Voc on used panels, MPPT Voc limit etc, temperature coefficient

[CaliSunHarvester]

Posting this as I was hijacking someone's thread where the temperature coefficient was mentioned.

In my ~2 week old installation I apparently run the risk of sending too high voltage to the charge controller. I consider myself very lucky reading the other thread and catching the temperature issue there (it was barely mentioned until I hijacked it).

Inverter charger is a Sungoldpower 10kW with Voc.max of 500V

Panels are REC 260 with Voc of 37.8V

I put 13 panels on string.. 13 x 37.8 = 491.4V
Which is under 500V

But now I learn that these panels' Voc increases by 0.25% for every degree Celsius dropped.

Inverter hasn't thrown a fault code in 2 weeks of operation.

 

[zanydroid]

I think something weird is going on with either the panels or the way you are testing. Maybe you got a lot that were for the most part 1/3 burned out. And that would mean you got bad panels but it saved your string.

Does the inverter graph the operating voltage over time? Assuming no shade that would show the Vmpp of the string (as long as it is drawing power, otherwise it will go to Voc)

Do you have or are willing to buy a panel tester that can check the output specs of spare panel in current lighting conditions? I don’t have one so I can’t recommend one but there’s probably thread.

 

[CaliSunHarvester]

What is the overhead of removing one panel?

Nothing shaded.

The panels are installed on a low slope metal overhang roof. The house itself has a flat roof and it's only 8ft high. So rather easy to get to the panels. I don't like panels on roofs. Esp not on flat ones.

Is that voltage with MPPT active?

Is that with no cells shaded? That voltage is pretty weird, you have 60 cell panels there, Voc should not be that low. Unless somehow somebody shorted one bypass diode on

The measurements were taken on disconnected panels. I measured the 260V at the breaker.
A bit later, as the sun was coming out, I measured 400V at the breaker.

Early in the am, not much light, I measured on the single panel:

38.1V while 10 degrees Celsius

A bit later:

35.5V while 15 degrees Celsius

I put this single panel perfectly facing the rising sun. And almost vertical.

The string on the overhang is almost flat, facing south.

So, definitely old panels can deliver "full" Voc and in fact above the 25C rating

 

[v_green57]

Posting this as I was hijacking someone's thread where the temperature coefficient was mentioned.

In my ~2 week old installation I apparently run the risk of sending too high voltage to the charge controller. I consider myself very lucky reading the other thread and catching the temperature issue there (it was barely mentioned until I hijacked it).

Inverter charger is a Sungoldpower 10kW with Voc.max of 500V

Panels are REC 260 with Voc of 37.8V

I put 13 panels on string.. 13 x 37.8 = 491.4V
Which is under 500V

But now I learn that these panels' Voc increases by 0.25% for every degree Celsius dropped.

Inverter hasn't thrown a fault code in 2 weeks of operation.

VOC is exactly that - voltage OPEN circuit.

As soon as you connect a panel or array to anything that draws current, you are no longer meeting that criteria.

I experimented with this some years ago and found that VOC can be knocked down significantly with a surprisingly small current and that the effect is not linear.

The reason that you haven't had a problem is likely due to the 50 - 100W "idle" power draw that almost all cheap Chinese AIO's exhibit. Part of that will likely always come from the panels during daylight hours, even if your battery bank is fully charged. That power suck keeps the panels from achieving VOC, as the "open" criteria never happens.

Sidebar and opinion: it has been my experience that if you want long life and trouble-free operation of any electrical or mechanical system, you should design it to have significant "slack" between components. Connecting a 500V AIO to an array that *could* source 491V according to published specs does not meet that description. I would personally want at least a 20V "worst case" difference there. Running anything on a knife edge is never a good idea.

 

[CaliSunHarvester]

Around 11:15am here. Good sun (for early January).

Temperature 21C -- in the shade
Single panel almost vertical, facing the sun:
35.5V
Same panel very low slope:
34.5V
Same panel almost flat:
34.28V (see photo)

String of 13 panels, almost flat, breaker turned off. Measured voltage at breaker. This should be Voc: 390V (30V per panel)

Waited a bit. Turned breaker back on.
Vmpp fluctuates between 295V and 305V.

