Converting 50kW system from micro inverters to 8x SMA inverters

[nerdralph]

Shortly after sunup on a clear, windless and low dewpoint winter morning it's entirely possible to have the average cell temp. below the surrounding ambient air temp. by several degrees C.

That's a red herring. I specifically said "in bright sun", not at the break of dawn. For a given temperature, Voc decreases with decreased insolation.

[J.P.M.]

A few weeks ago I mentioned how on a cold winter day in bright sun, your panels will be at least 20C warmer than the air.

https://www.solarpaneltalk.com/forum/solar-panels-for-home/solar-panel-system-equipment/433763-temperature-correction-for-maximum-voltage-doesn-t-make-sense-in-the-real-world

I'd be careful about making assumptions and statements like that, especially if operating close to allowable string voltage limits.

The difference in temperature between an array cell and the ambient air depends primarily on the POA irradiance on the array, the magnitude of the wind vector, the dew point temp. and a couple of other things having to do with something called the effective radiant sky temperature.

Shortly after sunup on a clear, windless and low dewpoint winter morning it's entirely possible to have the average cell temp. below the surrounding ambient air temp. by several degrees C.

Reason: Primarily because of low(er) radiant sky temps. under those conditions, thermal radiation heat transfer from the panels to the sky under those conditions may well be greater than the slight amount of POA irradiance on the array under those conditions.

It's all about doing an energy (heat) balance on the array.

That's the physics of it.

[nerdralph]

The -12.333 C is the lowest temperature recorded by my weather station in January of this year, that is located next to the panels.

I don't know what the odds are that all of the following will be true:

-12C
Daylight
No clouds and bright sun
Inverter not already producing power


But to be safe, I should probably just do 14 panels per string and then have the last 4 panels in a string of their own.

A few weeks ago I mentioned how on a cold winter day in bright sun, your panels will be at least 20C warmer than the air.

https://www.solarpaneltalk.com/forum/solar-panels-for-home/solar-panel-system-equipment/433763-temperature-correction-for-maximum-voltage-doesn-t-make-sense-in-the-real-world

[oregon_phil]

bcroe in an earlier post commented "Here strings with 720 cells (12 panels of 60) run MPPT at 360V in summer. But in cold Feb, MPPT can hit 420V. That means
before inverter turn on, Voc can be 525V. "

SMA MPPT range is 100 to 550VDC. With 14 panels you could potentially start up at 601VDC on a cold winter day which would put you outside of MPPT range. 12 panels would put you into MPPT range like Bruce's setup. My SMA inverter with OptiTrac/Shadefix checks for global MPP every 5 minutes to make sure it isn't in a local MPP rut. My longest string is 10 panels so I don't have a MPPT max voltage problem. But lets say your inverter starts up with 14 panels at 601VDC. I'm guessing the inverter picks some power value based on voltage(?) and just stays there. Since voltage never goes below 550VDC, MPPT never kicks in (endless do loop). I'm not an expert in this so others can fill in the gaps.

You are doing a lot of research! Good luck!

[pclausen]

Appreciate the feedback. So I found the full specs of the panels along with this calculator, and got the following results:



The -12.333 C is the lowest temperature recorded by my weather station in January of this year, that is located next to the panels.

I don't know what the odds are that all of the following will be true:

-12C
Daylight
No clouds and bright sun
Inverter not already producing power


But to be safe, I should probably just do 14 panels per string and then have the last 4 panels in a string of their own.

3 x 15 panels are the above Talesun 270 Watt panels (6 of them are actually Talesun 275 Watt panels, but close enough) The newly added bottom row is 15 white label 250 Watt panels, so I don't want to mix those with the Talesun panels as it would drag those strings down to 250 Watts per panel. So my thinking is that if I don't go 4x15 strings, to do the following:

3 x 14 Talesun 270 watt strings
1 x 14 white label 250 watt string
1 x 4 string with 3x Talesun 270 watt and 1x white label 250 watt

[oregon_phil]

I found a spec sheet that shows your panel's temperature coefficient Voc =-0.31%/C.

My panel's reference air temperature is +25C. My guess is that you chose your lowest air temperature to be -11C (12.2F). The difference from between +25C to -11C = -36C.

