diy solar

diy solar

Reverse Bias due to shading and Bypass Diodes

Bud Martin

Solar Wizard
Joined
Aug 27, 2020
Messages
4,844
Interesting reading from Sunpower. I was looking to learn about shading effect and bypass diode, I did not know about heating issue due to shading, I member one of the member shows the pictures of burnt up cell that melt the panel.
Reverse Bias and Bypass Diodes
When cells are put into reverse bias, instead of converting photons to electricity, they convert electricity to heat. This happens any time the current generated by the rest of the cells in the string of modules exceeds the current that a cell can support. Common causes of reverse bias are shading from leaves, chimneys, or the buildup of soiling along the bottom of a module.
If too much heat is dissipated from a reverse bias cell, particularly if that heat is concentrated to a small defect region on the solar cell, the heating can cause failure of the other components within the module, such as the backsheet, glass, or encapsulant. Conventional Modules rely on bypass diodes to deactivate the portion(s) of the module where one or more cells are shaded, typically resulting in a 33% power drop. While this has a negative effect on energy output, it mitigates the heating issue. One tradeoff of this design is that the reliability of the module is tied to the bypass diode.
Diode life depends on several factors such as diode quality, junction box handling, and module installation; but, all diodes will eventually fail with use. This is because an activated diode runs at an elevated temperature, reducing the remaining life of the diode [6], [7]. Depending on how a diode fails, it can permanently remove a substring from the module or allow a shaded cell to run unmitigated in reverse bias, leading to module failure.
SunPower’s cells run in reverse bias at much lower temperatures, so bypass diodes are not required to ensure long term reliability (Table 1). However, when enough cells are in reverse bias, the cumulative power loss from the shaded cells can exceed the power produced from the cells in forward bias, so SunPower includes diode protection to enhance energy yield.
 

Attachments

  • wp-sunpower-module-40-year-useful-life.pdf
    615.5 KB · Views: 10
Last edited:
Yep. From a practical aspect to many DIY'ers who are using older panels, these stand out as to why one should test them for current output. Or perhaps part of a larger preventative-maintenance check.

The most common thing is that panels contain at least TWO bypass diodes, one for each "panel within the frame." Electrically, your single panel looks like two in parallel inside the junction box.

If you are only getting HALF of your rated output, then it is most likely that one of the bypass diodes has burnt up or degraded.

One common method of blowing a bypass diode is if one tries to attach a panel to a battery without a solar charge controller, and hooks it up in the wrong polarity. Now the bypass diode(s) acts like a fuse and blows up spectacularly. Moral: don't do this.

High quality panels have diodes that are electrically matched at the factory to be nearly as similar to each other as they can.

It is worth one's while as part of a preventative maintenance routine to open the junction box and check for loose connections to the bypass diodes. While they may have been tight initially, constant thermal temperature expansion (night/day) can loosen the contacts. Catch-22 happens now. Terminals get hotter, more thermal expansion and eventually diode failure.

These diodes are usually what are known as "Schottky" diodes. For DIY'ers, other than a PM check, they might wish to opt for much larger ones than those that initially came with the system to handle heat better, especially with economy panels. Ie, if your panel is only capable of 6 amps output, and you find only a 10amp diode in there, if you have space for a new one, fit in a 20 amp bypass diode. Handles heat better.

A search in many online stores for "Solar diodes" usually turns up these Schottky-type diodes.

The truly dedicated will buy a handful, measure their forward voltage drop, and try to use the two that are as closely matched to each other.

Glad you brought this up - many panels, especially if they are old have degraded bypass diodes, loose connections, or are simply running right at the ragged edge.
 
Yep. From a practical aspect to many DIY'ers who are using older panels, these stand out as to why one should test them for current output. Or perhaps part of a larger preventative-maintenance check.

The most common thing is that panels contain at least TWO bypass diodes, one for each "panel within the frame." Electrically, your single panel looks like two in parallel inside the junction box.

If you are only getting HALF of your rated output, then it is most likely that one of the bypass diodes has burnt up or degraded.

One common method of blowing a bypass diode is if one tries to attach a panel to a battery without a solar charge controller, and hooks it up in the wrong polarity. Now the bypass diode(s) acts like a fuse and blows up spectacularly. Moral: don't do this.

High quality panels have diodes that are electrically matched at the factory to be nearly as similar to each other as they can.

It is worth one's while as part of a preventative maintenance routine to open the junction box and check for loose connections to the bypass diodes. While they may have been tight initially, constant thermal temperature expansion (night/day) can loosen the contacts. Catch-22 happens now. Terminals get hotter, more thermal expansion and eventually diode failure.

