Hi Bjorn, I have checked your previous posts and not found any previous report of these crashes or any requests for assistance to resolve them, if you choose to implement such a âbluntâ workaround without any attempts to help resolve the issue, you must understand why you and now Mike have these problems.
We can only resolve issues when there is good information to locate and correct the cause.
Your list of dates shows 8 crashes over 98 days, thatâs 8 missed opportunities to fix the issue, the â20 daysâ might be a useful piece of info if thatâs the max time a system will run, but it might also be a red herring as your crashes average 12.25 days and have happened at 5 and 8 days, but sharing this info helps identify patterns and helps diagnose the issues. It is in fact far more useful than sharing a method of âsweeping it under the matâ so that many more users fall foul of the issue down the line.
Iâm guessing you are not based in Barrow, Alaska, where the sun comes up in May and doesnât set until August 
Jesting aside, âthe graph is clearly wrongâ is the key bit of info there, it is not believable therefore it is clearly a fault. If it was showing zero production for a couple of days, that is believable, Is the data correct? Has something tripped? Has the inverter packed up? Has a shadow been cast over a small part of each string? Or is it just very overcast?
The way the data is presented isnât prefect, perhaps it could be improved, but showing zero as a fault indication when zero is a valid value doesnât work, Iâm all in favor of a positive fault indication, but as you say thatâs another topic.
Thereâs nothing âfunnyâ about that, the lines are simply drawn between datapoints, if the last datapoint for a 48hr period is 2000 and the first datapoint after that 48hr period is 1000, there will be a line that stretches from 2000 to 1000 over that 48hr period, had it gone offline at night (no pv) and come back on at night, it would indeed show zero for that period. The reason it changes when you zoom is because the data set has excluded the datapoints that the line previously started and ended at, you have a dataset full of ânullâ datapoints, there is nothing to draw a line from or to.
