Hi All

I can see this topic has been covered before, and most of you know about my huge tree issue...

Situation is: 2 sets of 3 x 280W panels in series, the sets are placed one above each other on roof. Today being a public holiday and bright sunlight, i noticed that as soon as the lower panels were even partly shaded the solar power in to the invertor dropped to almost nothing (43W), even though over 800W of panels still in bright sunlight. I know both sets work well, because I can get well over 1500W when all good.

1. I am concerned that the panels may be back feeding into each other, causing damage later on.
2. Should I install diodes at the connection blocks, to isolate each string?

Panels are supposed to have diodes in already.

An idea, costly, is to have a panel per controller.

Just now, The Terrible Triplett said:

Panels are supposed to have diodes in already.

Yes, but they are bypass diodes. They sit in parallel with the solar cells, reverse biased. They are there so that if a cell goes bad, then the current goes through the diode instead of through the bad cell, causing it to heat up.

Solar panels don't have reverse-flow series diodes as far as I know. I know in the 90s some panels came with a diode in each panel that you can either wire in or not depending on what terminal you chose to use, but I haven't seen that in a long time.

I was sitting here trying to think what happens if you have two cells (or two strings of solar cells) in parallel and one is higher than the other... I have yet to fetch a piece of paper and figure it out... :-)

As I understand it, when using an MPPT, the MPPT should adjust the load (and thereby the panel voltage) so as to obtain the maximum power. This would mean that in a two-string array, I would expect it to optimise for the array with the higher capacity. It should simply push the voltage too high for the shaded string to make anything, but the other string should be fine.

It seems my thinking might have been profoundly mistaken.

http://www.pveducation.org/pvcdrom/modules/bypass-diodes

Might be better to use two separate charge controllers, or try a series diode. Might have to be a beefy diode though :-)

And here is a guy studying the same issue. Basically, what happens if one string makes power and the other doesn't... does the current flow backwards through the other one? Basically, how resistant is a PV cell to reverse current when it is still receiving incoming light, but not enough to push the voltage up.

http://forums.whirlpool.net.au/archive/1525193

If this guy has it right, then reverse-block diodes will make a big difference.

Come to think of it, I have two 20W panels that I can test it with... :-)

Most of the above assumptions were my understanding as well (MPPT controller and panel diodes)... Will look into some reverse block diodes, not even sure where to source, but got to start somewhere.

Ordered some, not sure if they 100% right but from what I remember from electronics, these should be sufficient. I understand there will be a voltage/power drop over these, and one sunny day I will test to see if it's significant. Ordered 5, only need 2, so if they work I may have available for someone else.

Source: http://www.mantech.co.za/Stock.aspx?Query=13M0338and

7 minutes ago, KLEVA said:

Ordered some

That's a silicon diode. 0.6V-0.7V drop, at say 6 amps, so you're looking at a good 4W of dissipated heat :-)

Further research last night indicates that a solar cell behaves like a diode in the dark, and that like most diodes, it has a reverse breakdown voltage at which point you will destroy it. That is why the bypass diodes are so important in long strings, because one shaded cell in a very long string might see very reverse voltages. It also has a reverse leakage current, which is apparently quite low.

At this point an electronic engineer would have wrapped his head around what is going on, but I still have to figure out what that means for reverse current from the batteries. The stackoverflow post I read on the topic seemed to suggest that because they act like diodes in the dark, you should be safe... and then the second paragraph said the series-diode would be essential. So obviously I'm still missing something :-)

Okay, so I did a practical experiment, that is to say, I verified this empirically.

A solar cell behaves like a diode when it is dark, and it passes current in the direction you would normally connect it to a battery. If you reverse it, it passes no current, as you would expect. There is a caveat though: You have to exceed the bias voltage of the cell, just like you have to with a normal silicon diode. Below this voltage, there is a small bit of leakage.

To test this, I used a small 12V panel, which makes around 20V open circuit. I then connected it to a power supply, positive to positive and negative to negative (as you would wire it to a battery) and set the current limit to 100mA. Starting at 12V, I measured leakage current of around 10mA, and this went up to about 30mA at 18V. But then, when I cranked it to 24V (which is above the panel's OC voltage) things changed. The current limiter came on. I cranked it up a little, but at 300mA it was still on and I decided it is sufficient to prove the point.

So, a few things we can learn from this.

1. It would appear that if you have two strings in parallel and one of them is shaded, that string turns into a forward biased diode (or to be exact, a great many forward biased diodes in series) that sucks the life out of the other one.

2. If the panel is connected directly to the battery (no diode), what usually happens is that the battery voltage (12V) for example isn't high enough to bias the solar cells (which might only happen at 20V, for example), and you deal only with a little bit of leakage. Raise the voltage (with another charger for example) and things change a little. Remember that post about the Morningstar that let out the smoke? I'm wondering it this might have had something to do with it.

3. Adding a series Diode might well solve the problem.

So now I'm waiting for @KLEVA to confirm, once he's received his diodes.

I will have to look into my own system as well. I see exactly the same thing. Ideally I should use two MPPTs, but I already own the third or fourth most expensive MPPT (I think only the top end MorningStar and Outback models outrank the BlueSolar 150/70) so the diode approach might be the way to go.

 

EDIT: There is something important that I forgot. Solar modules are constant current devices. If it still gets SOME light it will make a voltage very close to the normal Vmp, it just won't be able to deliver much current. With no significant difference in voltage between the two strings, there should be no reason for power to flow from one string into the other.

 

Sadly today is a cloudy day in Cape Town, so there is no going outside to try it out.

Yay! I'll be the Guinea Pig 

Today we had sun, so I did the experiment again. I hooked my two 20W panels to a 12V battery, and measured the current from each. One provided 240mA and the other provided 300mA. It was late afternoon and I just had the panels flat on the ground... hence the low power figures. I then measured the combined output, which was what I expected: 540mA. I then shaded each panel in turn to see what would happen, and the results suggest that no current is running through the shaded panel. When I cover the 300mA panel, I get the expected 240mA from the other one, and when I cover the other panel, I get 300mA.

In this experiment, we have to note that I have the panels rigged directly to the battery, so I'm pulling them down to a non-biasing voltage. The result is therefore to be expected.

I then did a second experiment. I wired the two panels together without wiring them to the batteries. I put an ampere meter in the positive line so I can see how much current passes from the one panel to the other. I then shaded one of the panels in turn, and got an 80mA feedback through the other panel. In this setup, the panels are running at Voc, so it biases the shaded panel and some current runs through there.

I will have to do the experiment again, this time with an MPPT. Reason for that, The MPPT will attempt to hold the panels at Vmp, and at Vmp it is likely that the shaded panel will be biased forward and some current will go through there. The MPPT will probably then pick a lower voltage such that more of the current from the non-shaded string comes through and to avoid biasing the other one. That is what I would expect at least.

TL;DR: It is not at all clear that parallel strings need a series diode. Not yet.