2 panels acting strangely

[Rabrsniver]

I have 2 Sharp 24V 165 watt panels wired as 48 volts, going through an Outback FM60.

I was rewiring today, and had something weird happen. My VOC was about 62V at the panels. When I hooked up the new wires, the FM60 didn't show any voltage.

I checked the wires for breaks or shorts, and they're fine. So I went back to the panels and rechecked things. I was only getting 13.6V. - way too low.

So I separated the 2 panels and checked them individually. One reads 13.6V, and the other 0 volts. I know the FM60 is working because I hooked some other panels up to it.

What happened? Did I short out the diode(s) or what?

Help!
John

[Flux]

It seems as though you may have blown the parallel diodes. there is probably one diode across each 12v section. You may have shorted one on the panel that gives 13v and both on the other one. You should be able to get at the diodes and change them.

My suspicion is that you reversed the connections and the input capacitors in the controller passed a large current pulse through the conducting diodes, nothing in the panels themselves could do this. I assume there are no series diodes or they would have blocked the current.
Flux

[Rabrsniver]

Hi Flux,

Thanks for the info. I have removed the cover where the connections are in the panels(s). Of course they are full of silicone.

When I have them cleaned out, what exactly am I looking for? There are 4 soldered connections. 2 of them go to the cables and I assume the other 2 come from the cells, right?

What do the diodes look like, and what happens if I just bypass them? I am out on an island with no cheap transportation to "civilization."

I DO have some diodes here that were from another project. Do they have to be a specific kind/capacity?

Thanks,
John

[Flux]

I haven't used Sharp panels, some makers have the diodes where they are accesible others may not.

It seems from your description that you may not be able to get at the diodes from the connector box.

Normally the diodes are black cylindrical things with a band at one end. Usually there are shunt diodes across sections of the panel to prevent total loss of power if one cell is shaded. From your description I suspect these are the ones causing trouble. Other manufacturers have one series diode in series with the output to reduce drain at night and thease are more use for panels connected in parallel.

For your series connection you can manage perfectly well without the series diodes. If desperate you will probably manage without the parallel ones as long as there is no shading.

Withiout knowing what Sharp do I can't offer much advice about where to look for the trouble. You should certainly find any series diodes in the connector box and it may be that one is open circuit on the dead panel. the low voltage one must have a shunt diode failure or something has happened to the cells or connections. Whether or not you can fix things depends on what has actually failed and how much chance you have of getting to the trouble spot.

I hope someone familiar with Sharp panels may be able to help you.

You can use any diodes of sufficient current rating, normally the voltage rating is more than adequate, I haven't seen any for sale under 100v for a long time.

Flux

[Rabrsniver]

Hi Flux,

I just finished getting the silicone out. There are 3 diodes in each panel. They have 3 leads on each diode. The ones on either side are fairly easy to see where they're soldered, but the middle one has a VERY short lead and goes down where I can't see the connection.

How do I test the diodes? I know they're made to only send current one way, but I don't know what they should read on a voltmeter. I have one of the cheap Harbor Freight multimeters, and it has a diode test setting, but I don't know what a "good" reading is. Of course it also has regular continuity resistance settings.

The diodes I have are 400V, 20A. I only have 5, so I want to make sure I don't replace one that is good, since I am one short of the total of 6 that I have in the panels, and it looks like it won't be too easy to replace them.

Thanks,
John

[Flux]

We are into a bit more guesswork now. I suspect that if they have 3 leads they are a dual package ( TO220 /TO247 transistor type package )

They may just have the 2 diode sections wired in parallel or possibly some are being used as 2 diodes for better control of shading.

If you can trace the wiring you should find one diode unit connected in series with one of the panel output leads. If so that is a series diode.

As nobody so far has come up with any information on Sharp panels we are somewhat in the dark. I think I would start with the panel that gives 13v, put it in good light and prod around with your meter on the volts range. Put one probe on the middle leg of each diode and probe the other 2 legs in turn. I suspect you will find the 13v across one of the diodes. That should be a parallel diode and you should be able to deduce the other one from the layout. Remove the other one that looks to be the remaining parallel one and test it. The middle leg is common so it should show as a diode from the middle leg to either outer one. If it is shorted it will show zero in either direction.

