Wiring a large array to minimise the effect of shading?

grizzlyj

Tea pot tester
Hi folks,

Our soon to be arriving camper will be having a roof full of eighteen semi flexy solar panels, paired up to give 24v when the feed enters the camper.

Does it make any difference how each string of nine will be wired, so if one panel is shaded it won't knock down that whole half's output?

Cheers

Jason
 

DiploStrat

Expedition Leader
Was chatting with a gentleman who does solar for yachts - where shading is a tremendous problem. His ultimate, cost is no object, approach is a series of small solar controllers, each fed by a single panel. Thus each panel is in parallel and there is no loss from blocking capacitors. Extreme, but it works.

Probably not cost effective for most of us.
 

dwh

Tail-End Charlie
Well, when rigging solar (or batteries), "string" means "in series". So when you say "string of nine", that means "string of nine panels wired in series", and of course those two strings would then be wired in parallel...because if the two strings of nine were wired in series, then it wouldn't be two strings of nine, it would be one string of eighteen.

Then again, when you say, "paired up to give 24v". That implies that they are 12v panels wired as 9 series strings of two panels per string, and the nine series strings would then be paralleled.



The problem with shading, is that it not only drops the amperage, it also drops the voltage. If the voltage drops "too much", then it will be too low to charge a battery. Say you have a 24v battery bank. You need to have the array always putting out around 30v or more or you won't be able to push the "24v nominal" battery bank up to full charge voltage, which for a "24v nominal" battery bank would probably be at least 28.8v.



Here's how a typical solar panel is laid out internally:

Bypass-diodes-in-PV-panel.jpg



Note the three bypass diodes (orange blocks). Those are there to protect shaded cells. If you feed power through a solar cell, then it becomes a radiant heating element. So in this pic, there are 24 cells. They are all wired in series. Say one is shaded. Well now you are feeding the power of the other 23 cells through that one shaded cell and it's operating as a heating element. It might overheat and be damaged.

So the bypass diodes will bypass part of the series string of 24 cells if the voltage of that section drops too far.

Solar cells all put out around 1/2 volt. So in this example, 24 cells in series would put out 12v. Rigged with 3 bypass diodes, if one section is shaded (or maybe even just one cell in a section), then that section gets bypassed, and now there are only 16 cells in series (the other 8 have been cut out). Okay fine, but the voltage of 16 1/2 volt cells in series is only 8v.

So using the solar panel in the pic, which puts out a max of 12v, if we're trying to charge a "12v nominal" battery, we need to get it up to say 14.4v. So there's no way the panel in the pic could actually charge a 12v battery. We'd need two of them rigged in series. That would give us an array voltage of 24v.

Now say we get shading and one section of one panel is bypassed. That drops the array voltage from 24v down to 20v (8 cells got cut out at 1/2v per cell). If another section got shaded and bypassed, it would drop the array voltage to 16v. Either way, it's still enough to charge a "12v nominal" battery. But if one more section gets bypassed, the array voltage would be down to 12v, and that's not enough.


Now, in real life, "12v nominal" solar panels have at least 30 cells (15v Vmp) or 32 cells (16 Vmp). Most common is 36 cells for a Vmp of 18v. With three bypass diodes (12 cells per diode), the voltage of an 18v panel with one section bypassed would drop to 12v, and with two sections bypassed would drop to 6v. So any shade that causes a section bypass, also causes the voltage of the panel to drop far enough that it won't charge a 12v battery.

So let's say you have two 18v panels (wired in parallel for an array voltage of 18v), and one has a section bypass. That panel is now down to 12v. It isn't producing a high enough voltage to charge the battery, so it basically does nothing. You've now lost 50% of your array output.

But if you rigged them in series, then you would have an array voltage of 36v. One section bypassed, would drop the array voltage to 30v, which would still be plenty to charge a 12v battery. Another section bypassed, and your array voltage is down to 24v, which is still enough. Another section bypassed, and you're down to 18v, which is still enough.

