Your comments about the position of the external sensor rung lots of bells.  My installer had difficulties finding a spot which complied with Vaillant requirements.  I am going to have to re consider my options.  That said, Vaillant must know of the micro climate problem and yet they stipulate mounting on the building.

The micro climate effect is only going to matter if, as in your case, the offset is variable and even then its not going to make a vast difference particularly as you will automatically compensate for the 'average' offset when you adjust the WC curve.  But it definitely would be better to avoid if possible.  Some HPs have the OAT sensor mounted on the casing which of course is then affected by the HP itself, arguably worse.

I cant see how it will do any harm to mount it on a fixed structure away from the building if there happens to be one convenient and the wireless range (if you have the wireless one) is sufficient, or on a North facing wall away from a window so any microclimate effect is minimised.     My installer found a good spot on a projecting bit of the garage roof. 

Your comments about the position of the external sensor rung lots of bells.  My installer had difficulties finding a spot which complied with Vaillant requirements.  I am going to have to re consider my options.  That said, Vaillant must know of the micro climate problem and yet they stipulate mounting on the building.

incidentally I have managed to mentally park the deltaT figure of 2 and turned it into a positive.  I also now know the pump doesn’t modulate to achieve DT of 5 as I had previously believed.  I am going up a steep learning curve.

When the outside temperature drops further and the lounge thermometer will not rise above 18.5C, I adjust the heat curve to 0.65. When it is 0C or below and there are strong winds I raise it to 0.7. My heat curve has been dropped back to 0.6 since mid-January.

I am in the fortunate position of having independent monitoring (using OEM) of the heat pump and all temperatures.

One thing I noticed quite early on was that the outdoor temperatures, as reported by the Vaillant app and measured with their temp sensor, was often over-reading by a couple of degrees relative to the outdoor temp sensor connected to my raspberry Pi. My Vaillant installer had mounted their temp sensor on the house wall close to the front door, under a porch. Open to the elements yes, north facing yes, but possibly still in a local micro-climate influenced by the house. My OEM temp sensor was only a few feet away, but in a shrub away from the house. In colder weather, these two sensors could differ a couple of degrees, but this difference was much less in milder weather. 

I realised that this would result in flow temperatures that were a little too cold when the weather got cold. 

I moved the Vaillant sensor, mounting it on a fence several meters from the house.

This completely removed the discrepancy in the readings, and my house as remained very stable in temperature since, without needing any fiddling with the WC settings.

In my system (Arotherm+ 7kW), the system is actually unable to run with an actual flow temperature less than about 35C.  At it's minimum compressor power, it consumes about 700W of electricity and outputs about 2.5kW. In order for my radiators to dissipate 2.5 kW of power, the water needs to be at about 35C. It simply cannot run at a steady state any cooler than that.

If the target flow temp (from the WC curve) is only say 30C, then the compressor will run at its minimum power, with the flow temp rising to around 35C. After a while, the ARotherm decides that the 'degree minutes' has reached zero, and it just turns off completely (leaving the water pump running). After a while, the degree-minutes has reduced down to -90, and it turns on the compressor again. The cycling continues. This typical of all heat pumps. The important point here is that the average actual flow temperature will be close to the target flow temp, but averaged over the cycle period. There are plenty of times when the flow temp is well below the target flow temp.

You cannot see, from the SensoComfort or the app, whether the compressor is actually running, or whether it is waiting doing nothing. You can only tell that listening to the outdoor unit or delving into the Live Monitor values on the 'Appliance Interface' (usually by your HW cylinder.).

The other point to make, in response to @majordennisbloodnok, is that the Arotherm+ does nothing whatsoever to control flowrate to achieve any specific flow/return delta-T. It runs the pump at a constant flow rate allowing delta-T to change depending on current power output. If the system is designed to have a delta-T of 5 at design conditions (max power), then most of the time delta-T will be well less than that. Mine is usually about 2.

Don't worry about delta-T. It's largely irrelevant.

At tickover, it will drop to 22W.

If its only consuming 22W of electricity, then the compressor is not running at all, just the water pump. This is not 'idling', its basically off.

That would be 3.5kW of electrical power in. The power output into the house will at least double that - often more. A 7kW Arotherm+ will never consume more than 3.6kW of electrical power.  The 7kW refers to the rated output power, not input power.

When the outside temperature rises, the heat pump is very effective in adjusting down its effort, using less energy. At tickover, it will drop to 22W. My old gas boiler only ran at one temperature when it was on.

I had a digital thermometer and used it on all the radiators aiming the dot at top centre of the radiators. This showed that all the radiators were at roughly the same temperature. While the target temperature in the sensoCOMFORT was 19C, it often did not get to that, or so it seemed.

I purchased a cheap digital thermometer and checked the termperatures in all the rooms. This showed a uniform temperature of around 19C, I then sat it on the coffee table in the lounge where it displayed 19C even when the sensoCOMFORT in the hall showed 18.5C or lower. After that I largely ignore the sensoCOMFORT and keep an eye on the thermometer. The heat curve when the outside temperature drops to about 10C is set at 0.6 and this seems to work well. The flow temperatures always seem a bit lower than the curve suggests but the room temperature is stable at 19C. When the outside temperature drops further and the lounge thermometer will not rise above 18.5C, I adjust the heat curve to 0.65. When it is 0C or below and there are strong winds I raise it to 0.7. My heat curve has been dropped back to 0.6 since mid-January.

What I don’t comprehend is why the flow temperatures are nearly always below the temperature expected from the curve.  So in my initial post, flow was 25C when the outside temperature suggested it would be above 30C.  The room temperature hasn’t really changed (19.6 against a set temp of 20C) and the heating on.  The flow T is now 28 but still below the value I expected from the curve - roughly 2hrs later.  It’s this non correlation between the flow temp indicated from the curve (based on ambient temperature) and what the flow is actually indication under “status” which I am trying to comprehend.

To comment on this I would need to know a lot more about your system/house.  Even then 'spot' observations of FT are always near impossible to interpret, plots vs time are needed.  Perhaps your installer enabled 'active' room temp mod mode which does an element of auto-adaption of the WC curve, however thats just one of many possible explanations.

A related factor is that I have been told by Vaillant technical that the system is designed to not overshoot the target temperature and backs off the flow temperature as the target is approached.  

I believe thats the case (assuming that the 'temperature' in question is the flow temperature).  There are however some circumstances when it will overshoot, deliberately so far as I can tell. 

In room temp mod = expanded it shuts off altogether when the measured room temp exceeds set temp plus a small amount (I cant remember how much).  I use expanded mode at the end of the season when solar gain tends to be a dominant factor, and inactive mode for the rest (ie the majority) of the season. 

The exact algorithm of room temp mode = active is not disclosed, which is why its difficult to recommend it.  Basically it does some autoadaption of the WC parameters but within undisclosed limits.  I presume, but dont know, that it works well in at least some houses provided the WC is set manually in the right ballpark and the homeowner doesn't fiddle too much

The openenergymonitor forum has some technical descriptions of the Vaillant modes, created I believe by 'reverse engineering'