Now the question, why does a single panel have 34.28V and in a string of 13, it's only 30V?

It's about 130' of wire (round trip). 10 gauge.
4.5Amps per AIO. Voltage calculator says 0.31% drop = 0.1V per panel only. But I'm losing more than 4V.

No other hardware between the panels and the breaker except 52 of those MC4 connectors (26 pairs).

Does it mean that I made poor MC4s?

Edit: added the photos.
Also the second photo shows brown marks (burn marks??) on the back of the single panel. Not all panels have these. Do they matter?

 

[CaliSunHarvester]

I would personally want at least a 20V "worst case" difference there. Running anything on a knife edge is never a good idea.

I agree. I'm deciding whether to go with 11 panels (fully save even with unlikely bright light at freezing temperature) or 12 panels (probably safe). The AIO does pull ~100W at all time.

Right now, I try to see why I receive only 30V per panel at the breaker when a single panel without 100'+ of wire delivers 34V. This is probably the reason why I never saw a over-voltage fault code.

These panels are rated 260W.. 13*260W = 3380W. AIO actually puts 1350W in the batteries/ loads, sigh.

 

[zanydroid]

I experimented with this some years ago and found that VOC can be knocked down significantly with a surprisingly small current and that the effect is not linear.

When you say the effect is not linear/surprising, do you mean it's counterintuitive to how solar panels are supposed to behave at low illumination? (IE the Voltage - current curve at the irradiance). I don't actually have a good mental picture of the graph at a very early morning situation.

WRT the voltage you see in Open vs non-Open when connected to an MPPT. It's possible that you are seeing average V below the danger zone but instantaneously it jumps above that as the circuit pulses to open. So you still don't want to trust zero damage happening, based on an average reading that is lower than what the calculations say you can see...

String of 13 panels, almost flat, breaker turned off. Measured voltage at breaker. This should be Voc: 390V (30V per panel)

Is the voltage what you expect with disconnect off?

Is there a DC SPD in the circuit? Maybe it's leaking a little.

Burn marks - maybe this is causing the bypass, because that segment of the module is broken. I don't know if that means bypass diode always activates. Quite possibly. Or, maybe somebody replaced the bypass diode with a jumper because the section was busted.

4.5Amps per AIO. Voltage calculator says 0.31% drop = 0.1V per panel only. But I'm losing more than 4V.

0V drop when open circuit.

Now the question, why does a single panel have 34.28V and in a string of 13, it's only 30V?

390 / (34.28 * 13) = 87% of expected
55V gap. Is there 1.66 completely bypassed panels, or 5 segments that are bypassed?

Segment (not sure if this is the right terminology) is 1/3 of the panel. So 1/3 of the voltage if they're being bypassed.

(I'm assuming these are full cut cells)

Do you have panel level monitoring?

How important is the missing capacity to you vs splitting the array to binary search for where the problem is? IE measure half the array at a time by jumpering across down to 6s / 7s segments. Or disassemble the array and measure voltage one panel at a time. Did you do Quality Control the panels before installing them?

 

[zanydroid]

Does it mean that I made poor MC4s?

I think your intuition about electricity is off here. If you made poor MC4s to the extent that you are getting 4V of drop across one. That is 20W of heat at 5A. That would quite likely melt the MC4 quite quickly.

Did you use genuine MC4s? Were the panels terminated with those? I doubt it would cause 4V, I'm just making sure you followed code.

 

[CaliSunHarvester]

Is the voltage what you expect with disconnect off?

Yes, the 390V is kind of expected for Voc of the entire string - it was 400V a few hours earlier in the day when it was cooler.

Is there a DC SPD in the circuit? Maybe it's leaking a little

No SPD or other hardware yet. I have that planned. Only 10AWG wire and MC4 connectors.

Did you use genuine MC4s? Were the panels terminated with those? I doubt it would cause 4V, I'm just making sure you followed code.