So rigging the two panels in parallel, a single section bypass in one panel drops the entire array output by 1/2. But if rigged in series, a single section bypass only drops the array output 1/6. Two section bypasses drops the array output 1/3. Both are better than 1/2.

Thus, in general, for shade tolerance, series is always better.


But...(and there's always a "but")...

Rigging the array for a higher voltage, increases the voltage mismatch between array and battery, and the greater the mismatch, the greater the inefficiency, and the less power you harvest. And that will happen constantly. So the voltage mismatch can cause the system to work constantly at a lower efficiency. Bad news.

So now it depends on the charge controller. With a PWM controller, the voltage mismatch matters, so you don't want to rig the array voltage any higher than you have to. So with 18v panels charging a 12v battery, you would always rig the array in parallel.

But with an MPPT controller, the voltage mismatch no longer matters, since the MPPT adjusts the voltage, which PWM can't do. If you've got 18 x 18v panels, you could actually rig them in a single series string of 18 and have an array voltage of 324v. With such a high array voltage, you could have a whole bunch of sections bypassed due to shading, and still have a voltage plenty high to charge a 12v, 24v, 48v or even 96v battery bank.

And that is how you would always do it with an MPPT controller and an array mounted at a fixed location, like a house.


But...

If you rig the entire array in series, if there is any break in the circuit anywhere, the whole array stops working. There's not much chance of that happening to an array mounted to the planet, but an array mounted on a vehicle is a different story. Vibration, tree limbs, etc. can cause a loose or broken connection. So for a mobile installation, you want redundancy.


You didn't say what voltage your solar panels operate at. You didn't say how many watts they produce. You didn't say if you are going to use MPPT.

I'm going to assume based on what you said, that they are "12v nominal" solar panels, so you'll need at least two in series to get 24v. Okay, then you will have to have series strings. Again, the smallest you could get away with are 9 series strings of two panels per string, and the nine series strings paralleled.

Say they are actually 18 Vmp with three bypass diodes. Two in series would give a Vmp of 36v, and again, for a 24v battery bank, the array has to stay at 30v or more. A single section bypass on a series string of two panels would drop the output of that string of two panels to 30v. Fine. Another section bypass and the voltage of the series string of two panels, drops to 24v, and that's too low to charge a 24v battery bank.


So now you've got to work out a tradeoff between shade tolerance, and redundancy.

If you rigged 3 panels in series, you would have 6 series strings which you could then parallel. If any one string of 3 panels quit working, then you would lose 1/6 of the array output. Pretty good in terms of redundancy. And the string of three would have an array voltage of 54v. A single section bypass on a string would drop that to 48v. Two would drop it to 42v. Three to 36v. Four to 30v and that's as low as you can go and still charge a 24v battery. So pretty good in terms of shade tolerance as well.

4 panels per series string wouldn't work, because you'd end up with one string that only has two panels in it. The series strings have to have either 2, 3, 6 or 9 panels per string.

Two panels per string gives a lot of redundancy, but not as much shade tolerance. Rigged that way if you lose a string, you only lose 1/9 of the array output, but only two section bypasses and that string is out of commission. You could rig for 6 or 9 panels per string and increase shade tolerance, at the cost of decreasing redundancy. If using 6 panels per string, if a string stops working, you lose 1/3 the array output. If using 9 per string and a string quits, you lose 1/2 the array output.

But rigging 3 panels per string, if you lose a string, you only lose 1/6 of the array output and you've still got good shade tolerance.


So that's probably how I'd rig it -3 panels in a series string, 6 series strings paralleled.


(Disclaimer: Provided of course that you are using an MPPT charge controller and that my assumptions (wild-assed guesses) about your solar panel and battery bank voltages are correct.)




[And...to those nit-pickers out there who've actually been paying attention to what I've said over the years, I *have* somewhat revised my thinking on the series vs. parallel question in recent years. So yes, you caught me. :D ]
 

Martyn

Supporting Sponsor, Overland Certified OC0018
If you are worried about the effects of shading on glass panels you need to read this article https://www.flexopowerusa.com/blogs/news/117385475-partial-shading-devastating-for-glass-panels

We reproduced this test in house and came up with basically the same conclusion, except we used a smaller piece of cardboard to achieve the same results.

