Matching two strings of Evergreeens

System · September 2007

My 3.8 Kw grittie system has been online for about two months now, and doing pretty good. 30 Evergreen 180w/170w mixed, split into 2 strings, giving me 20+ KwH/day. Lately I decided to fine-tune the balance. Installed an ammeter in each string. Noticed that at lower amps (2 or less) they are dead-on balanced. But when the amperage goes up, an imbalance develops. When string "A" shows 7.5A, string "B" might show 6.2A. I tried taking one panel from the strong side, and adding it to the weak side. Hardly any difference. So... I think I have either a resistive connection, or maybe a "bad" panel (?).

The question is... which one is the prime suspect, and what is the best and safest way to pinpoint this bad actor? I've tried measuring voltage drop across connections while the system is running, but I run the risk of tripping the GDFI fuse (it's expensive, Ive found - hehe). Is there some more creative - or simpler - method of finding what's slowing down my electrons on one string and not the other?

Thanks.

Solar Guppy · September 2007

Re: Matching two strings of Evergreeens

First are the strings each 15 panels? , if not , like 16/14 , the reading you have makes sense

Next, what is the voltage of each string open circuit? pull both hots from the inverter and measure the voltage, they should be with a few volts or so of each other. If not, the string with the lower voltage might have a bad panel.

Next, what is the "mix" ... are the 180's all in one string and the 170's in the other? this could be an issuse as the 180's put out a slighty higher voltage than the 170's and with 15 in a string it adds up. When the inverter tracks the COMBINED vmp, it might favor the 180's and run the array at a higher voltage and the 170's string will run at reduced current. In this case you better off having a 50/50 mix per string.

Is there ANY shading, from a branch or power or phone line when the current difference is measured?, ANY shading will cause the difference like you are measuring.

As for debugging, if its a roof mounted array, it can be next to impossible as you have to isolate panels to find the cause, if possible, swap half of on string with the half of the other and that will point to which 8 of the 30 to start looking at closer.

BTW, 20kWh from a 3.8 array is very good performance

niel · September 2007

Re: Matching two strings of Evergreeens

if it is a case of one string pulling the other down then try one string at a time and see if it is still low. if it stays low the other string isn't pulling it down and there may be another reason for it. it could very well be a small difference such as that a bad connection could cause and need not be several volts difference as it could be in 10ths of a volt and cause a current difference. it could also be a difference in the length of the wire runs and at what gauge wire those runs are using. ultimately you may be looking to fine tune an already fine working system, but if you have the will and patience to do that that is good. if you find something let us know.

System2 · September 2007

Re: Matching two strings of Evergreeens

Neil,
I've isolated the strings, and there is not much change in the differential. Right now, at 1:35 local, I've 7.6A on one string, and 6.6A on the other. I shut one string or the other down, and the needles barely budge.
Solar Guppy,
I started out with 15 on each string, with the 170w panels distributed almost evenly (3x170w)+(12x180w) AND (4x170w)+(11x180w). Thinking it was the lower wattage panels (VERY slight difference) I played the 'shell game', and switched a few around. It made a statistically insignificant difference.
As far as shading, there is some, but only in the morning and evening. This imbalance is most prominent during peak output, when the sun is dead-on, and no shadows. I also have the two strings "checkerboarded" (interwoven), so that any disadvantage from shadows will effect no more than one panel on a string before effecting a panel on the complimentary, or opposing, string.
None of these panels are roof mounted, so I can access the rear of each quite easily. 18 are on a custom-designed and built "shade arbor", and the other 12 are thrown on top of a pre-existing patio cover. All thirty have PLENTY of air on the back side for cooling.
I am thinking that there is either: 1) a single "bad" panel, or 2) a resistive connection somewhere. Currently I am sticking probes in the cables while the system is running, and checking for voltage drop across MC connectors. So far I haven't found anything significant (<.05V). I DID discover a 2.05V drop across a wire nut connetor in the cut-off box, so I soldered that and stuck the nut back on, just for grins. My next step was going to be measure the voltage across each panel while on line, but this is where I start to stumble. I figure if a panel has a restriction (internal impedance above normal), the voltage across it will be HIGHER than the others, given the same current across all panels. Conventional thinking would say a weak panel has "weak", or lower voltage. Anyway, I have measure about 10 panels so far, and come up with variations WAY outside what I would imagine (keeping a keen eye on the ammeter and the sun, making sure the fluctuations are not due to varying INPUT). When i say WAY outside, I'm talking 10.0 volts or more from one to the next... Is there some device that I can plug each panel into, one at a time, that will show voltage (and current) under load? A device that I can control the load, to simulate MPP? This standing under each panel with a voltmeter in one hand, while simultaneously trying to poke two cables with two test leads... there's GOT to be a better way... I will take any and all suggestions.

