Coulomb's panel repair bench

[coulomb]

I felt the urge today to snap a picture at the Coulomb Castle, of the bench used for solar panel repair:



[ Right click and View Image for more detail ]

I picked up ten 8 year old Sunpower 200 W all-black panels that had been installed without ventilation under them, and had consequently overheated. They also did not have the required 100k resistor to earth, so they are also affected by Potential Induced Degradation, which I'm hoping is reversible in this case. I got them at about 25c per watt. It looks like it will be a reasonable deal, but only if I value my time very lowly (and also my wife's time; she expressed and interest and has been helping a lot.)

Visible on the bench:

* Heat gun, set to 125°C, for softening the backing, black potting, and EVA (Ethyl Vinyl Acetate, the stuff that goes between the cells and between the cells and the glass).
* Temperature controlled soldering iron, with a "battle axe" shaped tip.
* Many repairs, with duct tape covering silicone "pods". They have a letter and digit code indicating their position. Three repairs have been marked up but not yet started (B2/C2/D2).
* Magnifying lamp with 60 LEDs, used mainly as a light source. Occasionally I'll actually look through it.
* Power supply and multimeter, used to check for shorts after a repair, and to find joints that are sensitive to pressure. The multimeter probes are pushed into the junction box at the back of the panel (at right of the photo). This junction box is very useful since it allows access to the joins between sets of 24 cells (these are 72 cell panels).
* Silver and gold felt tip pens; these are used to make visible lines and markings on the black backing and duct tape.
* A blanket covering the table, to minimise scratches. This may later be used as a wet blanket in PID reduction experiments.
* Flat blade jeweller's screw driver, used for removing the backing at the start of a repair, and scratching a small area to solder for the repair itself.
* A chopstick (near the F1 repair), used for pushing softened EVA and black potting away from the back of cells. I much prefer the ones (I think from Noodle Box) that have a triangular end. The intention is for expert chopstick users to be able to control the sticks better; it turns out to be a wonderful wedge shape for scraping without damaging. (Though it's still possible to crack the edges of cells.)
* Some rolls of duct tape. As you know, this stuff holds the universe together.
* Plastic "stanley" knife, for cutting the black backing, and also some black plastic under the straps that join cells; with my repair technique, some of this plastic that sticks out has to be trimmed away.
* Small needle nose pliers which close very well. These are used to prise the black backing off.
* Markings on the back of the panel for cell borders, and three repairs marked up ready to cut the backing material.

Not visible in this photo:
* Silicone caulking gun. Silicone replaces the EVA and waterproofs the repair. The duct tape basically holds the silicone in position, and will no doubt come off in time.
* A dressmaking pin held in a clip lead. One of the tests after a repair requires a connection to a strip of copper that is only about 1 mm wide; the pin helps make a precise connection, and also helps penetrate any oxide on the copper, and/or vestiges of EVA.

The original thought is that each panel may need up to six or so repairs. Unfortunately, there are very few panels with less than 10 repairs, and most of those will need many more. Several panels have more than half the junctions repaired; several "thirds" (sets of 24 cells terminated in the junction box) needed all 22 joins repaired. (The outside connections seem to be much more reliable; I have only 3 or 4 connections to "row L" (right at the bottom of the panels), and none so far to "row @" (at the very top of the panels).

Thanks are due to Weber for getting me started with the repairs, and the idea to attempt them in the first place.

[ Edit: Sorry, there is very little EV content here, except that perhaps one day these panels will charge an EV. I post this here because I know that many EVers are interested in solar power. ]

[acmotor]

.......
I picked up ten 8 year old Sunpower 200 W all-black panels that had been installed without ventilation under them, and had consequently overheated. They also did not have the required 100k resistor to earth, so they are also affected by Potential Induced Degradation, which I'm hoping is reversible in this case....

Can you give us some more detail on the PID. Are you saying electrolytic corrosion ?
Is this from DC potential between PV elements and frame ?
Does it involve moisture ?
Where is the 100k usually installed ?
Why do you think damage is reversable ?
So many questions..... It is interesting.

