How are PV panels divided into sections for the bypass diodes?

[vtpeaknik]

I need to decide whether to mount my 2 new panels, each about 3 by 5 feet with Vmp about 28V, in horizontal ("landscape") or vertical ("portrait") orientation.  This is on a roof with not a lot of slope.  When it snows, then partially melts, the bottom part of each panel will have some snow on it even as the top part is bare.  Ideally, the bottom part would be bypassed by a diode.  Are such panels normally wired so that the by-passable sections are along the long axis of the panels?  There are other constraints and considerations that have to do with the mounting arrangement and snow and wind loads.

[OperaHouse]

I believe they are generally done side to side.  Itg is a moot point if the panel is not specified at 12 abd 24V.

[Rob Beckers]

vtpeak, manufacturers actually use both orientations for bypass diodes, sometimes changing methods for different models from the same manufacturer (Suntech being one of those), so the foolproof way is to check the documentation.

If  you don't have that, most modules are divided into 2 rows of cells per bypass diode, when holding the panel up in landscape orientation, top-to-bottom (most are 6 rows of cells high in landscape, so there's a diode every two rows). What this means, for most panels, is that in portrait shading a row of cells along the bottom will zero output for the entire panel.

You can also open up the junction box, and physically trace how the diodes are hooked up.

Hope this helps!

-Rob-

[vtpeaknik]

The junction box appears to be "sealed" and I don't want to damage it as the panels are new and warranteed.  The panels are SV-T-205.  The documentation I've managed to find only says that it has 2 diodes (only 2?!) but no further details.  Can one figure it out by following the conductor traces that are visible from the glass side?

Would experimenting with partially shading a panel help figure it out?  I havn't mounted them on the roof yet, so could experiment on the ground.  I'm thinking of connecting a panel to a dummy load and using a large piece of cardboard to shade portions of the panel while measuring the voltage (or current) on the load.  I have on hand a couple of very large (100 watts or so) 9-ohm resistors, and one of them, or the two in series (18 ohms), may be a good low-impedance load for such testing on a cloudy day, when the panels would only yield a fraction of an amp and the voltage on the load would reveal the amps.  But is cloudy-day operation similar to sunny conditions in the way the bypassed sections work?

[vtpeaknik]

PS: This panel has 108 cells, arranged 6 by 18.  Looking at the "bus" conductors on the short edges of the panel, it seems that 2 cells are in parallel along one short edge, and the other 4 in the same row of 6 are in parallel.  On the opposite edge, it's 4 and 2 (swapped).  Is it a good guess that it's two parallel strings of cells, going down the length of the panel, then back, then down again?  That would be 54 cells in series for a total voltage of about 32 (if each is 0.6V), which fits the open-circuit voltage claimed for this panel.

If that's the case, and given that it's supposed to have only two bypass diodes, would it be one across a smaller secton, 1/3 of the cells, and the other across a larger section, the other 2/3?  If that's the case, then shading 1/3 of the panel, i.e. 2 rows of cells along a long edge, would reduce the power by either 1/3 or 2/3, depending on whether the shaded 1/3 is part of the smaller or larger section?

I might do some shading experiments, but need some theory to guide them...

[Rob Beckers]

Sounds like one of the 'odd' panels; that's a configuration I've not seen before (then again, almost all panels I deal with are 60 and 72 cells)...

Sure, experimenting is good and should tell you how the strings are hooked up. If it's two strings in series, with diodes, all you need is two pieces of cardboard the size of a cell, and a little tape (fold it over so it sticks both sides) to hold the cardboard on. Block one cell in a string and that string will go to bypass, the total panel voltage will drop by the size of that string. You'll need to pull some current to activate the diode, and this is something I've not tried to test (normally the docs tell you how the strings on a panel are arranged). My guess is that 0.5A should be plenty.

So, the way to find out how strings are connected is to block a cell, see voltage drop, block another cell: If the voltage drops to nearly nothing you hit the other string. Move the cardboard around to see what cells belong to that string. If the cardboard is on the same string as the one you initially blocked there will be no additional effect on voltage. Hopefully you get the gist of what I'm trying to convey...

