Looking for some unbiased advice

[sensij]

I see what you mean: at +45C surface temperature 31Vmp panel with -120 mV/C coefficient will produce only 28.6V which is only enough for charging without room for equalization. Hope OP gets himself MPPT controller eventually and all this would become irrelevant . Short of that 72 cell panels would be way to go.

Most mppt controllers don't have a boost function, so you'd have the same problem with them.

My Canadian Solar 290's have a 32.1 Vmp. Mid-day mpp voltage in summertime routinely drops below 27 V. They would do a very poor job of taking care of a 24 V battery.

[max2k]

Most mppt controllers don't have a boost function, so you'd have the same problem with them.

My Canadian Solar 290's have a 32.1 Vmp. Mid-day mpp voltage in summertime routinely drops below 27 V. They would do a very poor job of taking care of a 24 V battery.

something is off with the numbers: if we take your panels are 40C above 25C STC 5.1V x 100 /30V/40C = 0.425%/C which is about x1.5 times worse than its rating. OTOH may be they get heated up all the way to 85C.

but then OP could connect panels in series without losing efficiency provided of course it can take at least 2x panels Voc at the coldest temp plus some (>100V max). I like MPPT controllers for this flexibility.

[sensij]

something is off with the numbers: if we take your panels are 40C above 25C STC 5.1V x 100 /30V/40C = 0.425%/C which is about x1.5 times worse than its rating. OTOH may be they get heated up all the way to 85C.

but then OP could connect panels in series without losing efficiency provided of course it can take at least 2x panels Voc at the coldest temp plus some. I like MPPT controllers for this flexibility.

Something is off with your understanding, I think.

Coefficient of Voc is 0.3%/deg C.

Voc = 39.3 V
0.3% * 39.3 = .118 V / Deg C.
5.1 V / .118 = 43.25 deg C

43.25 + 25 deg = 68 deg C.

Vmp won't exactly follow Voc, but close enough.

For what its worth, I suspect the optimizers are adding some heat, so my cell temps might be a bit higher than models would estimate, in the 60's rather than the 50's, more or less canceling out any outperformance they achieve by eliminating mismatch in the string.

[max2k]

Something is off with your understanding, I think.

Coefficient of Voc is 0.3%/deg C.

Voc = 39.3 V
0.3% * 39.3 = .118 V / Deg C.
5.1 V / .118 = 43.25 deg C

43.25 + 25 deg = 68 deg C.

Vmp won't exactly follow Voc, but close enough.

For what its worth, I suspect the optimizers are adding some heat, so my cell temps might be a bit higher than models would estimate, in the 60's rather than the 50's, more or less canceling out any outperformance they achieve by eliminating mismatch in the string.

I agree it was stretch on my part to use Voc temp coefficient in Vmp related calculation. I found Vmp coefficient is rarely specified simply because it is very close to power temp coefficient since Imp coefficient is almost x10 times lower. For your panel I think it is listed as -0.41%/C which would give cell temperature: 25 + 5.1 x 100 / 32.1 / 0.41 = 64 C - close enough as you said

Point still remains the output voltage would not be sufficient for charging 24V lead acid. Panel would be pulling battery up to Voc at that temperature but at sharply falling current past Vmp value.

[J.P.M.]

I agree it was stretch on my part to use Voc temp coefficient in Vmp related calculation. I found Vmp coefficient is rarely specified simply because it is very close to power temp coefficient since Imp coefficient is almost x10 times lower. For your panel I think it is listed as -0.41%/C which would give cell temperature: 25 + 5.1 x 100 / 32.1 / 0.41 = 64 C - close enough as you said

Point still remains the output voltage would not be sufficient for charging 24V lead acid. Panel would be pulling battery up to Voc at that temperature but at sharply falling current past Vmp value.

For those interested, and as a comment or two on Voltage as f(array temp.) and its possible use to estimate cell temps.: On the way to fouling estimates, I've found my array's Voltage as read at the inverter to be quite well correlated with measured and averaged panel temps. as described previously. Kind of a sticky business given that I've measured panel and cell temps. that generally vary over a panel by a deg. C. or two, with average panel/panel temp. variation of several deg. C. over an array, leading to trailing edge.

As a very empirical set of 57 data sets, in quasi steady state (cloudless skies with measurements taken as close to the time of minimum incidence angle as possible), each of my my arrays 2, 8 panel strings undergo an average voltage drop of -1.477 V/deg. C. with a std. dev. of 0.07, N = 60. P.O.A. range 830 to 995 W/m^2. A lot of the variation is f(Vmpp vs. P.O.A.) as can be seen from how the locus of MPPs change as f(irradiance) on a V vs. current graph.

The published Sunpower Voltage change/deg. C. for my panels = - 0.1766 V./deg. K. (or C.), which gives 8 * 0.1766 = - 1.412 V/string per deg. K. temp. change.
The difference of (-1.477) - (1.412) = 0.065 V/deg. C. is what I might expect from instrument/measurement error, panel variation, line/wiring losses and other sources. The two strings are usually within a couple of Volts - a lot of which I can see as being due to panel temp. differences.

Under full sun (P.O.A > ~ 900 W/m^2), my array's average temp. runs pretty consistently at about 27-29 C., above that roof ambient air temp. with a std. dev. of 1.7 C., for N = 67 data sets of clean array readings (different from the 57 data sets referenced above. The ambient roof air temp. BTW, tends to run ~~ 3 - 5 C. above the ground level ambient air temp.

A derived heat transfer coeff. for the array, which, while not strictly a heat transfer coeff. the way I calc. it because it includes glazing reflection losses which I can't account for any other way at present, runs at an average of 33.4 W/m^2 deg, C., std. dev. 2.4 W/m^2 deg. C.