[coulomb]

I forgot to mention why I'm doing these repairs. It's because thermal cycling, exacerbated by the lack of ventilation, has induced cracks in the joins from cell to cell. Uniquely to Sunpower (as far as I know, as of mid 2013), the cells are "all back contact", meaning that there is never a need to get to the front of the cell (which is nigh impossible, without a factory to put it back together again). So I bridge from two cells to the connecting strap with solder.

The ones that need repairing are sometimes obvious, causing a bulging under the black backing. Others can be found by pushing power into the cells with a power supply at 2-3 amps, and looking for hot spots with a non-contact thermometer.

Can you give us some more detail on the PID. Are you saying electrolytic corrosion ?

No, not corrosion.

Is this from DC potential between PV elements and frame ?

Yes. It's been said that ions get trapped under the glass, forming a sort of giant gate, turning the whole panel into a sort of giant MOSFET, and the potential turns off the transistor. I don't pretend to understand it in detail.

Does it involve moisture ?

I don't think so, though perhaps moisture migrates through the glass, or at least ions do, and these ions need water to get there.

Where is the 100k usually installed ?

Oops! I forgot to mention this important fact. With my Sunpower panels, which apparently have P-type cells, you connect the positive end of the string to earth (mains earth will do) via a ~ 7 W 100 k resistor (for up to about 450 V strings). It's conveniently done in the inverter box where the DC and AC cables terminate.

Note that most other manufacturer's cells are apparently N-type, and for them you need to connect the negative end to ground via a similar resistor. Check with your manufacturer's documentation before doing this!

Why do you think damage is reversable ?

Sunpower more or less discovered this sub-type of PID; it's reported in the literature. They have done extensive testing, and found it to be reversible. Of course, they have a commercial interest to find this, but no-one seems to be challenging this, and there are plenty of examples in the field. For example, Weber found his solar output dropping off, found the 100k resistor solution on the web, installed it, and his Sunpower arrays made a complete recovery in a few days.

This paper confirms that Sunpower PID is reversible: http://www.solon.com/export/sites/defau ... al_PID.pdf

So many questions..... It is interesting.

Yes, this one was not predicted by the manufacturers, and the solution was found after many installations started showing reduced output. They were very fortunate to find such an easy and inexpensive solution. You have to wonder what other problems will be found, especially with the cheaper panels. I worry about the replacement of EVA with cheaper plastics; there are too many panels out there that are turning yellow or even brown after only a few years. Some panels are exhibiting burn spots; the experts haven't figured that one out yet.

Interested readers can find more information on the Green Tech forum of whirlpool:
http://forums.whirlpool.net.au/forum/143

I can't quickly find the post about the burn spots. It might be a few pages back in the 'Dodgy solar installs "don't do" Pt 2' thread (note this is part two and is already at 71 pages; there are a lot of dodgy solar installs out there). [ Edit: fount it here .]

More reading: http://www.pi-berlin.com/images/pdf/pub ... atings.pdf . (Thanks to Weber for the links.) It states that modules that have not degraded to less than 15% of normal output can be recovered to 96% of normal output.

[coulomb]

It is interesting.

Now you've done it! For that, you get subjected to more pictures

This is how the repairs start:


As noted earlier, I mark with a silver pen now. Once started, pliers and the heat gun applied to the backing removes the backing cleanly in about half a minute. Heating the "hinge" and creasing it keeps it out of the way.

These are the measurements I use now:


When the EVA and black potting (it's quite thin and soft) are pushed away (I push away much less these days; these are early photos), I see this:



The green areas are roughly where I solder, though I now remove the black plastic that extends from under the strap, keeping away from the short circuit hazards (marked in red).

The white dots in the right area of the photo are bumps in the glass. I think that the cells are supposed to slide over these bumps; they force a certain minimum thickness of EVA between the front of the cells and the back of the glass. I'm told this is necessary because of the large difference in coefficient of expansion between the cells and the glass. That would also be the reason for the split at the end of the strap (so the strap can cope with the gap between cells changing as the temperature changes).

[acmotor]

OK, so if PID is from potential between PV elements and frames, how does the 100k at the GCI address this potential ? Are you assuming a grounded metal roof or mounting bars that are deliberately electrically connected and also connected to electrical earth (as they should be under ASxxxx for solar installations ?) What of the anodised frames ? What of transformerless CGI's vs transformer isolated GCI's ? Should GCI's internally be controlling the float volatge of PV's away from earth potential ?
Thanks for bringing this topic to my attention.