One more thing: Do not connect/disconnect the solar connectors (MC4, Tyco, or whatever is on there) under load! They are not meant for this, and it will burn a hole in the plating that keeps the pins from oxidizing. So, either flip the panel over (cells facing the ground), then hook up your resistor etc., or connect a lead to the connectors first, then the resistor to that.

-Rob-

[vtpeaknik]

Ah, so that's why those MC connectors come with labels "do not disconnect under load".  Thanks for the tips.

I've done some measurements, and although the numbers are rough due to constantly changing cloud cover (from heavy to heavier), I think I got the answers I need:

The panel is SV-T-205.  Max power is supposed to be 205W at about 28V 8A.  Under the clouds I measured Voc 30V, Isc 0.5A - about 15 watts.  With loads of 9/18/36 ohms the voltage was 4.4/8.7/17.1.  That converts to a power of about 2/4/8 watts.  Clearly the higher resistance was closer to the max power point under those conditions...  As the clouds thickened the 17 volts decreased more towards 12V and eventually it started to rain and I retreated indoors.  But I managed to measure the following part-shade results with the 36-ohm load:

Shading a strip 2 cells wide (1/3 of the panel) along the long side of the panel, the voltage decreased to about 12V.  It was the same for such a strip on either side.  Shading a wider strip, 4 cells wide, or 2/3 of the panel, along the long edge, got the volts down to about 8.  Shading 1/3 of the panel (6 rows of cells) along the short edge farthest from the junction box, dropped the voltage to less than one volt.

Conclusion: for this panel, it is clearly partitioned along the long axis.  The results indicate perhaps 3 diodes even though the spec sheet says 2.  And it is clearly better to mount it in "landscape" orientation rather than "portrait" if I expect the lower edge of the panel to be sometimes covered with snow and ice even as the upper part becomes clear of snow.  (The location will not allow me to reach the panels and clean off the snow.)  I'll have two panels in series, and if each gives me about 9V under load using just the top 1/3 of the panel, the resulting 18V (from 2 panels in series) should be usable via the MPPT charge controller.

[Rob Beckers]

Sounds like a 'standard' panel...
Yes, you're right about being better of mounting panels like this in landscape when snow is involved. This is one consideration here for flat-roof PV jobs (we have lots of know, I'm in Canada). Even if the snow covers the bottom of the panel, the other 2/3 are still working. In portrait it would knock out the whole panel, even though less of the panel surface would be covered in snow.

-Rob-

[rumleyfips]

I just tried this experiment. Panel from Honeybee solar , Toronto , 245 w nominal.
It ran 2 batteries from 24.6 to 30 volts real quiuck at 8+ amps in full sun. Cloud cover dropped this to 25.8 and 1 amp.

Covering the bottom of the panel, two rows on one side and a diagonal corner all resulted in 0 amps. It would seem that orientation is not imortant.

We only had enough snow to plow 5 or 6 times in Nova Scotia this winter. In each case it melted within a few (~3) days.

Now I have to figure out who is right; those who say put the rows close together so snow will slide off or the ones who say leave lots of space to give the snow somewhere to go.

[Rob Beckers]

Rumleyfips, direct battery charging adds another wrinkle to the equation: If the panel voltage drops below the battery voltage you'll get 0 Amps, even if 2/3 of the panel is still working. For MPPT (grid-tie, or battery) this is not the case, and orientation matters, as snow tends to collect at the bottom of the panel and in landscape mode that often leaves you with 2/3 of the panel still working fine (bypassing the lower 1/3).

-RoB-

[rumleyfips]

Bob:

Thanks; I get it now. This morning with lower batteries I saw25% power with either one cell or one long row covered.

Portrait it is and I'll splurge on a extra piece of unirac.

John

[ghurd]

I am in the snow belt.
It snows a lot, but it doesn't get horrible cold.

Usually, here, if there is enough light to do much charging, the exposed cells and parts of cells, make enough heat enough to melt the snow and ice (or heat enough so snow/ice evaporates faster).
G-