FYI I’m now on 09.02 and integral is working as designed (and reported by @Andre_K). Hope to now see reduced cycling on default settings, as well as have the ability to extend cycles even further by reducing “compr start heat” to -100.
Fair play to Vaillant for resolving this within just 2 days!
Wow, that’s great & really quick. Did they change the board and did you have to pay?
Think i’m finally getting my head around this issue now.
Here’s a trace from my 351.06.07 5kW
I’m not sure Integral dropped to -89, I think cos i’m in Setback… but it’s a good starting point.
11:10 Integral -89
11:32 Integral 0
(yellow line on graph)
Flow Temp versus Target Flow Temp across that 22 mins: 39.
(difference is purple, ie 2.3 at 11:21)
ie, this is the time & DT where flow temp is higher than target flow temp (the yellow shading!)
So the Integral changes twice as quick as it should?
I do see the Integral changing every 30 seconds when I drill into the data points. Which isn’t right? It should be every minute according to the manual.
The energy balance is the integral from the difference between the flow temperature actual value and targe tvalue, which is added up every minute. If the set heat deficit is reached,the heat pump starts.
If the supplied heat volume corresponds to the heat deficit, the heat pump is switched off.
I have also played with “Compr. start heating from” and tuned this down to -100, but my unit ignores this and starts at -60 regardless.
Hard to see on your graph, but yes, those are the two issues. I added a calculated integral on mine which helps visualize it. While the lines offsets won’t match, you can at least compare the gradient which is what you are interested in. (the division step needs to be based on your sample rate. If it’s 15s this should be 0.25, 30s then 0.5 etc.)
Yes, new board (external only). Replaced under warranty.
The integral can and does change more often than once per minute (polling it every 10s over ebus you will see this), it’s the magnitude of the change that 's wrong on the old firmware. If you have a 2°C difference between flow and return temperature, the integral should change by 2°min over the course of 1 minute. The old firmware apparently doubles this, showing a change of 4°min.
In your graph, between 11:25 and 11:30, you have a desired flow temp around 26°C and an actual flow temp of 27.5. Over 5 minutes, the integral should accumulate by 1.5° * 5 min = 7.5°min. Eyeballing your integral, it rises from around -24 to -9, which is a delta of 15°min and hence twice as much as it should be.
On a related note: Your ebus flow&return temperatures seem to make sense and not have a terrible offset like mine. I’ll use that as evidence to hopefully convince Vaillant of a repair.
What offset do you see? I have the following:
Compressor off and 840L/hr flow:
Compressor on and 1200/hr flow
This probably means that Vaillant is slightly under-reporting output and COP in my case, as the 0.3C offset in return temp means lower delta-t vs. heat-meter.
I wonder if there is a way to improve that return temp by adjusting it’s position, using some heat-transfer paste, or insulating around it?
There’s a roughly 1.2 °C more or less constant offset between them, easily to be seen by the remaining delta when the compressor is off. Yello/Blue are Vaillant ebus measurements, Red/Green my own PT1000.
@dfeist @Zarch Thanks for those graphs!! There’s quite a spread there. I wouldn’t mind about an absolute offset but the delta between them when the compressor is off is very telling. I have the suspicion that this might also contribute to my cycling problem since the system always thinks the flow temperature is too high. Let’s see what Vaillant has to say.
If it’s alright with you I will add our three graphs to the Vaillant inbuilt metering vs MID thread for reference as it was discussed there as well.
Great stuff. I think these threads should become mandatory reading for Vaillant’s engineering team ;).
Shame I couldn’t get exactly same scenario to overlay calculation on 
I have confirmed the same Energy Integral bug on the the Heat Pump Controller live monitor.
Target Flow was showing 27.5C
Current Flow Temp was showing 30C
So 2.5C difference.
At the same time the Integral at was -58
I started a stopwatch and timed one minute.
At the end of the minute the Integral reading was -53 (5 difference).
So double the 2.5C difference it should have been.
This backs up the same data i’m seeing being pulled from the ebus.
Which is inline with my post on this deg30sec issue way back, before we knew from the excellent contributions here that the old firmware is buggy. Hope you are under warranty as otherwise Vaillant will quote a large sum to bring your system back to the intended specifications.
Some comparison runs with different firmware versions and “comp start heat from” values. In all cases these are 4hr periods with roughly the same initial flow temperature and target flow temperature.
06.07 Firmware (“comp start heat from”: -60)
Time to first cycle: 28min
Cylces: 4
Cycle length: 30min, 37min, 33min, 33min
Time between cycles: 20min
COP: 5.24
09.02 Firmware (“comp start heat from”: -60)
Time to first cycle: 45min
Cylces: 2 1/3
Cycle length: 1hr, 48min
Time between cycles: 31min, 33min
COP: 5.42
06.07 Firmware (“comp start heat from”: -100)
(This firmware ignores this setting)
09.02 Firmware (“comp start heat from”: -100)
Time to first cycle: 70min
Cylces: 1 1/5
Cycle length: 100min
Time between cycles: 55min
COP: 4.92 (no comparable as temp dropped to 6C)
So is apparenlty this is the heat-curve formula:
Source: Home · cyberthom42/vaillant-arotherm-plus Wiki · GitHub
I’m thinking I should be able to use this (along with ebusd) to implement my own weather compensation for cooling(which Vaillant controller doesn’t do it seems) externally on raspberry pi. 
Would this be the correct cooling formula?
I fitted a slightly different one which fits to <0.1°C error.
Interesting, hadn’t seen yours! Did you have any thoughts about how to adapt it for cooling?