I'm not sure about the construction of the semi flexible panels you are using, but a repeating this test could produce interesting results.
 

DaveInDenver

Middle Income Semi-Redneck
That Flexopower test is a good demonstration of the concepts dwh talks about, the series and parallel combination of cells into strings to make modules.

The position that they placed the cardboard takes a cell from each string out of the circuit, which has the effect of causing the bypass diode on each string to conduct. This is by design to protect the panel from overheating the shaded cells that would otherwise have current forced through them by the illuminated cells. There are ways to configure modules to make it less so, but it's something that all polycrystalline modules will do at some level.

Flexopower uses smaller modules and parallels more of them to achieve the current spec on the array. So the two shaded modules have also gone to zero but since they have the equivalent of 12 panels the whole system output isn't impacted as badly.

The nature of how a panel is shaded will affect them differently. A more fair comparison would be to put a 10% filter over the glass panel since that's not a 10% shading, but 100% shading on 10% of the module area with an array size of 1. The Flexopower has two modules with 100% shading and 10 with 0% shading and an array size of 12.

Either way is fine but it's more an observation than anything. You can achieve better results for shading like this using several smaller panels to configure a larger parallel array or even better is to use multiple panels with dedicated controllers to feed the battery so that each array is optimized.
 
Last edited:

dwh

Tail-End Charlie
20160331_114612_HDR_resized_medium.jpg


Yup. 36 cells, two, possibly three bypass diodes. No surprise.


It's a bit of an apples and oranges comparison. The unit pictured below is made using thin film instead of glass cells, so it's a totally different design.

Couple of important differences. First is price. Glass is down around a dollar a watt, but I'd bet the flexo is quite a bit more than that.

Another is the construction. No one is gonna screw down 18 flexos in their roof and expect the nylon backing to survive for long.

Another is size/energy density. Thin film is less efficient, so as far as I know, to get the same wattage, you need more surface area.

20160331_114452_HDR_resized_medium.jpg



But here's what bothers me about that comparison test...

Lay both panels out in the sun. Cover 10%. Test with a meter. One panel looks way better in terms of amps.

But that doesn't mean either one is putting out enough voltage to charge a battery.

They show that the glass panel dropped amps 78%. Yea, tested with a meter. But if you try to charge a battery with it, you get nothing because the panel voltage is lower than the battery voltage. So really it's the same net effect as 100% total loss.

Under partial shading, the important number isn't the amps, it's the volts. So what if the flexo tests out at 2.9a - doesn't matter if the voltage dropped below battery voltage.

So what are the voltages of those panels, partially shaded? Time to do the test again.
 

DaveInDenver

Middle Income Semi-Redneck
The glass panel is a Solarworld 85W, which has two bypass diodes. The placement of the cardboard wasn't random. It's actually a very good panel, made in Germany too.

This is how the cells are laid out in the module. The cardboard was positioned to block the four cells in the middle at the end, compromising both strings.

image21_mid.jpg
 

DaveInDenver

Middle Income Semi-Redneck
But here's what bothers me about that comparison test...

Lay both panels out in the sun. Cover 10%. Test with a meter. One panel looks way better in terms of amps.

But that doesn't mean either one is putting out enough voltage to charge a battery.

They show that the glass panel dropped amps 78%. Yea, tested with a meter. But if you try to charge a battery with it, you get nothing because the panel voltage is lower than the battery voltage. So really it's the same net effect as 100% total loss.
I believe the Flexopower is an array of 12 6.5W foldable modules that they've bussed. The way they covered two would remove at most 17% of the capacity from the array, so their claimed numbers are probably justifiable under load. The bus voltage would still be 19.2V from the 10 remaining fully illuminated panels and you'd still have 83% of the specified current. I also agree the glass panel would actually be effectively 0V under load.

You could build the same array using 12 Power Film R7 panels that you fasten to a fabric backing. Those are 14" x 23", so laid out 2 x 6 would be an array somewhere around 84" x 46" (plus borders). It would cost you somewhere north of $1,000 to do that. But the Flexopower 79W panel is $670 and slightly smaller, so the actual parts are obviously different.

