I’m not familiar with the graph but the yellow highlighted section looks like regular defrost cycles. That’s when the heat from the radiator pipe work briefly runs back through the outdoor unit to melt the ice which condenses on the grill at the back. All normal behaviour.
What do the various coloured areas along the top indicate ? 

At the top is what zone is working. Z1, Z2, Both or Idle

Below is the same graph with operation mode. frost protection is kicking in where the arrows are. (a lot of arrows)

I have never seen this behaviour before. It also looks like the rise in perceived outdoor temp is bringing the flow temp down as I would expect because this directly effects the compensation curve.

So the blue section shows both zones working, then at just gone 7am one of those zones switches off, after which the defrost kicks in on a frequent basis? Any idea why the defrost isn’t shown during the early hours when both zones are on? When is your hot water heated?

I’m just trying to see the whole picture here.
Meanwhile I think you should keep all zones on at all times; that would allow the system to settle into a steady state which helps with efficiency I believe.

Hello ​@razor I’m still an onlooker on heat pumps. Where is the temperature sensor and can it be exposed to accidental heat contamination. It may be my eyesight, but the defrost cycle looks as if it precedes the temperature spike and is not following it.

Can I clarify. I’ve called it the defrost cycle where I should have called it a flow temperature dip, but the issue remains, why is outside temperature fluctuating so wildly.

This is another Ecodan, data from the heatpumpmonitor.org over a few hours with a similarly low temperature. I think it’s safe to say Trystan’s heat pump will be running as well as it should, but unfortunately Razor’s system has several problems which we’ve looked at in a previous thread. However, though these graphs are from different monitoring systems they do both show defrosts.

Where the sensor is placed is a very good question. Trystan’s only shows a very small rise compared to Razor’s. The flow temperatures are different but it’s a surprise to see so many and such large spikes.

As regards the temperature dip preceding the outdoor sensor rise, I think that makes sense - the controller must trigger the defrost and sends the heated flow water in reverse, so the flow temperature drops as it’s sent back into the frosted outdoor unit. The sensor out there receives the heat as the frost melts (delayed something to do with latent heat - long time since I did this at school...) but the sensor picks up the message that the ice has melted and it’s now good to send the flow in the normal direction. Guesswork!!

I have been playing with weather compensation on a gas boiler with internet outside temperature used. Now this is frowned on as not being sufficiently accurate, though it has served my purpose. The alternative sensor on the wall comes with many caveats if it is to be more accurate. To me I suspect a degree of inaccuracy is acceptable as long as the trend is consistent. However, what should not be present is feedback, heat pump affecting sensor affecting heat pump and the spikes here peak at over 10c temperature difference. It might as well use internet temperature.
The outside temperature on Trystan Lea’s Ecodan is better and unlikely to induce feedback, but should temperature error be there at all in any detectable amount? As an outsider I’m impressed by the monitoring available on these systems, I’m less impressed by what seems to be the standard of installation.
Relevant here is how often is the signal is sampled on a heat pump, if only say every 10 minutes or averaged then spikes wouldn’t be noticed by the system though I wouldn’t be able to sleep knowing it was going on.
OK I’ve had my gripe, I’ll give it a rest now.

That doesn’t sound like a gripe to me ! I’ll see if I can find out where the sensor might be on the Ecodan. I’m wondering if Razor’s is not correctly positioned, given the other ‘issues’ with their installation. I have no idea how the info from the sensor is picked up by the controller but it wouldn’t be every 10 minutes, more like seconds! 
Trystan is one of the creators of the heatpumpmonitor.org system and they are also monitoring a gas boiler ! All extremely informative, I use it to see what my hp is doing as the Daikin info is sparse to say the least. 
Sorry to hijack your post Razor, but this might lead to getting some improvement for you.

I have no way of knowing if this is where the outdoor sensor on Razor’s Ecodan is but this is from a Mitsubishi ASHP manual from a possibly older model. Seems too easy…

​@razor is there something like this on the back of your outdoor unit ?