Take one of your good spare diodes and check it with the meter on the diode range, try it both ways to compare. Normally the thing reads about 0.6 in the conducting direction ( nominal diode volt drop) and goes full scale in the blocking direction. When you know what your meter shows you can test the diodes from the panel.

If as I suspect the two halves of these diodes are in parallel you can replace them with your spares, just connect them the same way round as the good one on the 13v panel using your meter to decide on the conducting direction.

If Sharp have done something a bit more sophisticated and actually used the individual diodes of the packages then you won't have enough spares to do the job as you will need 2 to replace each dual unit.

Trying to sort this without ever having seen a Sharp panel and not having it in front me is far from easy but I hope I have given you enough idea to get somewhere..

Flux

[oztules]

If you damaged the diodes, then there is a fair chance all of them are bypass diodes.
If they were blocking, then you could not have damaged them in the first place (the blocking diode would have stopped you)

A possible test for short circuit diodes without invasive surgery, may be as follows.

All cells seem not to be  created equal.... in a shorted panel exposed to the sun, you will find that some cells heat differently (substantially sometimes). Some will get hot, others will not. This tells us that current is flowing in the panel, and heating up the cells with slightly higher resistance than the rest... we have a short. (bypass diodes??)

If the panel remains uniformly even, then there is probably no current flowing, and an open circuit is present. (blocking diodes??).

...although my experience is diodes usually fail short... not open. I'm guessing bypass diodes.

If shading is not an issue, then no diodes and driving your controller will yield the same results  as before..... If shading is a problem pull em all out (diodes) and replace them with your spares. At least try to bypass each 36 cell section. (4 diodes total here for 2 panels (144 cells total)).



...............oztules

[Rabrsniver]

Hi Oztules,

As you have read my previous posts, I  have 3 diodes in each 24 volt, 165 watt Sharp panel. Do you know how to test them? I have one of the "cheapy" Harbor Freight mulitmeters which has a diode test position. But I don't know which of the 3 leads on the diodes to attach the meter leads to, and what reading I should get for a good diode?

Also, I can't seem to see where the middle lead of the diode is attached. (Or is it not used when the diode is used for blocking?) This 3rd lead seems to head straight back toward the panel, whereas the other two are just soldered in series between each pair of terminals.

As you can probably tell, electronics is not my strong point. I know a diode has an anode and cathode, but what is the 3rd lead? A ground? Al lot of the diodes I've seen on-line only have 2 leads which makes more sense to me. It is just attached in line and only allows current to go in one direction.

Also, I only have 5 diodes here to use as replacements, so I don't want to replace any that are still good.

Since the one panel shows NO voltage, does that mean all 3 diodes have blown? The other panel shows only 13.6 volts, so to me that means that at least one of the diodes is till good.

John

[Flux]

Normally 3 lead diodes are dual units with a common connection and they are usually in a transistor type package. Any chance of a picture or a part number.

These 3 lead versions are often schottky or fast diodes, schottky would be more likely here. I can't tell from your description but it seems as though the centre lead is actually connected to the cells and the other two leads are an external connection. You could try prodding on the centre leads with your multimeter, if they are series diodes and open circuit you may find volts there.


I took it that 2 would be bypass diodes across 36 cells as Oz mentions. I thought it most likely that the 3rd would be a series diode. Any series diode should be obviously in series with an output lead ( normally positive).

I am not aware of a 3 lead package with the middle lead not used. There are a lot of single diodes in the TO220 package but they normally only have 2 leads.

Have a good look and try to establish where the output leads go and see if there is an obvious series diode.

If you can remove a diode easily then try testing it with your meter. The centre lead will be common and it should have a diode between the centre lead and each outside one. You will easily be able to distinguish between open circuit ( both ways round on the meter probes) short circuit and a diode will show open circuit one way and partial conduction in the other.

I have not seen these 3 lead diodes used in any panels I have dealt with so I really am guessing whether the sections are in parallel to form a single diode or whether they are doing something more cunning and using the individual sections.

You could tryOz's suggestion of looking for changes in surface temperature as you have different fault conditions in the two panels and there may be a difference in behaviour, one section of the partly working one won't have a short and if you find that section cooler then it is a fair bet the others are shorted and that is back to parallel diodes.

Flux

[mab]

If you damaged the diodes, then there is a fair chance all of them are bypass diodes.
If they were blocking, then you could not have damaged them in the first place (the blocking diode would have stopped you)

I have to disagree with that statement - If they were damaged by reverse polarity connection to a capacitor as Flux suggests, then all the diodes would've been forward biased.

I have one of the "cheapy" Harbor Freight mulitmeters which has a diode test position. But I don't know which of the 3 leads on the diodes to attach the meter leads to, and what reading I should get for a good diode?

I've got a similar meter, the way the diode test works is it supplies a constant 1mA and reads mV across the test load so you might expect the following readings:-

meter probes not touching; reads overrange (1 in 1000's column; other digits blank)
meter probes shored together; reads zero ( or as near as makes no odds).
meter probes connected to 10 ohm resistor; reads 10

And for a diode that's NOT connected to the solar panel:-

connected to a working reverse biased diode; reads overrange
connected to a working forward biased diode; reads typically between 200 and 900 (i.e. the normal forward voltage drop of the diode)

connected to a blown (open circuit) diode; reads overrange whichever way you connect it to the diode
connected to a blown (short circuit) diode; reads zero whichever way you connect it to the diode.

It's difficult to say what reading you will get with the diodes connected to the panel, but if (as suspected) the bypass diodes are shorted, then I would expect reading substantially less than 200; I would suggest you test in weak daylight if practical.

hope this helps

mab

[Madscientist267]

Yes, the blocking diode would be forward biased if reverse polarity is applied to the module, but the bypass diodes would be as well, so its highly likely that at least one of them was damaged.

Not to increase the 'oh crap' factor any more than necessary, but IF there was damage to the bypass diodes under reverse polarity and they opened up:

Reversal can potentially damage the entire module, depending on the voltage applied (and current available, which could be potentially much higher from a battery vs. from within the panel itself).

A snip from this article explains what could have happened:

If they cannot resist the voltage, then a strong current flows due to the effect of the electric "breakthrough". In this event, charge carriers typically are accelerated in a strong electrical field produced inside the diode within the solar cell. That generates new pairs of charge carriers and can thus lead to a high, and in the worst case uncontrollable current that can destroy the solar cell and the entire module.

So it's possible that not only the diode(s) were damaged, but the entire panel as well. Not to say empirically that this is what has occurred, as a shorted bypass diode is very possible as well.

All of the above of course is assuming polarity reversal was the cause.

Other possibilities here come to mind... Ruled out lightning?

Steve

[Rabrsniver]

Thanks to everyone for their help/suggestions. I will go out and try to test the diodes and see what I come up with, now that I have a better understanding of what I'm dealing with.

John

[oztules]

"I have to disagree with that statement - If they were damaged by reverse polarity connection to a capacitor as Flux suggests, then all the diodes would've been forward biased."

Yes that is quite true.

I didn't think that through at all did I.... not very bright sometimes  I'm afraid.

Well corrected mab.




................oztules

[Steve Trumann]

I may not be the smartest tec out there. But why don't you just put a jumper across the suspected diode, and check for voltage. If you get your voltage, then the diode is bad.

Steve Trumann

[Madscientist267]

That would work for the blocking diode. For the bypass diodes, the scenario would be a little different, and possibly meaningless:

Jumping the blocker would return the panels to 'normal', assuming the diode is open, and nothing else is wrong. A shorted blocking diode would not give way to the symptoms provided - The only result in that case would be drainback at night.

Jumping the bypass diodes would effectively short the section(s) of panel that it/they are associated with, bringing the output to zero, for the associated section. Might be able to unveil functioning components this way - good sections being shorted would cause the overall output of the entire panel to drop significantly.

