Solar panels producing correct Isc, correct Voc produce v. little into a load

[independent]

Hello all.

I have a real problem. I have bought some solar panels and a few of them exhibit this behaviour. Does anyone have any experience with this and do they know what causes it?

That is, the faulty panels produce a rated short circuit current and correct open circuit voltage. However, when connected to a load, they produce very little current, around a fifth of what they are supposed to. I used a Sanwa DMM and then borrowed Fluke to check my meter against. Also, several Solar panels have been used to check the faulty panels against and these ones work all as predicted producing more or less their rated output in the same circumstances. Cables have been checked. Sun is shining--check.

However, the 5x that don't work properly exhibit this strange behaviour. Low output with even the slightest load, in this case, either power resistors or a brand new 12v 10Ah NiMh battery.
20v Voc, 1.4A Isc, but into a battery 200-250mA if lucky. The battery is not full and all other panels that are known working easily produce more current into the battery (at their rated output). Using 30ohm worth of power resistors rates the output of this panel at around 5w as opposed to 25w. Ammeter on the battery tells exactly the same story. Ammeter on power resistors is the same.
With power resistors, graphing the output on another one of the faulty panels produces a very flat line power output curve of very little output. 

Here is something I found that says a similar thing can happen:
http://www.altestore.com/forums/Renewable-Energy/Technical-Discussion-Other/Low-output-problem-from-solar-panels/index.php/topic,1344.msg4744.html#msg4744

That is a panel can have a good short circuit current but actual output can be a fraction the Isc.

[OperaHouse]

I have a pair of 75W photowatt panels like that.  So much for guarantees.  Fortunantly they are tapped at 6V so two of them can make one good panel.  I found open circuit was about a volt and a half low.  I also bought some new underspec 5W panels that were relabeled 3W that act the same way.  More common than you think.  There is no fix. other than putting another in series and running a switcher at power point to get the most of what is left. Try the circuit I recently posted.

[vtpeaknik]

But what's the physical explanation for the behavior of these panels?  Including when the load is a resistor, not a battery?

If these panels are rated for a max-power point of 17V 1.4A (24W), that would correspond to a load of 17/1.4 = 12 ohms.  A load of 30 ohms, on a healthy panel, would accept about 18/30 = 0.6A or 11W.  (I've added one volt there as a WAG for the rise in voltage due to the lesser load.)  The measured output in this case was 5W, so about a half of that.

I have one conjecture: some of the cells have high internal resistance.  That does not affect the open-circuit voltage.  When a short circuit is applied, the section including the high-resistance cell(s) gets bypassed by bypass diode(s).  That still allows the cells in the non-bypassed section(s) to output their full short-circuit current.

If that's the true explanation, then there should still be significant output into a low-enough resistance.  If one half of the panel is bypassed, that should leave about 8 volts, which into a 30-ohm load would yield 0.26A or 2W.  The reported output was actually higher than that.  Perhaps 1/3 of the panel is bypassed, that should yield about 10.5V, or 0.35A (4W) into a 30 ohm load.  Change that resistance to 8 ohms and the output would rise to 1.3A (14W).

In that case, like OH said, it's still OK to use as an "8V" panel (2/3 of its original "12V" nominal output), and putting two such panels in series, feeding a 12V battery via an MPPT controller, would yield about 60% of their promised wattage.  Using a simpler controller, or wiring the 2-in-series pair in parallel with (paralleled singly) healthy panels, it would be more like 50% of their supposed wattage.

[OperaHouse]

20V OC is just not that high and would indicate some bad cells.  Some cells partially short and only produce a portion of what they would normally.  When panel draw exceeds that amount the cell rurns into a diode with associated drop.  ISC generally remains constant up to the knee.  I have it easy as I justg out on an electronic load and direct reading power meter.  Then O just read max power.  You can keep increasing the resistance from 2 to probably 6 ohms and look for the point the current quickly drops.  A power transistor can easily handle 20W with a 3 ohm resistor in series.  hen you could just dial up the current.  Good project, UNO watmeter/load.

[independent]

Can't tap at in the middle of the panels to test for locating bad cells.

Will try lower resistances today as a load. The power resistors are really only to watch the current collapse under any load and monitoring the current going into a battery produces the same result. That is, low current output into the battery. The other known good panels I have available do not behave in this way and do not collapse under a load of 30ohm or 20ohm.

My feeling is this is sorted, bad cells of some kind.

The only other thing I was wondering could possibly produce this effect was if there could be a way for a diode could fail mysteriously resulting in this behaviour, shorting out the panel when under load. I always assumed a Schottky diode was in series to prevent back-current, it is what I've seen on panels I've owned. However, when my friend opened the junction box on his panel yesterday I saw the diode was across the +'ve and -'ve. I don't have access to the j-box on my panels. But if a diode was across the terminals in this manner is there a small chance for failing to happen in this way?