These diodes are usually what are known as "Schottky" diodes. For DIY'ers, other than a PM check, they might wish to opt for much larger ones than those that initially came with the system to handle heat better, especially with economy panels. Ie, if your panel is only capable of 6 amps output, and you find only a 10amp diode in there, if you have space for a new one, fit in a 20 amp bypass diode. Handles heat better.

A search in many online stores for "Solar diodes" usually turns up these Schottky-type diodes.

The truly dedicated will buy a handful, measure their forward voltage drop, and try to use the two that are as closely matched to each other.

Glad you brought this up - many panels, especially if they are old have degraded bypass diodes, loose connections, or are simply running right at the ragged edge.
Hi @Substrate, thanks for this detailed reply! This is only somewhat related to the topic of this thread, but here goes:

I recently installed 4 210W panels in 2S2P and am getting almost exactly 3/4 of the rated output, suggesting one panel is bad. It seems like you know about opening up the electrical connection boxes on the back of panels to diagnose things, in particular bypass diodes. My first thought was that one panel's PV diodes must be open-circuited and all the current of the other panel in the 2S with it is passing through the bypass diode. Or, maybe a bypass diode was installed backwards or is shorted. The Vmpp per panel is specified at 33.55V, but instead of ~67V for the 2S2P array at solar noon, my MPPT is tuning to about 59V. I ordered a clamp ammeter so I might try a non-invasive test of short-circuit current for each panel before I open up the electronics and/or RMA the panel. Any tips would be appreciated.

EDIT: I should add, "before I RMA the panel" ... if I find one which has obviously lower short-circuit current. I don't have the ammeter yet.
 
Last edited by a moderator:
Schottky?
I might use those as reverse-polarity blocking diodes, but not as bypass diodes.
Schottky dioes have lower forward voltage (reduce loss when used in series with panel), but higher reverse leakage (loss and overheating when used in parallel with panel.)

"your single panel looks like two in parallel inside the junction box"
Some do (one open circuit would cut current in half.) These are likely to have half cells.

Most panels are like several lower voltages in series. A shorted diode would reduce voltage.
 
Hi @Substrate, thanks for this detailed reply! This is only somewhat related to the topic of this thread, but here goes:

I recently installed 4 210W panels in 2S2P and am getting almost exactly 3/4 of the rated output, suggesting one panel is bad. It seems like you know about opening up the electrical connection boxes on the back of panels to diagnose things, in particular bypass diodes. My first thought was that one panel's PV diodes must be open-circuited and all the current of the other panel in the 2S with it is passing through the bypass diode. Or, maybe a bypass diode was installed backwards or is shorted. The Vmpp per panel is specified at 33.55V, but instead of ~67V for the 2S2P array at solar noon, my MPPT is tuning to about 59V. I ordered a clamp ammeter so I might try a non-invasive test of short-circuit current for each panel before I open up the electronics and/or RMA the panel. Any tips would be appreciated.

EDIT: I should add, "before I RMA the panel" ... if I find one which has obviously lower short-circuit current. I don't have the ammeter yet.

Wired as 2s2p, if any one panel is open circuit or shorted, you'll get only 1/2 the power because it will kill performance of a neighbor.

Try 2s1p, then the other 2s1p. If each of the gives 3/4 the rated output of 2 panels, you're just observing that actual power is 3/4 of STC rating. Under ideal lighting hopefully that'll be 85% to 90% of STC.

The other thing that often reduces power produced is battery being fairly full. But in that case you would see voltage higher than Vmp, not lower.
 
Schottky?
I might use those as reverse-polarity blocking diodes, but not as bypass diodes.
Schottky dioes have lower forward voltage (reduce loss when used in series with panel), but higher reverse leakage (loss and overheating when used in parallel with panel.)
What's the alternative? So I guess you're saying forward bypass dissipation of 0.3V * 10A (let's say) is less than reverse leakage power of 50V * 1mA? I suspect they dissipate more power when "active" in bypassing current, than they do when leaking in normal operation, but I've just made up all the numbers of this post, so I'm curious what values you're assuming.
 
Wired as 2s2p, if any one panel is open circuit or shorted, you'll get only 1/2 the power because it will kill performance of a neighbor.
So if one panel's PV is open but its bypass diodes are active, so I effectively have (2s + 1s) in parallel minus diode losses, you think I would get significantly less than 3/4 nameplate power? Please note tone: this is a real question, not an argument.
 
So if one panel's PV is open but its bypass diodes are active, so I effectively have (2s + 1s) in parallel minus diode losses, you think I would get significantly less than 3/4 nameplate power? Please note tone: this is a real question, not an argument.
I guess to answer my own question, if the bypass diode of one of the two in 2s were active, then the voltage on the other panel would be ~ 59V - V_bypass ~= 58V, which would make it not produce much power at all because it's so high of a bias voltage. So perhaps @Hedges is right and if one panel was getting bypassed, it would cut out 50% of the production...

I guess a lesser alternative hypothesis is that *half* of a panel is getting bypassed, if @Substrate is right about them being divided in half. I don't know how 24V "RV" size panels are wired; I'd need to open up the electronics box to look at whether there is 1 or 2 bypass diodes.
 
So if one panel's PV is open but its bypass diodes are active, so I effectively have (2s + 1s) in parallel minus diode losses, you think I would get significantly less than 3/4 nameplate power? Please note tone: this is a real question, not an argument.

Yes, if you have 2s || 1s, you get the voltage of 1s and about the current of 2p. Maybe a bit closer to 1 x Imp + 1 x Isc, but basically 1s2p.

Like I said, disconnect 1 string (do that while no current flowing) and see what you get from 2s. Then reconnect that 1 string and disconnect the other (again while no current flowing) and check the other 2s. If there is one panel out of four that's bad, will be obvious.
 
What's the alternative? So I guess you're saying forward bypass dissipation of 0.3V * 10A (let's say) is less than reverse leakage power of 50V * 1mA? I suspect they dissipate more power when "active" in bypassing current, than they do when leaking in normal operation, but I've just made up all the numbers of this post, so I'm curious what values you're assuming.

I don't use any blocking diodes (which would be to prevent backfeed into a shaded string) because I consider the backfeed current to be quite negligible.

If a blocking diode is used, it's forward voltage x current is power being dissipated in normal use. I actually haven't found any diodes Schottky or otherwise with particularly low voltage drop at 5 to 10A sorts of current.
There are "ideal diodes" using a FET to provide millivolt on resistance instead of 0.3 to 1.5V drop.

Reverse leakage of a bypass diode would be power dissipated during normal operation. So I would use a silicon junction diode there to keep leakage low.
When used in forward conducting direction I don't care much about power loss, but the heat needs to be dissipated so it doesn't fail (as happens for some brand panels.) Thermal path to spread the heat over area would be the key. Reduced power dissipation would make that easier.
 
My SunPower panel (72 cells) has 3 string (24 cells per string) connected in series, and 3 bypass diode one per string.
Similar to this setup with schottky diode.
 
Last edited:
There are "ideal diodes" using a FET to provide millivolt on resistance instead of 0.3 to 1.5V drop.
Yeah, that makes sense to use an ideal diode (basically gate-boosted NFET). I didn't know they used those for PV due to needing the charge pump circuit, but it makes sense. For Schottky, here's an example part which is probably not the right one to use, but anyway: https://assets.nexperia.com/documents/data-sheet/PMEG100V100ELPD.pdf

That's 100V, 10A, 770mV forward voltage = 7.7W dissipated which is 2-3x over the maximum thermal dissipation of the part (1.66-3.75W). But probably a typical 350W panel is more like 50V 7A. So I guess using ideal diodes is probably what they do. But, leakage is tiny: 0.2uA reverse current which is approximately zero power dissipated. This is a "low leakage" Schottky, but still, they don't really leak much.

I wasn't talking about reverse blocking diodes, I thought those were not in widespread use inside panels, for efficiency reasons, and may only be added outside panels in certain installations. I was only talking about bypass diodes.

Edited.
 
Last edited by a moderator:
Like I said, disconnect 1 string (do that while no current flowing) and see what you get from 2s. Then reconnect that 1 string and disconnect the other (again while no current flowing) and check the other 2s. If there is one panel out of four that's bad, will be obvious.
Thanks. That was one strategy I was planning to try. The other one is to just short circuit each panel individually and look at Isc with a clamp ammeter.
 
Yeah, that makes sense to use an ideal diode (basically gate-boosted NFET). I didn't know they used those for PV due to needing the charge pump circuit, but it makes sense. For Schottky, here's an example part which is probably not the right one to use, but anyway: https://assets.nexperia.com/documents/data-sheet/PMEG100V100ELPD.pdf

That's 100V, 10A, 770mV forward voltage = 7.7W dissipated which is 2-3x over the maximum thermal dissipation of the part (1.66-3.75W). But probably a typical 350W panel is more like 50V 7A. So I guess using ideal diodes is probably what they do. But, leakage is tiny: 0.2uA reverse current which is approximately zero power dissipated. This is a "low leakage" Schottky, but still, they don't really leak much.

I wasn't talking about reverse blocking diodes, I thought those were not in widespread use inside panels, for efficiency reasons, and may only be added outside panels in certain installations. I was only talking about bypass diodes.

Edited.