So in your repair (you are using environmentally friendly green solder) to repair the electrical connection between bus bars and PV elements, yes ? You expect that better thermal management will control differential expansions and not fatigue this connection in the future ?
Are you just bulk soldering or soldering on a wire capable of some flexure ?
(clear openning there for more pics !!!) and well done on the component level repair. The electrons that flow will be tagged with an extra level of satisfaction !

[coulomb]

OK, so if PID is from potential between PV elements and frames, how does the 100k at the GCI address this potential ?

I believe that at the nano-amps required, panel (module) frames "find" an adequate earth.

Are you assuming a grounded metal roof or mounting bars that are deliberately electrically connected ...

Deliberate earthing doesn't seem to be needed. Anodising would not matter.

What of transformerless CGI's vs transformer isolated GCI's ?

Ah, I assume that PID susceptible panels like these should not be used with transformerless inverters. It's a bit of a bummer for anyone who used transformerless inverters with susceptible panels before this was discovered.

Should GCI's internally be controlling the float voltage of PV's away from earth potential ?

Well, with transformerless, there is little choice    But I'd say no, because the inverter doesn't know if it will be connected to P-type or N-type cells. Also, many modules are being certified by Fraunhofer (they seem to be deep into this) to be "PID immune".

... to repair the electrical connection between bus bars and PV elements, yes ?

I wouldn't call these bus bars; they merely connect one cell to the next. Mostly, these aren't even the long strips of metal that run to the junction box, or even from one row to the next. (Though the rarely repaired "row L" does connect from one row to the next.)

You expect that better thermal management will control differential expansions and not fatigue this connection in the future ?

Yes, the soldering seems to be way more robust than the factory connections, which seem to be some sort of press fit or something. Of course, I could be completely ruining the cells with the heat of soldering; I can get 140 W from some panels after repair, and I hope for more following depolarisation.

Are you just bulk soldering or soldering on a wire capable of some flexure ?

Bulk soldering.

(clear opening there for more pics !!!)

I just ducked out to the factory floor; my wife hasn't got any ready just yet. Soon perhaps.

[coulomb]

Are you just bulk soldering or soldering on a wire capable of some flexure ?
(clear opening there for more pics !!!)

Here is a completed repair, before re-sealing:


You can see the two "tusks" of solder clearly. I usually make mine a little shorter than that. The solder connects the end of the tinned copper strap to the top surface of the cell. This seems to be copper for Sunpower cells; for other cells, this may be aluminium, which would obviously not solder.

The "tiger stripes" [ edit: on the solder tusks ] are some sort of reflection of the cell P- and N- stripes (whatever they are); the solder is actually mirror shiny and smooth.

In this photo, you can see inside the red oval where the strap has missed the pad it is intended to connect to entirely. I don't know whether this is a manufacturing defect (the panel can still work using the other two redundant connections per cell), or if this has happened as a result of excess heating.

Note how the stripes on the left are much thinner than the stripes on the right. This is supposedly due to the difference in mobility of holes and electrons. I know it sure makes it hard to test for shorts on the left hand side (in this photo). The hole in the backing sheet is 40 mm x 40 mm.

[acmotor]

Nice job, I think. Initial visial reaction is frightening.
Does the solder flux require removal ?
Traditional 60/40 solder ?
More pics of the seal up required. This is a bit like open heart surgery !

[coulomb]

Does the solder flux require removal ?

No.

Traditional 60/40 solder ?

Yes. I'm afraid it isn't even lead-free. But that would have required a slightly higher temperature, and I'm scared enough of the possible effects of the high temperature already.

[coulomb]

I thought I'd update with my latest techniques before I forget them and they are lost.

I used to use a "template" to mark the approximate cell boundaries on the back of the panel. Now I've decided that this is too prone to error, and I prefer to measure from the edges of the panel instead now. This allows me to use a 30 x 30 mm area that is peeled back for the repair. I can use a single piece of 50 mm duct tape instead of two pieces, saving time. It also saves a bit of tape and silicone. I've used at least 8 300g tubes of silicone, and 8 rolls of duct tape. I prefer the softer tape, as it seems to mould the silicone into a more pleasing shape. I use black tape now to better match the backing material, though it's possibly between grey and black.