The REC panels came with MC4 connectors. I only made 4 crimps myself.. the end of my string (2 connectors, + and -) and then a 3 foot extension piece for the + cable (2 more connectors). I have made these connectors before for a different installation, a 3s4p array, same panels, and Voc for that array is indeed almost 3*37.8V. It's 108V which is 3*36V.. pretty close. And it has a combiner box which explains the 1.8V drop maybe.

I don't know if they are *genuine* but they look the same as the ones on the panels, they connect and disconnect easily.

binary search for where the problem is? [..] Did you do Quality Control the panels before installing them?

binary search.. memories from CS come back. No, I am not really looking to do that. My quality control was to measure Voc on the top 2 panels in a stack of 30 and the bottom one. They were quite dirty. I don't recall burn marks on the installed ones but it is possible. From washing them, I remember that 4 or 5 of the 30 had such burn marks.

How about using one of these laser thermometers and measure all connections, looking for hot spots?

Do you have panel level monitoring

No panel level data.. in regards to those brown spots.. I don't remember seeing them on any of the panels that are mounted yet. So far, one string of 13 panels is mounted and 13 others are waiting to get put up. Reducing the string to 11 or 12 is just a matter of re-assigning 1 or 2 to the 2nd string and then add 9 instead of 13 to complete the 2nd string.

I appreciate your calculations / thoughts about the burn marks and bypassed segments. Only one issue. The single panel with all the burn marks produces the highest Voc! Or so it seems. I did not measure the installed panels. But I took another one from my stack - it has no burn marks.

Left one in picture has the burn marks and measures consistently 1V higher than the "clean" one on the right. I measured both back and forth and the numbers were like this with 0.2V variation over 5 minutes:

Left 35.96V with burn marks on back
Right 34.73V "impeccable"

 

[zanydroid]

Yes, the 390V is kind of expected for Voc of the entire string - it was 400V a few hours earlier in the day when it was cooler.

Earlier in the thread you have 491V calculated. Why is 390V expected? What is the difference?

If you test the open circuit voltage (with all loads disconnected) in full sun and it is 390V then you are probably OK relative to your 500V. The voltage should only go down with load, but I would try to confirm that with more people on the forum.

As for finding the problem panels (if any). That would let you harvest more power. For instance if you have up to 5 partially busted panels, you could swap out some of them to get your string voltage higher.

If you really do have busted panels, you can actually use that to max out your string better. If there is a panel with 1/3 bypassed, it will have 2/3 VOC, which lets you subtract 10V from a full 13s. Effectively a 12.66s string instead of going down to 12s. Presuming of course that the damage is nicely contained, EG the other 66% of the module performs to spec.

I don't know if they are *genuine* but they look the same as the ones on the panels, they connect and disconnect easily.

I'm not an expert, but the copies and genuine ones I've seen, the copies have very similar plastic mold but genuine ones had some extra printing on them. If you bought the genuine ones yourself from a supplier (and it was marked as such), then there's not much you can do beyond that. When I made jumpers I did some resistance testing a few times but you can't easily do that against connectors on a solar panel.

binary search.. memories from CS come back.

I'm a software engineer so that's my goto approach if I have a problem that looks like this.

How about using one of these laser thermometers and measure all connections, looking for hot spots?

Could be a good idea.

Left one in picture has the burn marks and measures consistently 1V higher than the "clean" one on the right. I measured both back and forth and the numbers were like this with 0.2V variation over 5 minutes:

Not sure what to expect (the only thing I can think of here is to compare that with the tolerance on the spec sheet. on my bought as new panels it actually came with a electrical test report for every individual panel on the pallet. cool but I didn't really care as long as it was close enough). I think using a decent quality panel tester would test both the VOC and V_mpp/P_mpp, I remember reading some posts about how open circuit will not catch all cell faults, & you have to put load through them.

 

[v_green57]

"When you say the effect is not linear/surprising, do you mean it's counterintuitive to how solar panels are supposed to behave at low illumination?"
=======================================
No, that's not what I was getting at.

