Measuring PV panel performance


I See Electromagnetic Fields!
Mar 28, 2020
I suppose I could buy a solar panel tester like Will reviewed, but that doesn't feel DIY enough to me.
Besides, I'm a cheapskate, and I try to do things in the most difficult (and expensive if the equipment wasn't on hand) way.

I'm presently working on an underperforming string of Sharp 165W Polysilicon. Underperforming like this 12s string is putting out about 68% the current of an identical paralleled 12s string.

I don't find Voc varying much.
Isc, lowest panel is only putting out 66% of highest.

When measuring Isc, you want to avoid arcing and burning contacts. You can use a suitable switch (visible blade knife switch would be good), or cover the panel or rotate it away from the sun.
In my case, Isc only flagged 2 panels as underperforming. It missed two other underperformers, and one marginal.

Measuring Imp requires a circuit like MPPT controller which finds optimum point on I/V curve. What I did was calculate what resistance would put the panel at this point. Could have used power rating of panel (PTC is about 85% of STC), but since I had observed Vmp on GT inverter while operating in a string and measured Imp of the good string, I used those values instead. I wanted 7 ohms, and a couple space heaters in parallel gave 6.5 ohms so I used that.

Average I(load) of 12 panels was 3.97A and lowest was 2.72A
Power (with 6.5 ohm Rload), one was 80% of average. Four were 70% or less. Seven were 94% and higher.
Recomputing with just the "good" panels, average is 121W (measured between Noon and 2:00 PM, some haze and smoke in California)
Good panel wattage ranged from 93% to 108% of average.
One poor panel was 81% of average, bad were 38%, 41%, 58%, 71% of average power
(This doesn't mean their maximum power was that low, just power when loaded with 6.5 ohms. )

Important note: Isc was not low for the three panels with measured load (into 6.5 ohms) that were 58%, 71%, 81% of average good panels. It was measurement of I(load) and V(load) which detected them.

I have spares I will measure (got late in the day and haze caused them to read low), and swap those in. These are used panels with connectors cut off, so I have to install new MC3.
Then, I will look into warranty. I bought these 18 panels new in about 2004. Sharp's performance warranty is 25 years, 10 years 90% output and 15 more years 80% output. It says ship to authorized dealer (locator doesn't find one within maximum radius). It doesn't say what they would do for me.

For anyone putting up a "new" array of used panels, once you have made a list of V(load) & I(load) (or Isc) for your 13 panels, you can select which four to put in each string. Current from each string will be approximately your lowest measured I(load) of panels in that string. Don't worry so much about Voc or Vmp (Vload) variations; the 4s strings should be close enough for that not to matter. If there is much current variation among panels, matching within strings should increase array power.

Details to follow ...
The array being tested. Originally 9s2p installed around 2004 or 2005, now 12s2p with the column of 6 panels on the left acquired later.
These Sharp NE-Q5E2U panels are paralleled with 24s AstroPower AP120 into one Sunny Boy 5000US GT PV inverter. One one occasion, inverter reported 357 Vmp, Astropower delivered 3.71A Isc, Sharp (panels on left of array) delivered 3.20A Isc, Sharp (panels on right of array) delivered 2.17A Isc.

array IMG_1634.jpg

Tester, measuring Rload. Voltmeter (here set to ohms) is across MC cables for PV input.
Heavy-duty knife switch connects PV to power strip with oil-filled radiators as load.
General-duty knife switch shorts for Isc.

The heavy-duty switch is rated 600VDC 30A, so could be used to test a string of panels.
General-duty switch is not DC rated. It would probably work fine with single panel voltage & current, but I always open circuit with heavy-duty switch.
If I was testing series string, the general duty switch would still be adequate to hold off Voc or carry Isc, but couldn't be expected to interrupt current so heavy duty switch must be used.

A suitable series connection of several heaters could be used as load for this 2000W array, but their resistance would need to be same/similar so no one gets more than 120V applied. Their switches and thermostats are not suitable to interrupt DC, so thermostats are turned all the way up and switches are set without load using ohm meter.

Rload IMG_1639.jpg

Measuring Voc with heavy duty switch open, Isc with both switches closed, V(load) & I(load) with heavy duty switch closed, general duty switch open

Voc IMG_1646.jpg Isc IMG_1647.jpg Vload IMG_1648.jpg
Test results tabulated.
These measurements aren't with one Standard Sun and aren't exactly at maximum power point.
It appears one panel may be marginally at 80% performance, the warranted performance until around 2029.
Four panels are distinctly underperforming, and I will see what warranty is good for.

It is important to observe that that the easy Isc measurement identifies some but not all of the underperformers.
Two panels had good Isc, but I(load) and V(load) were low.