Thanks, guys.

mike95490 · September 2007

Re: Matching two strings of Evergreeens

Just be careful probing the PV wires with needle probes.

1) you are working with LETHAL voltages - you won't get a "buzz" like you do with AC. you last thought will be "Ouch", as your muscles lock up till you cook.

2) each needle probe ruins the water tight insulation, and will let air and moisture into the wire, where in 10 years or so, they will be crumbly bits of green copper oxide, not copper wire.

3) don't disconnect the MC connectors while they are powered, you will BBQ them with arcing

4) have you TRIED all 180W on 1 string ? What is the difference between the panels watt rating, is the voltage different, or the amperage different ? A lower amp panel, I think will limit the total current in the string, to only the current it can pass. If the difference is voltage, that's better, because it will just add in the string, but the lower voltage string (are the 2 strings paralleled int 1 inverter, or each has it's own inverter ?) can drag the voltage of the higher string down to it's voltage, if they are just paralleled .

System2 · September 2007

Re: Matching two strings of Evergreeens

Mike,
Thanks for you input. The 180w and 170w panels differ in both voltage AND amperage, to the tune of 3.3% amperage, and 2.3% voltage, or 0.23A and 0.6v raw. These are peak power numbers. DIFF(Voc) is 0.6%, or 0.2v, and DIFF(Isc) is 3.3%, or 0.23A. All well within what some manufacturers give for their tolerance ratings of a SINGLE model. So, I am not terribly concerned about mixing the panels. That being said, I have taken pains to mix them based on their specs, not on actual individual-panel measurements. I did check them all prior to installation for Voc uniformity (all +/- 1.0 volt, but not Isc uniformity.
Today, right now, I am currently showing 7.6A (0.65A OVER the peak power amperage) on one string (???), and 6.6A (0.35A UNDER ppa) on the other. It's almost as if one of the ammeters is "off", but this would not explain their being "even" at lower amperages, and "even" at zero (The meters are zero-adjustable mechanical, and they ARE zeroed out).
What would be nice is a way to check them INDIVIDUALLY under peak load condition. Even better, a way to check for internal impedance. But I digress. . .
Maybe I should shut up, and count my blessings. After all, I AM getting over 20KwH (on a good day). It's just this nagging feeling (sorry, Honey) that the system is not working at peak efficiency, and not knowing how to go about fixing it.

System2 · September 2007

Re: Matching two strings of Evergreeens

Mike,
1) It sounds like you've felt the sting of DC before... LOL (I have too~)
2) Ruining the seal... yes, I realize it's not a good thing.
3) disconnecting while under power... yeah, I accidentally did that... ONCE.. frightening, that arcing sound.. reminds me of when the Emperor was frying Luke Skywalker (!) (that particular MC connector was the FIRST connector I checked for "resistive condition"..it was still good... less than 0.02v drop)
4) I've gone through MANY combinations, starting with 12, then 13,14,15, and16 on a single string, as I was building the system, until I got more panels ready to go online. then I split them into two strings of 15. And yes, they are in parallel, feeding one Sunny Boy 3800U GT inverter. That's why I'm so anal about trying to match (and therefore maximize) the current.

Again, thanks for your input, and will take any more suggestions.

Solar Guppy · September 2007

Re: Matching two strings of Evergreeens

I would check the amp meters, sound like your using some mechanical thing, use a dvm and verify, based on what you have written, I bet thats the issue not the panels. Easy test, swap the hot legs and see if the current reading move, if not the meteres and all of this work was for not!

If current difference follows the string, then its simply a divide and conquer approach. Run a single string and swap half the panels. In two tests you know which quarter of the panels has the weak panel ... three more tests you know which one it is.

Roderick · September 2007

Re: Matching two strings of Evergreeens

If you are able to disconnect the strings one at a time, is the *voltage* under load very different? That would confirm the theory of different operating point. But, as you say, there would still be the tedious job of rebalancing which panels to put where, or finding the weak one, if any.

niel · September 2007

Re: Matching two strings of Evergreeens

just get down to it and measure the isc with a dmm on the individual pvs in the suspect string making sure you've disconnected them from any loads of course. if they don't give the proper short circuit current then you have your answer and if they do then you've eliminated the most espensive part from being at fault.

Matching two strings of Evergreeens

Comments

Categories