How it measured out was:

A small current draw dropped the VOC by X. Drawing twice the amt. of current did not result in 2X the voltage drop. I don't remember what the actual curve would look like. Testing was done (as always) in full sun with an adjusted panel angle perpendicular to it (max. radiation).

Now, I did this test using a pile of power resistors I had lying about. Those are a simple straight-line resistive load. A MPPT charge controller is essentially a constantly changing load while it fishes around looking for the MPPT point.

Anyway, the point still is that AIO's have a pretty large idle current which I think would tend to never let VOC actually be present. But I would NEVER EVER count on that.

Live by the "20V" rule and be safe.

 

[zanydroid]

"When you say the effect is not linear/surprising, do you mean it's counterintuitive to how solar panels are supposed to behave at low illumination?"
=======================================
No, that's not what I was getting at.

How it measured out was:

A small current draw dropped the VOC by X. Drawing twice the amt. of current did not result in 2X the voltage drop. I don't remember what the actual curve would look like. Testing was done (as always) in full sun with an adjusted panel angle perpendicular to it (max. radiation).

Cool, thanks for clarifying.

Now, I did this test using a pile of power resistors I had lying about. Those are a simple straight-line resistive load. A MPPT charge controller is essentially a constantly changing load while it fishes around looking for the MPPT point.

There's also an issue with the fact that it is likely approximating that load with switching rather than a linear circuit. When switched off the voltage will rise. If you see here (Diagram below Step 12), if the transistors are turned off in the right way (and ignoring the backflow prevention / power supply that's always on to power the electronics in standby) the panels will see a 100kOhm load.

1kW Arduino MPPT Solar Charge Controller (ESP32 + WiFi)

1kW Arduino MPPT Solar Charge Controller (ESP32 + WiFi): Build a 1kW WiFi MPPT Solar Charge Controller, equipped with phone app datalogging telemetry! (Android & IoS) It is compatible with 80V 30A solar panel setups and all battery chemistries up to 50V. The project is based on an Arduino ESP32...

www.instructables.com

Anyway, the point still is that AIO's have a pretty large idle current which I think would tend to never let VOC actually be present. But I would NEVER EVER count on that.

Live by the "20V" rule and be safe.

OP is also measuring open circuit with disconnect opened, so no load for sure. And no worries about what the instantaneous vs averaged voltage would be. There are much fewer assumptions here with this experimental approach, and it does not need to worry about WTF is going on with the AIO.

 

[42OhmsPA]

I would rewire to 6s2p or run another set of wires and run 7s string to mppt 1 and 6s to the other.

Not sure if it's helpful but I attached a chart of my startup this morning using the same inverter from SRNE.
MPPT 1 is 7s2p 287 Voc (weeks old 370W)
MPPT 2 is 9s2p 362.6 Voc (years old 315w)

 

[RCinFLA]

Most PV MPPT controllers initially test for Voc before they begin their initial MPPT search. To speed up the search they start the search below Vmp at about 70-75% of what they measured for Voc.

They often will not start up their SCC DC-to-DC converter if Voc is outside their PV input max or min voltage limits.

 

[sunshine_eggo]

Most PV MPPT controllers initially test for Voc before they begin their initial MPPT search. To speed up the search they start the search below Vmp at about 70-75% of what they measured for Voc.

They often will not start up their SCC DC-to-DC converter if Voc is outside their PV input max or min voltage limits.

Outback FM-80 allow the user to customize this. It's defaults were right around those levels.

 

[zanydroid]

I would rewire to 6s2p or run another set of wires and run 7s string to mppt 1 and 6s to the other.

Not sure if it's helpful but I attached a chart of my startup this morning using the same inverter from SRNE.
MPPT 1 is 7s2p 287 Voc (weeks old 370W)
MPPT 2 is 9s2p 362.6 Voc (years old 315w)

Would you suggest the rewire "blindly" with the 13s configuration giving a strange Voc, or try to isolate first if there are some very undervoltage panels? There is a discrepancy of at least 50V across the 13s.