Sharp underproducing.jpg

The PV string with replaced panels is now producing same as the other.
It shows 4.36A (1439W), and has 5 (used) panels installed yesterday + 7 panels from my original purchase.
Other string shows 4.28A (1412W). It has 6 (used) panels installed last year + 6 from original purchase.
AstroPower 24x AP120 shows 5.25A (1733W)
Sunny Boy 5000US shows 4473W, 330Vmp. With 4584W PV input, 97.6% efficient.

string Sharp South current 090621 IMG_1657.jpg string Sharp North current 090621 IMG_1654.jpg Astropower current 090621 IMG_1662.jpg SB 5000US W V 090621 IMG_1671.jpg

The replaced panels are cleaner because I washed and rubbed them with a rag the day they were put up. Others were simply hosed down, some time ago.
One new panel was swapped into location of an existing good one, so longer cables would be available where needed to cross between columns. The replacements had cut cables, under half the normal length.

array 5 panels replaced IMG_1651.jpg

Replacing MC3 connectors on the fat cables (which had been cut) of used Sharp panels was the most difficult part. With cable lube and bruising my fingers, I managed to stuff the 10 wires through their boots.
Wire gauge was too small, but bending double and maybe clipping off one strand it was snug.
The hex die for my ratchet crimper didn't fit, and AWG 8 for the HF hydraulic crimper was best. It would squish out ears of metal, so crimped partially, rotated, crimped again.

MC3 hydraulic crimp IMG_1649.jpg

Eventually I'll look into warranty from Sharp. Instructions are to carry in or ship to an authorized dealer, but none show up within 250 mile max radius of the web page. Concern of course is that shipping could exceed whatever they'll do for me. Warranty doesn't say what that is.

First, I'll see if my near-infrared cameras can see any luminescence when the panels are powered with DC. I'm curious about the two failure signatures (only some underperforming panels had low Isc) and I've read of cell luminescence highlighting portions of panel.

Im using those same exact panels that were warrantied out from Sharp San Jose, I have 40 or so of the NE-Q5EU that i bought for a song along with a few dozen NE-170 and NE-180 panels. the design in the NE-Q5EU was abandoned and the cell structure was changed in the NE-170 and NE-180’s . I got my panels from inside so it was cash and carry

I have 4 poor preformers in the batch of 165 watt panels and zero fails in the 170 and 180 watt panels.
I have heard of this from others in the area. Im sorta local to the bay area so

Just happen to have some shots of a mixed array of 165 watt NE-Q5EU and 170 watt NE-170,s

Should be obvious which is which......

Heard that they are doing swap out between these but please dont quote me on that

You would probably shed a tear as to what i paid so i wont say.....but it was a small fraction of what sellers are asking.

MC-3 connectors the easy way.........small pot of boiling soap....tongs... warm up the boots to 120-140 deg. or so, apply small amount of dish soap and slide it on.....easy as pie


  • 81D5332B-DF52-495D-A5DD-EC87BE1A8E8B.jpeg
    120.4 KB · Views: 7
  • 512C75A8-CA33-41B0-9D73-9A007129A9A1.jpeg
    1.6 MB · Views: 6
more NE-Q5EU 165 watt on one of my workshops......hadent yet got the real iron ridge hold down clamps for this array


  • 37018D97-5A0D-4020-B728-4FF877110852.jpeg
    188 KB · Views: 9
  • D0F210BA-A494-47CF-805F-C8597AF7FE9F.jpeg
    181.5 KB · Views: 10
You would probably shed a tear as to what i paid so i wont say.....but it was a small fraction of what sellers are asking.

I think I paid $4/watt for mine around 2003.
Recently, $0.35/watt for SunPower (I think someone else bought elsewhere for half that)

So Sharp took the panels in on warranty, then liquidated? I can see why they would be cheap!
Bought this batch from a real estate developer who was upgrading a 3.2 kw array to larger format 6 inch polycrystalline panels of unknown make. 66 feet of way old iron ridge solid rail thrown in....haul away as $20/panel including rails but hold down clamps were lost, I search out failed solar systems and overruns from builders who have some left over.
A dozen Suntech STP-280 at $50 ea brand new, left overs from a 30 house project.....drove to Sunol for those
The ones from Sharp were $20 ea no warranty no winin.....have several batches of these as well as some monocrystalline panels in the same frame, cant recall the exact number but Nx190 and Nx200 at $30. ea garranteed to meet specs but warranty

Eleven years ago, I bought 22 used Sharp panels (3600W) for $6000 delivered.
It seemed like a good deal at the time.

I bought the SunPower panels because 50% more power in same area, swapping boosts output.
My existing AstroPower and Sharp could go on a roof, if I can ever figure out what to do with all the watts.

I was an early adopter of GT PV, didn't realize it would drop so much and utility rates wouldn't rise terribly fast. I was betting on inflation. Which has occurred, just not for PV.
Some other things I've been better about buying used up front, like most of my electronic equipment (couldn't justify retail price of of vector network analyzers, spectrum analyzers, and the like.) But there are cheaper brands available now.
The panels on the shipping containers were Sharp warranty outs,

The NE-Q5EU panels on my workshop were bought from the real estate developer.

I bought two Strings of panels Chinese 6” 295 watt polycrystalline from Santa Rosa area that were a “failed system” for $40 / panel on the ”good string” if I would haul away the ”bad string“ as well.