For comparison, the 85W Solar World panel is about $140. Point is you have a trade off between cost and performance. You could also use four 20W glass panels to make an 80W array for maybe $200 that putting that same size cardboard would yield at most a 25% reduction (assuming you eliminate one module from the array). In that case the designer could claim 100% shading on one panel would still leave you 75% capacity. It's all about marketing speak.
 
Last edited:

dwh

Tail-End Charlie
The placement of the cardboard wasn't random.

Yea, that too. They carefully set it up to get a bypass on both strings. If they had shifted the cardboard left or right, it would have been only a single string bypass for a 50% drop in amperage.

But then the "comparison" wouldn't have looked as impressive.

Marketing smoke and mirrors.



(Not that it would matter in the real world - the glass panel even with only one string bypassed and only a 50% reduction in amperage, stiil would have a voltage too low to charge a battery, so it'd still be a total loss.)
 

DaveInDenver

Middle Income Semi-Redneck
Yea, that too. They carefully set it up to get a bypass on both strings. If they had shifted the cardboard left or right, it would have been only a single string bypass for a 50% drop in amperage.

But then the "comparison" wouldn't have looked as impressive.

Marketing smoke and mirrors.
Totally a marketing stunt.
(Not that it would matter in the real world - the glass panel even with only one string bypassed and only a 50% reduction in amperage, stiil would have a voltage too low to charge a battery, so it'd still be a total loss.)
That's true, with half the strings bypassed the output voltage would be half minus a forward diode drop but the current would remain the same so you could still extract significant power. In a 36-cell module with two diodes protecting 18-cells each the terminal open circuit voltage would still be ~10.1V with one cell fully shaded and the current would still be whatever the cells are rated, so say 4A with full illumination on a nominally 80W/19V panel. That would require using a buck/boost configuration for your charger but you should still be able to get power. It's not the best situation of course. You'll also have losses also in the bypass diode conducting, which could be a few watts worst case.
 
Last edited:

dwh

Tail-End Charlie
Yea sure if the charge controller had a boost converter and bumped up the voltage.

But I'm not aware of any commonly available charge controllers that do that. They all have buck converters to step the voltage down instead of boost converters to step it up.

So if the solar voltage drops below battery voltage...no joy.
 

dwh

Tail-End Charlie
They exist.

Yes David, I am quite well aware that they exist, thank you very much.

I said commonly available.

The OP is talking about eighteen PV modules on the roof. Even if they were only 50w each, that's still 900w nominal. Into a 24v battery bank we're talking about a 40a array. 80 if those are 100w modules.

While some obscure 10 or 15a charge controller might be technically interesting from an academic POV, I don't think it really applies in this situation.


(Unless of course, the OP wanted to do as the marine electrician that Diplo described and buy 18 charge controllers. Then sure, might as well buy the buck/boost while you're at it to harvest every possible watt. Makes perfect sense on a boat.)
 

LeishaShannon

Adventurer
Flexopower... hehehe.

Their 79W model is only slightly smaller than a 330W LG panel... (1.3m2 vs 1.6m2) but twice the price.

79W - 12% = 69W
330W - 78% = 72.6W

So even using their ridiculous test the LG panel would still produce more power...

And when there is no shadows you're making 4x as much power...

Who buys this crap?
 

Mwilliamshs

Explorer
...The problem with shading, is that it not only drops the amperage, it also drops the voltage. If the voltage drops "too much", then it will be too low to charge a battery. Say you have a 24v battery bank. You need to have the array always putting out around 30v or more or you won't be able to push the "24v nominal" battery bank up to full charge voltage, which for a "24v nominal" battery bank would probably be at least 28.8v.

Solar cells all put out around 1/2 volt. So in this example, 24 cells in series would put out 12v. Rigged with 3 bypass diodes, if one section is shaded (or maybe even just one cell in a section), then that section gets bypassed, and now there are only 16 cells in series (the other 8 have been cut out). Okay fine, but the voltage of 16 1/2 volt cells in series is only 8v.

Now say we get shading and one section of one panel is bypassed. That drops the array voltage from 24v down to 20v (8 cells got cut out at 1/2v per cell). If another section got shaded and bypassed, it would drop the array voltage to 16v. Either way, it's still enough to charge a "12v nominal" battery. But if one more section gets bypassed, the array voltage would be down to 12v, and that's not enough.


Now, in real life, "12v nominal" solar panels have at least 30 cells (15v Vmp) or 32 cells (16 Vmp). Most common is 36 cells for a Vmp of 18v. With three bypass diodes (12 cells per diode), the voltage of an 18v panel with one section bypassed would drop to 12v, and with two sections bypassed would drop to 6v. So any shade that causes a section bypass, also causes the voltage of the panel to drop far enough that it won't charge a 12v battery.

So let's say you have two 18v panels (wired in parallel for an array voltage of 18v), and one has a section bypass. That panel is now down to 12v. It isn't producing a high enough voltage to charge the battery, so it basically does nothing. You've now lost 50% of your array output.

But if you rigged them in series, then you would have an array voltage of 36v. One section bypassed, would drop the array voltage to 30v, which would still be plenty to charge a 12v battery. Another section bypassed, and your array voltage is down to 24v, which is still enough. Another section bypassed, and you're down to 18v, which is still enough.

So rigging the two panels in parallel, a single section bypass in one panel drops the entire array output by 1/2. But if rigged in series, a single section bypass only drops the array output 1/6. Two section bypasses drops the array output 1/3. Both are better than 1/2.

Thus, in general, for shade tolerance, series is always better.


But...(and there's always a "but")...

Rigging the array for a higher voltage, increases the voltage mismatch between array and battery, and the greater the mismatch, the greater the inefficiency, and the less power you harvest. And that will happen constantly. So the voltage mismatch can cause the system to work constantly at a lower efficiency. Bad news.

So now it depends on the charge controller. With a PWM controller, the voltage mismatch matters, so you don't want to rig the array voltage any higher than you have to. So with 18v panels charging a 12v battery, you would always rig the array in parallel.

But with an MPPT controller, the voltage mismatch no longer matters, since the MPPT adjusts the voltage, which PWM can't do. If you've got 18 x 18v panels, you could actually rig them in a single series string of 18 and have an array voltage of 324v. With such a high array voltage, you could have a whole bunch of sections bypassed due to shading, and still have a voltage plenty high to charge a 12v, 24v, 48v or even 96v battery bank.

And that is how you would always do it with an MPPT controller and an array mounted at a fixed location, like a house.


But...

If you rig the entire array in series, if there is any break in the circuit anywhere, the whole array stops working. There's not much chance of that happening to an array mounted to the planet, but an array mounted on a vehicle is a different story. Vibration, tree limbs, etc. can cause a loose or broken connection. So for a mobile installation, you want redundancy...

Say they are actually 18 Vmp with three bypass diodes. Two in series would give a Vmp of 36v, and again, for a 24v battery bank, the array has to stay at 30v or more. A single section bypass on a series string of two panels would drop the output of that string of two panels to 30v. Fine. Another section bypass and the voltage of the series string of two panels, drops to 24v, and that's too low to charge a 24v battery bank...

Different array but same question, 3x Renogy RNG-100D panels (100w, vmp 18.9V, imp 5.29A) and a 12v battery bank: with a PWM controller in series would be wasting efficiency due to a drastic voltage mismatch (54v vs 14.4v), right? But you'd not have to worry much about voltage drop due to shading, whereas in parallel you'd have ~18v and it wouldn't take much shade to drop that too low to charge, but your amperage would be stronger all the time, right? If I've understood that correctly, PWM benefits most from being parallel and MPPT benefits most from being in series? What if I found room/$ for 4 panels, would 2 paralleled strings of 2 panels in series with one another be better?
 

Forum statistics

Threads
185,529
Messages
2,875,555
Members
224,922
Latest member
Randy Towles
Top