Voc change would be = -36C * (-0.31%/C) = 11.16%

11.16% * 38.5 Volts = 4.3Volts

Voc at -11C = 38.5 + 4.3 = 42.8 Volts

15 string voltage = 42.8 * 15 = 642 Volts at -11C

Feel free to check my math. I haven't had my coffee yet.

There are quite a few Voc max calculators on line. All of them rely on your estimated lowest temperature among other variables.

Added: SMA 7.7 MPP voltage range is 100 to 550 Vdc.

[pclausen]

I was able to wring some more details out of the SMA designer and it shows the following for Voc @ Tmin:



So 642 V. Where does that figure come from? When I look at the label of those panels, I see this:



A bit of Googling reveals that the Temp. coefficient of Voc = -0.36%/C at STC.

Above panel has a Voc of 38.5V, so 38.5 x -0.36%/C = -0.2386V/C. Given the NOCT of 45C, at 0C that would in increase of 10.737 volts. So Voc @0C would be 49.237V.

But 49.237V x 15 panels would be a whopping 738.555 Volts where the SMA calculator is coming up with 642V, which is 42.8 Volts per panel.

What am I missing? It sounds like I should limit myself to 14 panels per string, but I'd like to better understand why.

[pclausen]

bcroe Good point about open voltage total. I added that to my summary:



So I come in a little below the 600V maximum input voltage spec of the SMA SB7.7 on my array with 15 panels per string.

That said, when using the SMA designer, it does give me the following error when I got from 14 to 15 panels in each string:



Not sure why since I'm below the 600V rating.

As an alternative, I suppose I could do 6 strings of 10 panels each and use all 3 inputs on each SB 7.7 inverter instead of just 2.

[pclausen]

How did you upload those images, did you use the "Upload attachments" or "Upload URL" buttons, or something else?

I did the following:



Click the picture icon along the bottom, paste in my url and then uncheck the "Retrieve remote file and reference locally".

[bcroe]

Here strings with 720 cells (12 panels of 60) run MPPT at 360V
in summer. But in cold Feb, MPPT can hit 420V. That means
before inverter turn on, Voc can be 525V. Inverter specs 600V
absolute max, I see that as a point where catastrophic inverter
failure could happen. Some run one more panel in the string,
but I would like to have some margain to failure. Things like
power lines and lightning strikes (yes has happened in Jan)
might add some more voltage.

Here a string is panels connected in series, connecting another
400V string in series causes the voltages to add up way beyond
the practical limits. The combiner box here handles 2 inverters
with 6 strings each, has a fuse for each string. You would need
more than that. The combiner box is certainly a convenient
starting point for any troubleshooting.

Any cracked panels I would consider probable failure points
before long, and keep them apart from the rest of the system.
Bruce Roe

[sdold]

How did you upload those images, did you use the "Upload attachments" or "Upload URL" buttons, or something else?

[pclausen]

I won't have any strings in parallel. All will be in series. Here's a summary of how many strings and the number of panels in each:



That last string with only 4 panels are the ones that are shaded in the AM and the glass is cracked in the panels. Maybe I would be better off keeping micro inverters on those 4 panels vs. turning them into a small string?

But yeah, I suppose having in line string fuses at the arrays would be a service issue, especially for the roof mounted arrays. They do make nice combiner boxes such as these:



But I have not seen something similar for individual string fuses being combined into a single box.

[bcroe]

If multiple strings are connected in parallel to an inverter, they need individual
fuses. Some inverters have fuse positions to do that. Or bring them together
in a combiner box with fuses, then make a heavy wire run to the inverter. In
line string fuses work, but could be a service problem to locate.

The new software will not allow me to show a combiner box. Bruce Roe

[pclausen]

Yeah, the SMA manual stresses to always disconnect the AC power before disconnecting the DC disconnect, as, like you say, that could draw a serious arc.

In keeping with KISS, perhaps I should think about using these where each string exit the array:

[bcroe]

An outside AC disconnect can work, that is what I have used to shut
things down (an extremely rare event). I hate to open an operating
DC switch, as the 400V 21A current draws such an arc. It will still
leave the DC disconnect inside, hope no inverter smokes.

String inverters KISS, some of the things going here could not be
done any other way. Bruce Roe