Testing showed two burned out MC-4 connectors.....
Yes they are getting more effecient, The Suntech array is a 5 percent more effecient than the Sharp panels, but they are too big for me to just flip around by myself. Almost took flying lessons with one of those 39 by 68 inch monsters, made a good sail....The Suntech,s are also 72 cell but the other big panels were 60 cell and would not fit in well, sold them for a very good profit. System owner did not look into fail mode enough, scrapped the panels for failed connectors....
My panel load tester........

A very used Sun automotive Alternator load tester which has 8” analog voltmeter and ammeters as well as a 300 amp carbon pile load tester. Read the meters and use a calculator........too old school for digital opened it up and changed a now has 6 voltage ranges as well as an ammeter reversing switch as it originally would display -30 to + 120 amps (don't remember the exact range but close) with a offset zero meter........its also a very useful monitor for my solar system. The ammeter uses a 50 mv. shunt
using various different shunts i get several smaller amp ranges.

8 inch analogue meters, easily seen across the room...the thing is bigger than the two classic’s

bought for $20. with a broken battery post shunt....
Last edited:
And some freebees, brand new, but scratched on installation First Solar Cadmium Telluride 75 watt panels 24” by 48” 9% efficiency 72 VOC


  • 1990C6AF-906F-4F3C-A724-074E632C3B52.jpeg
    67.4 KB · Views: 9
Put these into the simplest solar system.... 4 of these connected in parallel to 4 two tube 48 inch T-8 florescent light controllers, no batteries, no inverter, no switches, nada.....

no not led’s....florescent fixtures.........with T-8 florescent tubes

Sylvania switch mode ballasts rated 65 to 277 volts AC...and in tiny print 65 to 165 volts DC

Super simple, sun comes up, lights come on, sun sets, time to quit...

works in the fog.....

my neighbor asked me if i'm farming solar panels
Last edited:
Way back in the late ‘65-66 at G-E we were obtaining Arco Solar SM50’s and SM-55’s for the smoking deal of $950 each.....G-E was the communications contractor for ARCO both the solar division and the refineries located in the LA area, we used then on our communication sites on mountain tops in the greater LA basin, pics of a few of those that survived an ice shedding from the antenna tower, I still have them and they are at near full rated output.....


  • 17A21C7B-1921-43FC-9935-6DA3516A8E50.jpeg
    2.4 MB · Views: 10

For my recent off-grid project I picked up 6 x Longi 370W panels for 25c/W delivered.

I just load test them with the system and according to Solcast forecasts they are pumping out at least as much as they should at this time of year and for their orientation.
photoluminescence and electroluminescence for testing of panels?

I wasn't clear if the emissions would include shorter wavelength IR near visible, or only longer wavelength.

I have a CMOS camera which is able to show images under visible and IR illumination (but not emission from warm bodies such as motion sensors can detect.) I tried to use it to test electroluminescence of the PV panels I removed from my array.

Could have used a bench power supply, but the ones I have which could drive to Voc of this panel are only 1A.

So instead, I wired two PV panels in series and aimed them at the sun. They were wired through the same switch previously used for Voc/Isc/Vload/Iload testing, but this time the PV panel under test was used as a load, located in a dark room.
The two panels in series will never deliver more current than Isc. I expected to see near Voc and Isc (of the two illuminated panels.) It delivered 3.0A and about 55V, a bit higher than the 40V I expected. The panel being tested was the worst one, # 0064.

With room lights off, camera just showed noise. Image didn't change with current through the panel or not. So apparently not near infrared. No results obtained.

Maybe somebody with a FLIR camera can try the same?
Also get images of operating panels in sunlight. Maybe also open-circuit and short circuit?
It is likely FLIR has a visible-light filter so daylight imaging could work.
I was able to borrow an IR camera from a colleague, so I did some tests. Individual panels in a dark room, with current through them. PV array in the sun, with and without current flowing.

I considered using a bench supply, but instead connected two more PV panels of same model in series and laid them out in the sun. That provide up to Isc (given enough sunlight), and will drive the device under test to Voc, or something like that. Couple of switches, DMM, clamp ammeter, and I could measure Voc, Isc, Vload & Iload. Similar setup to how I tested the individual panels, but this time 2s PV string fed the switches and the DUT panel in dark room took the place of a resistive load.

Here are the images from several bad and one good panel.
I should have pre-conditioned all indoors, but instead I swapped the around with the panels I laid outside to serve as a supply. That biased temperature of some higher.

20210921-130627.jpg 20210921-132606.jpg 20210921-133720.jpg 20210921-141239.jpg 20210921-142047.jpg 20210921-143511.jpg

PV array open-circuit (with one bad panel reinstalled), followed by under load, which make the bad panel jump out.
Other panels in the array, not identified as bad by electrical testing, show similar signature of some cells hot.

20210921-150243.jpg 20210921-152045.jpg

(to be continued, limit of 10 attached images)


  • 20210921-135204.jpg
    149.1 KB · Views: 1
  • 20210921-132635.jpg
    160.3 KB · Views: 1
Last edited: