SMA 4000tl start up woes

[resybaby]

Gents, after a bit of help with my FIT system. Its having wobbles when starting up, throwing an 'Isolation Resistance' error and going into fault.

Been happening for a couple of weeks now, dry weather or wet.
System will not boot up automatically and start in the morning with both strings DC isolators on.
However, if i isolate one of the DC strings (labeled 3), as well as turn the AC feed off and back on, it will then start up ok using DC string (2) on its own.
If i miss out turning the AC feed off/on it wont start either way.

Same trouble this morning so i kept my eye on it for a while, after its initial failure to start. I turned off the DC string (3) only and left the AC feed on. The system wouldnt start as expected. I watched the display panel, and despite the DC(3) string being off it still showed a small voltage of 74volts on the display panel for that isolated DC string? Which i just dont understand, given it was isolated at the DC isolator. The other string (2), was showing 175v.

Having had a good explore as to why this is happening im coming to the conclusion that i have an 'iffy' joint (or panel) on the DC (3) string, and given the voltage im seeing on the invertors display is 74v, the 'iffy' bit is likely between panel 1 & 2 on that string.

Does that sound feasible, or am i barking up totally the wrong tree?

Annoyingly. If i isolate both DC isolators, as well as the AC for a few seconds, then turn the AC back on as well as the DC isolator on string (2) only everything boots up fine. Once running if i then turn the other DC isolator (3) on, it continues to work and doesnt trip anything at all, so it seems the 'isolation resistance' fault that im getting on initial start up 'disappears' once the invertors up and running already via string (2).

its a pita having to do this so looking to get it sorted if any of you guys could give me a pointer or two id appreciate it. Im guessing next step is to get myself a multimeter and test the strings, although ive worked out which is the iffy one.

Two strings of identical panels on each mpp. 4 x 50v VOC so both individually above invertor startup voltage.

[Tinbum]

To find the fault turn all the isolators off and measure the voltage from the + to earth and the - to earth on each string. (panel side of isolator so be very careful as will be live) The value you get gives an idea of where the isolation fault is.

If you have a display for the inverter eg webbox you should be able to see what the isolation resistance value being measured is.

[shirehorse99]

Sorry to hear that pal

Couple of questions

do you have an error code for that error the inverter is showing ?
Has it happened only since new panels were installed?
Do you have access to the mc4 connectors for each string?
Are there more than 1 string on each mppt of the inverter?

[nowty]

The reason it only faults on start up is because the inverter can only measure the resistance before connecting to the grid. Once connected to the grid and operating it measures earth leakage and only trips out if it goes over around 12mA.

Older panels can suffer from this issue through water ingress, mine do and I had squirrels chewing through some of the cable insulation. I have 2 workarounds.

1) Is lower the resistance threshold to the lowest possible allowed in the grid settings but you will need a Grid Guard code from SMA to do this.

2) I have my offending string connected via a heavy duty contactor controlled by a WiFi socket. The inverter starts up with the better string, then about an hour later the offending string connects by timer and all is well.

Sometimes if its pissing down with rain I cannot connect at all even with the better string.
One of the reasons I am going to replace my older panels in the near future.

To find in the string where the fault is you can use this technique which Timbum has already outlined,
https://manuals.sma.de/SBxx-1VL-40/en-US/391466379.html

[Countrypaul]

Nowty, who's the contractor you control with a WiFi switch?

[nowty]

Nowty, who's the contractor you control with a WiFi switch?

At at my Mums ut’ North so struggling with my iPhone.
I’ve edited it now.

[resybaby]

Thank you gents for all your comments and help.
Trip to Screwfix tomorrow for a multimeter to give Tinbums string tests a go. Embarrasing really but i dont possess such a device currently . No Webbox facility either, just the basic invertor.

Shire, yes, error code is 35 isolation resistance. Looked that up and it confirms wiring issue. Worked fine for a while when panels were changed but has got increasingly worse till a daily problem now. Im guessing water/corrosion in a connector. The only original wiring left up on the roof was the lengths from the isolators to the first mc4, everything else is new. Cant realy get to the connectors without removing all the panels and dont fancy that much, they are big and heavy. 4000tl, has two mpp connections (each capable of having two strings). Mine are split, one string to each mpp.

Nowty, have looked up the grid guard thing, few hoops to navigate to get such a code as an end user and id likely screw things up further if i ended up changing settings having got that far. Probably a bit much for novice me i think.
Interested in the contactor/timer approach - not being a sparks ive had a google to understand what one of those are, basicslly an enlarged relay i think? Have found a firm called Toughleads.co.uk that look to be the place to go for such an animal to be created. Effectively wire that in between left switched on DC isolator and invertor, them use timer to swtich it on automatically rather than do it myself.
Would appreciate a piccy of yours if possible, if you get a spare moment when back home, slow learner im afraid and pictures help me get my head around the practicalities and join the dots in my electrically limited mind.

Thanks again all for your help and input so far, no doubt i'll have some other barmy questions to ask before i get it sorted.

[Westie]

Hi

I had this issue on my 4000TL-20.

I also have two strings. IIRC it was in wet /damp weather as we have currently. When the error occurred I isolated each string in turn and found which string had the issue. I spotted a couple of MC4 connections lying on the wet tiles under the panels of that string. I suspected they may be the most likely point of failure , so using a long sticky thing I managed to tuck the mc4's between the back of the panels and mounting rails so they were no longer touching the wet roof. That solved the problem

[resybaby]

Hi

I had this issue on my 4000TL-20.

I also have two strings. IIRC it was in wet /damp weather as we have currently. When the error occurred I isolated each string in turn and found which string had the issue. I spotted a couple of MC4 connections lying on the wet tiles under the panels of that string. I suspected they may be the most likely point of failure , so using a long sticky thing I managed to tuck the mc4's between the back of the panels and mounting rails so they were no longer touching the wet roof. That solved the problem

Thanks Westie, i shall give that a go. Ideal tool easily to hand as well - the big telescopic window cleaning squeegy pole stick thing that i use to clean the seagull mess off with. Got to be worth a shot. Hospital first thing then i'll be out there.
thank you

[resybaby]

Well have been peering under the panels from the driveway with the binos this morning and could clearly see three MC4 connectors dangling down into the channels on the concrete roof tiles, no doubt getting a good soaking every time it rains.
One dangler is right between panel one and two of the offending string.
Clambered up on the roof to try and sort that, only to find because the panels are so huge my normal sized arms arent long enough to reach the MC4's to lift and zip tie them off the roof, access only being from the one short panel end, and very little roof to work from giving its pretty much covered.
So have, as a test, wedged them up with a couple of lengths of timber to see initially if the problem disappears or becomes less frequent.
Fingers crossed.