This is likely however to be house specific!

I had a couple of instances when the house started to get too cold - OAT rising slightly, wireless controller bang on the setpoint, ASHP flow temp lowered by 5 or 6 degrees. The thermal lag in the hallway meant that the temperature seen by the wireless controller didn't drop, so it kept lowering the flow temp - ended up going way too low for most of the rooms.

It also made no real difference to the "cycling".

When it was "in sync" with the house, AA mode did seem to be a bit cheaper. Too unpredictable though.

I think AA mode would work if it had multiple sensor inputs, and could take an "average" across the whole house. Or, maybe it'd work if I kept all the internal doors open to equalise the temperatures? Keeping the whole house at 21 degrees might get expensive though

For what its worth I had exactly the same experience with my Vaillant heat pump and have settled on operating in pure WC mode.

One of the challenges for the designers of AA mode (or whatever your heat pump calls it) is that it attempts to respond to changes in IAT.  However any change it makes in response have a delayed effect (because of the thermal mass of the house), leading potentially to instability in the control loop.  This is particularly true with houses that have UFH in a slab, and much less true if you have fancoils throughout.  Thats one reason why weather compensation works so well, the heat pump is responding directly to a leading indicator of the heat loss (the OAT) rather than to a lagging indicator (the IAT) - so there is no loop delay.  

You may or may not be aware that there is a whole branch of engineering called control theory which deals with this sort of (horrible) problem.  Where the characteristics of the system remain constant (other than the one variable for which you are trying to correct for) its possible (in many cases) to design a control system that stabilises well, using three terms (proportional, integral and differential) with appropriate weightings.  When the characteristics of the system (other than the one variable that you are trying to correct for) change in a timescale comparable to the response time of the system, this becomes increasingly difficult.  WC circumvents this problem by responding directly to the main variable that we are trying to correct for.

I had a couple of instances when the house started to get too cold - OAT rising slightly, wireless controller bang on the setpoint, ASHP flow temp lowered by 5 or 6 degrees. The thermal lag in the hallway meant that the temperature seen by the wireless controller didn't drop, so it kept lowering the flow temp - ended up going way too low for most of the rooms.

It also made no real difference to the "cycling".

When it was "in sync" with the house, AA mode did seem to be a bit cheaper. Too unpredictable though.

I think AA mode would work if it had multiple sensor inputs, and could take an "average" across the whole house. Or, maybe it'd work if I kept all the internal doors open to equalise the temperatures? Keeping the whole house at 21 degrees might get expensive though

Wireless remote in the hallway, and the hallway radiator "open".

Heating schedule set to "off" between 20:00 and 23:30.

OAT dipped to 9 degrees, back up at 10 degrees before dawn.

When the heating came on, the flow temp started off 5 degrees above the WC curve. That's fine, it "knew" the house needed more heat to make up for the "off" period. This is what I want it to do - run a bit harder during the 7p hours, after being off for 3.5 hours when it's 27p kWH.

By the time I got up just before 7, all the rooms were within 0.5 degrees of the setpoint, and everything was stable.

The graphs posted show that AA mode is actually working - as long as the target temp is very close to the actual temp, and the OAT is stable. The flow temp is up and down a little, and it looks like it's being controlled by the compressor duty cycle - same 20 minute cycles, but with much longer "off" periods than when in pure WC mode (a fixed 4 minutes)

Interesting little "blips" on the power consumption graph. Looks

like it fires up the pumps after being idle for a few minutes, checks the temps, and then starts the compressor.

I'll leave it alone for a few days, and see how it copes with swings in OAT. I'm not sure if it went crazy yesterday morning because the OAT dropped rapidly, or because I had the target temp set at 20 degrees - a temperature the hallway was realistically never going to reach.

I'm seriously impressed with the cost so far today - 80p and 6.26kWH since midnight, and that includes DHW up to 49 degrees. 

I'll switch back to WC mode, and remove the wireless remote if necessary.

I'm reasonably confident that once I've replaced one slightly undersized radiator, got the hallway radiator replumbed the right way round with new valves, and topped up the insulation over the bathroom and lounge, the rooms will be reasonably close in rate of heatloss.

As it stands, I'm having to run a little hotter to keep the worst rooms up to temperature.