Diode replacement

[cs.dk]

Hi,
Don't know if this question should be in beginners or technical. Mods may move it, if they feel for it

Anyway,
Got some solarpanels where the diodes are melted because of a bad internal connection in the box.
The diodes installed now are these; MIC 10A10 http://micindia.com:8080/micindia/resource/wcm/pdfs/10A05%20THRU%2010A10.pdf

I found this replacement from tme.eu, and i think they will be fine, unless i have missed something; Diotec P1000M http://www.tme.eu/en/Document/c5bb4a972f3aa89b44ae23a484a3fe84/p1000a.pdf

Will they do the job?

[madires]

I'd say yes.

[cs.dk]

Thanks, i'll order 500 pcs. and begin some repair. 
IIRC the panels are connected 24 in series to the inverter. Each panel is 30V, 8,2A. So 720 VDC in one string. (If that should be an important information)

[Kleinstein]

Cooling these size diodes is tricky. They usually require cooling through the leads to allow there rated current. So it may be important how they are mounted. At 8 A and a maybe 0.6 V drop this is already close to 5 W, thus more than the plastic part alone can dissipate.

On the other side, the diodes should only be needed if the panel in question is shaded. Usually this should not happen so much in full sun. Usually it is only under lower angle that some panels are in the shade - so normally the current would be lower than normal.

[cs.dk]

Thanks Kleinstein,
The original mounting is shown in the attached picture. There is a small plastic support under the diode leads (like the marked one in the picture). If a screw comes loose, or the connection gets bad, the support melts away, and the connection gets even worse.

The diodes them selve seem to be fine. I think every one of the melted ones, are from a loose connection. I was thinking about soldering them on the topplate instead. They have really many panels, where the diodes (or the whole plastic case) melts away.
A newer type with the diodes in springterminals seems to last much longer between fails.

[madires]

No metal counter plate? That would be a very poor design (or planned  obsolescence).

[cs.dk]

Nope, it's a plastic support.
I can take a picture later, when the diodes are removed, from a faulty one.

[Kleinstein]

Having 2 diodes in parallel at each panel indicates they had problems - however having didoes in parallel only has a limited effect, as most of the current tends to flow through the hotter one. This can be especially a problem with the P1000 replacement diode - this one seems to have a lower series resistance. So the current sharing might be worse with the replacement one.

Having 3 diode pairs so close together and effectively in series makes cooling even more tricky. Worst case (full sun to the other panels and one in the shade) this would mean about 15 W for this small plastic case.

[cs.dk]

Stupid question maybe, but how did you figure out the series resistance?

It seems like a shitty design after all.

[David Hess]

Why did they use such high voltage diodes?  The reverse voltage is only as high as each panel section can generate which appears to be about 10 volts in this case.

I might use one of these P600 packaged schottky rectifiers.

Matching the diodes for forward voltage drop at high current will help with current sharing.

[cs.dk]

I don't know the reason for their choise. But the diodes must stand the 700 VDC in the string, or?

Those schottky diodes are ~5x the price.