A big revelation came a month or so so when I found I was doing more repairs than I needed to do. Once I decided a joint needed repair, I did a complete repair on it, never checking that it actually needed it. I found that some intermittent faults would be sensitive to pressure from several cells away, so those far away cells would look like they were intermittent, when in many cases they were not.

So now I start at one end of a "third" (string of 24 cells). I pick a joint to test, and make a dot there (and circle it so I remember to silicone over it later). I push a small pin with a large head through the backing to the middle of the strap. I use a point 40 mm in from the cell border. This eliminates intermittent effects from any cells past this point. I press very firmly on joints up the the test point, and even on the test joint, looking for changes in voltage. (I have a power supply across the string of cells current limited to 2 amps.) Sometimes I'll do a sort of binary search technique. Once a joint is known to be bad, I repair it there and then before attempting to test any after it. That way I have a string of known good cells to work from (or almost good cells; I use 50 mV as my threshold of goodness, about 10% of the voltage of a cell).

But even when I'm sure I've found a bad joint, I now stop when the backing is off and I've pushed just enough black goo and EVA away to reveal contact area just to either side of the end of a connecting strap. That way I can measure the voltage across the joint. If it is less than 50 mV, I don't go any further with it. This happens many times, and saves a lot of time.

I find I usually don't want to close any repairs or explorations until I've finished a complete third. That way I can go back to that opened joint and do measurements on it.

When closing a repair, I put silicone over the top of the whole area that was peeled back. I run a finger over the back of it after the duct tape is on, to force the silicone to squeeze out evenly, so the repair looks good even if the tape comes off (or I take it off anyway).

When I have a pinhole that hasn't turned into a complete repair, I put a dob of silicone over the hold to seal it. I use a square of tape, about 50 x 50 mm, and cut from the middle of one side to the middle of the square. This allows me to adhere the tape to the backing in a sort of circular motion, with the last piece slightly overlapping the start, making a sort of conical blob of silicone. When this has completely dried in about 2 days, I take if off, leaving just the small, neat, conical blob of silicone. You can still see the dot and circle under the clear silicone.

I experimented with black silicone (still neutral cure of course) for one tube. It was pleasing enough, but in the end I decided I preferred the clear versions.

I hope that these notes will be useful to someone some day.

[coulomb]

Where are the pictures? Well, I didn't have a pinhole ready to cover when I felt the urge to post earlier.

    

A pinhole with its spot and circle, covered with a spot of silicone.

    
The piece of duct tape attached only at the bottom middle. I run my finger around the outside, tracing a neat circle as I go.

    

When finished, there is a small overlap, and a small run of silicon oozes out. I usually leave this as is. I leave it to set for two days; before that, the silicone isn't strong enough to resist the amazing sticking power of the duct tape. The final result is a well defined, slightly conical weatherproof cover for the pinholes. You can just see the ridge from where the duct tape was cut. The cover can be later removed with a metal ruler if necessary (for more measurements).

Here is one of the panels that had almost every possible joint repaired:

In the top right corner, you can see a rare "row @" repair. This one has all six of the relatively uncommon "row L" repairs (at the very bottom, no cells beneath).

Finally, a panel where two cells inexplicably became very hot in operation, some 15°C hotter than their neighbours. They seem to get hot right in the middle of the cells, so this has nothing to do with my repairs. These two on this one panel are the only example I've found of this. I found that the panel performs better, and the cells no longer overheat, if I bypass them:


It also sports some covers, many of these made with black silicone.

Edit: I've almost finished the ninth panel of ten. Weber thinks I'm nuts for persevering with this, and he's probably right.

[weber]

Weber thinks I'm nuts for persevering with this, and he's probably right.

But, my dear Doctor Coulomb, I can hardly complain about this most admirable quality of yours -- perseverence.

[PlanB]

Amen! I've met some amazing folks in my journey through life, some gifted academics in my student days & some hard headed businessmen later. The common factor in these disparate parties that stands out is the persistence & stamina. I wish I had more of both, pass me that Leaf brochure.