Perhaps interestingly, it's possible to get a dart throw at a panel's heat (transfer) loss coeff. from the N.O.C.T. conditions and the published N.O.C.T. My panels (Sunpower 327's) published N.O.C.T. = 45 C.. +/- 2 C. Since N.O.C.T. is measured at O.C. conditions, ( == 0 current), all 800 W/m^2 as mandated by the N.O.C.T. testing will be rejected as waste heat. So, panel heat transfer coeff. = 800/(45-20) = 32W/m^2 C. +/- 2.6 W/m^2 *deg. C.That number seems in the same ballpark with what I've come up with by temp. measurement of the panels, especially considering my wind velocity was higher (1.74 m/sec. vs. 1.0 for the N.O.C.T. test).

Also, and again FWIW, the temps. I've measured, along with the irradiance and wind measurements taken at the same time for several hundred data trials, compare surprisingly well (to my mind at least) with the Sandia method of estimating cell temps.

For the 57 data sets (34 sets in winter 2014-15, and 23 sets the following summer) done by measuring temps. of all 16 panels a total of 8 times each, 4 on either side of inverter voltage recordings, all done within ~ 12 -16 minutes on clear days around the time of minimum incidence angle, the average of (measured cell temp.) - (ambient air temp.) was 52.42 C. std. dev. 8.05 C. The Sandia calculated temp. diff for the same conditions 52.37 C., std. dev. 8.06 C. There was wider variation when winter and summer data was compared, but that's more than can be described here.

Bottom line: From a need to estimate array fouling, I've compared and correlated array Voltage as indicated by my inverter display with average array measured temps. and found array Voltage to be a good way to estimate array average temp. under clear skies. I've also verified that the Sandia method of estimating array average temps. under clear skies to be reasonably accurate against measurements. I also have estimates of lots of other array characteristics such as overall array efficiency, inverter efficiency, panel per deg. performance deterioration as f(temp.) and some rough estimates of array annual deterioration, as well as the reason I started this whole thing - how dirty an array gets, how the fouling changes performance and how cleaning and rain influence the fouling rates.

Take what you want of the above. Scrap the rest.

[bcroe]

The published Sunpower Voltage change/deg. C. for my panels = - 0.1766 V./deg. K. (or C.), which gives 8 * 0.1766 = - 1.412 V/string per deg. K. temp. change.
The difference of (-1.477) - (1.412) = 0.065 V/deg. C. is what I might expect from instrument/measurement error, panel variation, line/wiring losses and other sources. The two strings are usually within a couple of Volts - a lot of which I can see as being due to panel temp. differences.

If I understand, these were all done running at MPPT, no clipping? Still thinking about building the % of clipping
meter (useful for adjustable tilt panels) tied into panel temp sensing and the panel power curve. Bruce Roe

[kb58]

For the charge controller, does it have the charging profile for the batteries you want to use? Can the equipment handle the higher voltage of modern panels? Honestly, I think I'd start over. You're already tossing out the panels and batteries.

[J.P.M.]

If I understand, these were all done running at MPPT, no clipping? Still thinking about building the % of clipping
meter (useful for adjustable tilt panels) tied into panel temp sensing and the panel power curve. Bruce Roe

Bruce: Thank you.

Short answer is no, there was no clipping.

It seems from all the data I collected in this fashion on ~ 450 of the last 1,207 days when I took measurements because I was home and available to take measurements, and also skies were cloudless at time of minimum incidence angle on the array, the inverter capacity has always been greater than the input from the system. Depending somewhat on amb. and array temps., the inverter will clip if cloud reflection happens and P.O.A irradiance gets to ~~ > 1,080 W/m^2 or so. Other than that, the array will not produce enough to clip under cloudless skies primarily because the irradiance will not be high enough, even at normal (vertical) incidence angle. And, even if temps. did get low enough to increase cell efficiencies way over the 0.2005 STC efficiency, such temps. happen at my location only in the early A.M. when the P.O.A irradiance isn't anywhere near enough to cause array outputs much more than maybe 10 - 20 % of STC capacity.

For ~ 450 days when data was collected at the time of minimum incidence angle, the max. input to the inverter was 4,833 W. The max. output from the inverter was 4,699 W. Those two points were not on the same day.

The inverter clips at 5,033 W. I've observed that output being pinned on the inverter screen while simultaneously observing pyarnometer G.H.I. outputs of > 1,100 W/m^2 on more than a few occasions.

Array STC size is 327 * 16 = 5,237 W.

For all the time/effort you've put into your stuff, I'd get a pyranometer (or several) and put it (them) in the plane of the array(s). Then, keep track of string currents vs. irradiance and look for points where the two no longer look the same when graphed that'll probably be points where the string current is > spec sheet MPP current. That'll be where things may start to clip, or pretty close to it, and you'll have some numbers to back it up/play with. Given your various orientations, some of them at what look like large tilt angles, I suspect you probably have some clipping going on in the winter from increased albedo due to snow.

[AngliaUSA]

Thank you all for the sound advice .... taking all things into consideration and the age of my electronics I will most likely start from scratch on my cabin - off grid back up system.

I might be able to reuse the main disconnect switch, battery temperature sensor and digital monitor etc Plus all the heavy gauge wire and cables will get reused. But I'll get myself a new inverter, charge controller and batteries.

Thanx again.

[littleharbor]

Good idea. You shouldn't have any trouble finding buyers for the inverter and charge controller if they are still operational. That old Trace stuff holds it's value pretty well. I've seen the LCD faceplate alone, from your charge controller selling for 125.00 to 200.00 dollars on eBay They're not making them any more and it has caused the price of even used ones to go up pretty high.