 

[42OhmsPA]

Would you suggest the rewire "blindly" with the 13s configuration giving a strange Voc, or try to isolate first if there are some very undervoltage panels? There is a discrepancy of at least 50V across the 13s.

My biggest reasons for suggesting rewire are the potential to easily exceed the rating in cold temperatures if panels perform as rated. 38.1V measured at 10C is under 5V to the 'limit'.

The other reason is 425V is the top of the working voltage range.

I'm also interested in finding the root cause.
I suppose removing 3 panels from the series and measuring 10s voltage would be simple enough?

 

[CaliSunHarvester]

Hello, thank you to everyone who weighed in. Tbh, some went over my head at this point. Maybe I will understand those later.

I removed the last panel which was as easy as disconnecting and reconnecting 2 of those MC4s. Started to put up 3 more panels to get the 2nd string started. I plan for 2 strings of 12 panels each. The highest Voc that I measured was 38.1.. 12 * 38.1 = 457.2 is well below 500.

Even if the temperature would drop to freezing with some sunlight, that would be 10 more degrees = 2.5% increase..

457.2V * 1.025 = 468.63V

Going by the manufacturer's spec.. (see picture 37.8V) and assume a temperature of 0C:

25C drop = 25 * 0.25% = 6.25% increase
37.8V * 12 * 1.0625 = 481.95V



I suspect that nothing would ever have happened with the 13s, BUT I like to sleep well at night and 12s will be enough and no, I'm not motivated to measure each of the installed panels for their Vocs.

I bought those 30 panels for $1998 tax and shipping included.. that's 25c per Watt.. it's OK if they don't all perform perfectly.

Once I have those 24 panels up, initially I will have each string in a separate MPPT.

Later I will buy another ~30 panels and at that point, the initial strings will be 12s2p in one MPPT and some similar configuration for the other MPPT.

I could end up with 48 total such panels = 12,480W but realistically, they will probably put 5.5kW in the two MPPTs.

So, I will be overpaneling, but I feel I need it for those winter days.

I have learnt a very valuable lesson in the last 20 hours about the temperature impact on Voc. I feel quite lucky to have read @LLLL's thread.


Edit:
I am in Santa Cruz, California.
The lowest temperature since 1893 was 19F = -7C
32C drop = 32 * 0.25% = 8% increase
37.8V * 12 * 1.08 = 489.9V

Pretty sure it was 2am when the temperature read 19F.

 

[CaliSunHarvester]

So when I thought it's all good, it gets weirder.
Yesterday I removed one panel from the first string, it has 12 panels now. Vmpp dropped only slightly, it was still around 288V (yesterday afternoon).

Set up the 2nd string with 5 panels yesterday too. They will be on separate MPPTs.

This morning, I am curious about the Voc numbers. Turned off the breaker and measured at the breaker.

1st string.. I expect something over 300V. It reads -3.5V. Yes, minus.

2nd string shows 180V for Voc which is 36 per panel and a very reasonable number.

I guess I have to pick up all 12 panels on the 1st string and inspect each?

Not going to turn on the breaker for #1 at this time!

After that, I just walked up to our Prius which was last driven 20 hours ago and the fan was running. Mind you, it's in the shade. Thought someone forgot to turn off the engine.. opened door, no - the engine is off. Fan shut down in that moment too.

If my cell phone wouldn't work, I'd believe in one of those events that peppers warn of. Solar storm or so.

 

[MisterSandals]

Would you suggest the rewire "blindly" with the 13s configuration giving a strange Voc, or try to isolate first if there are some very undervoltage panels? There is a discrepancy of at least 50V across the 13s.

I wasn't sure who to quote but this seems best for my Voc thoughts/ramblings.

When measuring a single panel Voc, its completely isolated and easy to match specifications.

Once you start introducing other variables, like a lot of other panels in series, you are introducing resistance, possibly proportional to the length of the string. I suspect that with this added resistance, you aren't really measuring Voc but rather a lightly loaded set of panels.

I wonder if this means "some" current should be detected within the string?