Jumping the 'dead' sections however wouldn't tell you much of anything, as it wouldn't reveal WHY the section is not providing output. Could be a defective string of cells, solder joints, shorted bypass diode... Here's why:

If a bypass diode is open, it will have no effect directly (when the panel is in sunlight), as it would normally be reverse biased in that situation. Shorting it would do nothing at all in this case, as to the rest of the chain, it would just be a 'shaded' section. EDIT - This didn't really look clear. What I mean is, open or 'OK', the diode would have the same effect, all else being equal (all cells in the panel are otherwise functional, and the entire panel is exposed to full sunlight).

If the bypass diode in question is shorted (internally), then jumping it would reveal nothing either. Nothing from nothing is nothing.

A damaged string of cells with a working bypass diode will just cause the entire panel to act as if the string in question is shaded.

So, the moral of the story is, to effectively troubleshoot shorted/open bypass diodes, they need to be disconnected from the panel, or at least checked with the panel in complete darkness.

Not really practical to do in many real world scenarios and implements... Though if your thing is dark rooftops (for example), and your meter has a backlight, well, by all means.  

Problem here appears to be that the diodes aren't set up to ever be removable/replaceable. BIG time respect points lost for Sharp on this one, at least for me personally... but the potting was a big first clue unfortunately...

That being the case, he's kinda between a rock and a hard place.

I know one thing for sure, I never thought of it as being a problem, but I'll be checking to see what 'user' options there are in terms of what's inside the JBoxes of any new panel I ever buy... That's handy information to know!

Sucks that the discovery in this case was due to a component failure, still yet to be determined.

Steve

[Rabrsniver]

Well,

I did as much testing as I could, and determined that only one of the 6 showed any reading. (The middle one on the panel showing 13.6V) I figured at this point I had nothing to lose, so I dug out the rest of the silicone and removed the 5 "bad" diodes.

One weird thing (at least to me) was that the middle lead on every diode had been clipped off - attached nowhere. What I thought was weird is that the only reading I got on the good one was attaching the probes to the middle lead and one outside one. All the schematics I found online for 3 lead diodes, said that the one outside lead was "not used," just the anode and cathode.

Anyway, I soldered in the new ones, (clipping off the center lead) and I'm getting 41VOC and 40.8 on the other. Like I said, these are 24V panels, and the VOC rating of each 43.1.

The panel literature says they have a "10A fuse rating." I guess the way these diodes are wired they were acting as fuses?

Anyway, I'm back in business. Thanks to everyone for their input.

I'm NOT going to silicone the whole junction box back in. I will just silicone the cover well, just in case I run into this again. Digging out the silicone was about the hardest part of the job.

John

[Madscientist267]

Glad to hear you found the problem...

The center lead being clipped is because it really doesn't 'exist' to start with. Those are power Schottky diodes in a TO-220 case for heatsinking (whether or not there actually is a heatsink). That in and of itself shouldn't be considered a 'problem'.

The 10A fuse rating means they should be fused externally at 10A, not that there is one included in the panel. PVs aren't fused because intrinsically, they contain current limiting. You can short the output of a PV panel in full sunlight for the remainder of the life of the universe with no negative effect on the panel. The problem comes when something like this arises, where a battery is involved and miswiring or the like causes more current to flow somewhere than where it should. Then the fuse can protect things like wiring, and even diodes, if it's sized appropriately.

Even though I understand why they potted the whole JBox, to me there's no excuse for potting the diodes WITH it. The diodes should always be accessible. Big smack up side the head for Sharp on that one. That's just WRONG.

One other thing, when you can (if you haven't already), change your replacement diodes to Schottky; they have lower losses than standard PN junction diodes. Or better yet, install 'ideal' diodes...

Steve

[Flux]

I am glad you have found the problem.

The only 3 pin diodes I have come across are dual types and the middle pin is normally a common cathode. It is attached to the heatsink tab and if the connection is made to the tab by wire or by bolting the tab on to something then the pin can be clipped off.

Single TO220 diodes normally have 2 pins or occasionally 3 but with 2 linked. I think your diodes must have some connection to the big heatsink tab for them to work. Whatever you have done you have removed the shorts across the panel sections, unless you have sorted out the connections correctly you may just not have any diodes in place any more, that doesn't matter as long as the panels are not shaded. I can't see how you can replace with ordinary 2 pin diodes without having to connect one end to the mounting tab if the middle pin was clipped off.

Flux