Appreciate the help, thank you.

[OperaHouse]

Actually the fault is somewhat obvious if the panels are left in the sun for quite a while.  Bad cells will get a circular dark spot where it is shorting and heating up.  If you did a spice model of a diode with a current source in parallel with it that would simulate a bad cell.  Many cells have a slight defect.  Only seriously shorted cells cause this problem.  If you run the test I bet you will supply an amp up to about 11=13V.

[vtpeaknik]

Yup, if my hunch is correct then the current will not "collapse under any load" but will be an amp or more with the right resistor.   The reason one panel can't give much current to the battery, under the theory of one bypassed section with bad cells, is that once that section is bypassed the remaining set of cells has an open-circuit voltage that's hardly higher than the battery voltage, truly "collapsing" the current in that case.  But with a suitable resistor for a load (try 8 ohms) you should still see full current.  Another test (not requiring a suitable resistor) would be to connect two such panels in series and then through an ammeter to the battery, you should get the full current (over an amp).

[independent]

Here are the numbers.
Resistance with power resistors. Current as measured by Fluke DMM, Voltage across the output of the panel and VI watts
25w solar panel
5ohm 800mA 2v = 1.6w
10ohm 620mA 4v = 2.48w
20ohm 450mA 8.5v = 3.8w
30ohm 340mA 9.9v = 3.37w
40ohm 290mA  9.0v = 2.6w
50ohm 240mA  11.7v = 2.8w

Those numbers are a little wobbly, ie not perfect, but they are definitely good enough for the purposes here and show the flatness of the power curve of this panel.
A known 5w working panel is putting 360mA into a battery. This 25w panel above puts something like 220-240mA in the same light.

This is the same panel which short circuited 1.4A

[vtpeaknik]

OK.  The wattage curve may be kind of flat, but the V/I curve is not.  More current = less voltage.  But not quite as much current at 5 and 10 ohms as my theory predicted.  Anybody got a better theory?

OTOH, even the healthy panels are only giving you 360 mA into the battery, or 4.3W, if that's at 12 volts.  What battery voltage are these panels supposed to be designed for?  And is this the battery you intended to use these panels with?

To charge a "12V" NiMH battery fully, assuming it is 10 cells, you'd need to bring the voltage up close to 15V.  That may be very slow with these panels, even the healthy ones.  Try two sick ones in series?  Unless you're planning to skip using a controller and just counting on the charge to taper off as the battery gets close to full.  But that's probably not a good idea.

[independent]

360mA with a known good **5** watt panel into the battery. 240mA with the same **25w** panel as above into the same battery. These are tests are for all intents and purposes the SoC is the same with the battery and both panels

[OperaHouse]

Sorry but when I see numbers like this " 5ohm 800mA 2v = 1.6w" all your testing is suspect.  In case it isn't obvious do a little ohms law.
 
This looks like a pretty consistant 400 ma panel. I tend to doubt that 1.4A value.

[vtpeaknik]

Good catch OH.  The numbers for 5 & 10 ohms don't make sense.  E.g., 5 ohms X 0.8A = 4V not 2V.

Also I wonder what does a healthy panel of the same type as the sick ones yields?  Or are all of those type behaving the same?

[OperaHouse]

I believe some of the numbers are good.  A panel with one or two bad cells can be helped with electronic methods.  The current here drops off so quickly any further discussion is moot.

[independent]

The resistance of the power resistors was measured.
The voltages across the resistors was measured but I did get the 5ohm voltage from memory 5 mins after measuring it.
The currents were measured both into power resistors and the battery.
The power output was calculated in a rush. My bad.

However, this 25w panel is not putting 400mA into a battery. The panel is putting 240mA into a battery at the same time a known 5w panel is putting in 360mA into the same battery in exactly the same conditions. What else is there more to say on the matter? The 5,10,20,30,40, and 50 ohm currents were shown to illustrate the wackiness of the situation.

Try it vtpeatnik. Early on to check these panels I put known healthy panels on power resistors to make sure I was seeing what I was seeing. Using the voltage alone and the known values of the resistors to calculate power output is straightforward. The power curve follows a normal one you see everywhere. The volts stay high and more importantly the power resistors get hot.

I don't have a controlled lighting chamber so I've only been able to test these panels when the sun is out and I am able to do it, sometimes in the morning, other times in the afternoon, but always with clear sun and perpendicular to it. Hence variability with some of the numbers. However, that 1.4A number was observed with a Fluke meter as were all of the currents measured with the 25w panel. Early on I calculated the currents from the power resistors with the other panels I have behaving this way as I didn't have the Fluke. But the measurements are consistent enough for me even though they aren't perfectly aligned..
If it is of any help these are not crystalline panels

[OperaHouse]

I have no experience with Harbor Freight type panels.  Your panels are bad and there is no recovery.