Schottky diode reverse current goes up when hot, and it can get into thermal runaway.

Data sheet for that one lists maximum 3.5 mA at Vr = 60V and 150 degrees C. Graph indicates it would double at 100V vs. 60V.
Potentially as high as 7 mA at 100V, 700 mW. Temperature rise from 3 degree/watt to 90 degree/watt depending on thermal connections.
At least that is within the 3.5W rating.
It looks to me like no matter how stupidly this was used, it could withstand the reverse bias at up to 85 degree ambient.

Not that you would push it that hard. If you have 3 diodes in a panel, diode only sees 1/3 of Voc.

But I examine Schottky data sheets that way because I tracked down a product failure to a 40V, 1A Schottky diode that went into runaway operating at 24V reverse bias at room temperature. They don't all perform to what the specs appear to say.

A suitable bypass diode with conservative margin would be good for replacement, for people who lose a diode without the panel being ruined.
 
I've gone and measured them using the diode-tester on my multimeter and discovered I have two halves of two separate panels which are defective! This explains the 3/4" output power. Interestingly, both were on the same series string. EDIT: see followup below. They are not necessarily defective.

There are 2 strings and 2 bypass diodes in my (24V, half-width) panels. See image below.

I measured them in both polarities across each diode. D1 and D2 are the forward voltage of the bypass diodes. R1 and R2 are the reverse voltages. This doesn't make a ton of sense to me, but I suppose that minimal illumination in the PV diodes produces the reverse voltage (photo "sensor" style). The fact that 2 of 8 are outliers and open-circuited agrees with my 3/4-of-expected energy production. Measurements are in millivolts.

# D1 D2 R1 R2
1 184 176 858 875
2 190 162 878 inf
3 168 161 854 839
4 156 170 inf 809
 

Attachments

  • IMG_0649.JPG
    IMG_0649.JPG
    824.6 KB · Views: 13
Last edited by a moderator:
DigiKey has a nice writeup of a white paper, using active FET's as low loss bypass diodes

Blocking diode function is nearly always handled by the charge controller, also as a FET

 
Update: this may be inconclusive: the strings which do not reach "overload" by default, actually do reach overload when illuminated with a flashlight. So I think it is just that the diode tester uses a small voltage and there is uneven illumination. While there could still be open circuits, what I'm saying is that the above test doesn't adequately test for them.

Newpowa support responded fairly quickly and suggested I check short-circuit current for each panel. I'll do that tomorrow or Friday once I have my clamp ammeter.
 
I left 2 panels in series today and short-circuited the other two individually. Everything looks reasonable: 6.24A Isc, 5.6-5.7A Impp, 60.9V Vmpp versus 6.55A Isc, (2x) 33.5V Vmpp, 6.24A Impp on the panel sticker. Peak power production was roughly half of the prior 2 days, or a bit more because it was cooler outside today. This suggests there's no point in testing the other 2 panels tomorrow as they must have been producing the difference. It's just a bit disappointing, coming in at about 4kWh per day, versus what NREL estimates would be 138kWh/month or 4.45kWh/day for May. I guess 10% below estimated isn't too far off and it's early in May. Perhaps my 20A overcurrent breaker + 63A GFP breaker in series are dropping some volts before the PV hits the MPPT.

False alarm I guess, and thanks for the help/discussion above anyway. I learned something about bypass diodes opening up those enclosures and measuring them!

1620360294894.png
1620360320657.png
 
Last edited by a moderator:
The most common thing is that panels contain at least TWO bypass diodes, one for each "panel within the frame." Electrically, your single panel looks like two in parallel inside the junction box.

If you are only getting HALF of your rated output, then it is most likely that one of the bypass diodes has burnt up or degraded.
One interesting thing which seems obvious in retrospect but is pointed out in the video @Bud Martin linked is that each cell produces the short-circuit current in panels that have every cell connected in series. So testing short-circuit current doesn't tell you the whole picture. You could have 1 of 2 (or 2 of 3) strings completely dead within a panel, and the remaining string would produce the short circuit current and its current would flow through the bypass diodes, I think. So for this reason, he says and I agree, that short circuit current isn't really a valid way to check a used panel's functionality.

But, please tell me if I am misunderstanding this in some way.
 
Some day the current crop of MPT controllers might be called panel killers. HBpowerwall did some FLIR videos on direct water heating where due to excessive load, many cells overheated by going into or close to bypass. A lot happens before those bypass diodes activate. This was not due to shade! Cells are just crappy diodes with inperfections and it will be the weak ones that overheat. Over time it could be expected that these cells will get weaker or fail. When the current was reduced from the panel, many of these hot spots went away. These were all used panels that had years on them.
 
Back
Top