Gen 2, 3.6KW, 9.5 Battery. 915 firmware
Glorious sun all day and evening yesterday, but the inverter basically said "enough" at 1600 hours?
String voltage takes a sudden dive, which I guess is inverter throttling as we hit the 3.6KW export limit, but it's like it never recovers.
Battery is full at around 1300 (when export starts), and there's little house load. West facing array, some shading from about 5PM onwards.
Could it be an export limit set which is causing the inverter to throttle back when battery is full?
It may be worth checking f teh grid voltage was very high at the time. Last year when we arere suffering from hogh grid voltage the inverter would shut off the panels sometimes.
killythebid Never above 250V. National Grid changed a tap for us last year as we had that issue.
Very similar trace today, with the string voltage taking a dive after 1600.
It's a west facing array, so in total shade until 10:00, yet the string voltage is a happy 450v despite that.
As the panels warm up, there's a gradual voltage drop (due to temperature?), and then the power and voltage drops at 1600.
At this point there's some light shading from tree branches on some of the panels, and by around 1800 most of the panels are shaded - and the string voltage is then back to normal until sunset.
Is that what you'd expect? String voltage drops when there's partial shading, but recovers when the shading is consistent across all panels?
rjp Very similar trace today, with the string voltage taking a dive after 1600.
You don't have timed export to dump battery to grid at 4pm, or anything? The inverter has limited AC generation, and if it's saturated by battery discharge, there's nowhere for the solar power to go, so it has to be curtailed.
DD There's no timed discharge.
I think I've sussed the problem, and it is shading. At that time the panels get slightly blocked by a large oak on the edge of our garden.
What surprised me is that the effect of shading is only noticeable on bright days with clear sky: If the sky is clear, there's a fairly sharp shadow effect, and the string voltage and power output drops dramatically. But if the sun is behind a cloud at that time, the light is diffuse and the shadow is less distinct - and the panels seem to actually generate MORE power than in bright light.
I've overlaid a sunny day (Sunday, blue) on a more overcast day (Saturday,yellow) showing PV power and string voltage. After 5pm the panels generate more power on the overcast day! 
GivEnergy support don't see any issue with the inverter.
When I was going through the original quote process I also had a large tree that would shade some of the roof. Initially they limited the quote to 12 panels. I said 17 would fit and they pointed out the shading problem. The solution was to add optimisers to each panel. With 17 panels and optimisers I can see which panels are producing at any one time and can see the effects of the shadow passing over some of the array (even the chimney shadow shows up). But as each panel isn't limited to the power that the worst shaded one can generate I get good generation despite the tree shadow.
I am also quite surprised at how much is generated in diffuse light. We have SW facing panels and early morning when cloudy and the sun is in the east, I get quite good generation. It surpasses my house load at 6.30am and 1KW at 8.15am despite the sun being reflected onto the panels at those times.
I don't know if optimisers can be easily fitted post install?
Bottom right panel is currently shaded by the tree. Bottom left panel is partially shaded by the chimney.
rjp We do get see shading on clear sunny days in the morning as a chimney stack creates a hard shadow on the panel array in the mornings but clears away by midday. Like you we see much better generation if it is a bright but cloudy day when there is hardly any harsh shadow in the morning compared to a cloudless sky.
Windy Miller I don't know if optimisers can be easily fitted post install?
I can't see why you couldn't install TIGO's post install - would need to get up onto the roof though and lift all the panels etc?
I would suggest it’s worth investigating the MPPT on the inverter and confirming it will adapt regular so you always get max power. Some are batter than others and poorer ones can get “stuck”.
It does seem like optimisers could help, but these can have some downsides too, albeit less that what you’re seeing.
If it is bright but cloudy the panels probably run cooler so can generate more. We have only had one cloudless day this summer and the peak was not spectacular. We get higher peaks for short periods in between clouds.