faulty solar panel

[crispyone]

I have 2 banks each of 2 x 100W solar panels, and normally the output (measured as battery charging current) of each bank was roughly equal. During a recent cruise one bank showed only 50% of the other. I checked the voltage at the controllers - 18v for one bank, 37v at the other. I checked the voltage at the junction box on each of the two panels in the faulty bank - 18v on one, zero volts on the other.

After disconnecting the dead panel, and making several checks, still getting no volts, I removed the panel, which was bonded to the cabin.

Sod's law - when I checked the volts after removing the panel it was generating - yes, you've guessed right, 18v.

Any ideas what kind of fault could show such symptoms?

 

[RichardS]

Bad connection somewhere in the faulty panel or its junction box? Fixable if in the junction box but otherwise probably not.

Richard

 

[William_H]

The panel may indeed show 18 volts or even 20v for a panel designed for 12v system however this may only be seen with no load. You can connect an amp meter (more than 5 amps) across the panel (no other connections) when in the sun and you should get around 5 amps. If you do get that 5A or so then panel is OK. If not as said bad connection or bad diode in the panel junction box (this can be bypassed for test). However the panel contains many connections of the individual cells to one another under the glass or the cells them selves can become corroded if water gets in. You can sometimes see discolouration. In which bin it. olewill

 

[crispyone]

I don't understand the internal wiring of the panels (these are flexible panels) but there was continuity of the circuit through the faulty panel because it was wired in series with its brother.
I have tested both panels in the early morning sunshine and both are giving a small similar current reading.
As suggested, probably a bad connection or component and I am guessing this must be in the junction box itself.
The junction box has silvered blades coming in from left and from right, and a tiny black plastic component in the middle (is that the diode?).
Am I right in thinking that if there is continuity then only this component can be to blame? I assume I would have to replace the entire junction box which is marked PV-XC502, but that doesn't look like a viable DIY job.

 

[RichardS]

If you have a soldering iron you should be able to replace the diode or re-solder its connections if it has come loose or has a dry joint. Are there any markings on it?

Richard

 

[peteK]

There are 2 ways the diode is wired,across the terminals, to act as a bypass diode if the panel is wired in series with another panel
or in series to act as a blocking diode to prevent discharge at night.

 

[crispyone]

the diode is marked with a diagram like a Y showing 2 branches joining with a blocked arrow on each leg, and carries a number 20S0045.
the diode looks to be securely soldered.
the fault is the worst kind - intermittent !!

 

[crispyone]

There are 2 ways the diode is wired,across the terminals, to act as a bypass diode if the panel is wired in series with another panel
or in series to act as a blocking diode to prevent discharge at night.

in view that there was no volts but continuity of the circuit was maintained, does this imply it is the first type?

 

[William_H]

I think first thing to do in fault finding is to wire the suspect panel on its own to the controller or even straight to the battery and observe performance. If it is bad (or occasionally goes bad) then try bypassing the diode. (yes that is a diode a type commonly used in computer power supplies actually 2 diodes 3 connections doing both series reverse current blocking and parallel bypassing). So connect wires to the 2 silver strips that connect to the diode. No need to unsolder the diode. If panel is still bad then fault lies in the panel itself. If it is OK and diode is bad then replace it or just try 2 panels in series without diode. good luck olewill

 

[crispyone]

So connect wires to the 2 silver strips that connect to the diode. No need to unsolder the diode.

Many thanks for your advice. I have 3 silver strips, I assume I link them all together to negate the effect of the diode. Correct?
What is the risk if I do this permanently?

 

[William_H]

Many thanks for your advice. I have 3 silver strips, I assume I link them all together to negate the effect of the diode. Correct?
What is the risk if I do this permanently?

My guess is that you have in effect 2 panels in parallel each with its own diode. So one strip will go directly to one outlet wire the other 2 strips will go to the outside connections to the dual diode. The centre pin of the dual diode then goes to the other outlet wire. The little arrow marks on the diode should point to the pos end of the diode. Just looking at one such diode the arrows go to the centre pin which I think might be typical. If this is the kind I am thinking of. Square black plastic with a bolt hole in one end. (used for attach to a heat sink when used at high current.)
So if this is the case as I picture it. A picture would help. You connect the 2 strips that went to the diode outer pins together and they should give pos out the remaining strip will be as already connected to negative out. You can just check current out of each of these strips individually. Might shop a bad half of panel. Removal (bypass) of the diode should make no difference if panels are in series in any case using a controller should stop any battery current flowing back into panel in dark. good luck olewill

 

[crispyone]

Might shop a bad half of panel.

so there are 2 diodes are encased in the little black cube; it doesn't seem feasible that only one of them has failed 'cos I got zero volts from the complete panel when checking between the two cable terminations in the junction box containing the diodes, although I retained circuit continuity. Still scratching my head, and of course I may never find out 'cos the fault has not repeated (yet!!) since re-assembling the system. If I had lost continuity in the circuit the fault would have been obvious, but that was not the case.