Do SolarEdge (SE) Power Optimizers track temperature?

[Steeler.Fan]

I read that SE optimizers track temperature and shut down modules if T>85C (185F). Is the temperature of the optimizer recorded and accessible from the SE monitoring portal? I am curious to see how hot my roof gets.

[J.P.M.]

It isn't how hot the roof gets, it's the temp. of the panels that counts. Panel temp., or the temp. of any body for that matter, is determined by a heat balance on a panel or array (or an inverter, or anything else). Energy into a panel or array == energy out. Energy in is mostly from solar irradiance with smaller inputs from thermal radiation from surrounding objects and sometimes the ambient air if a panel cooler than the surroundings. Energy out is the power output and also the (usual) thermal losses to the surroundings. Those thermal losses are mostly a combination of what are called convective and thermal radiation losses. VERY roughly, for common amb.- panel surrounding temp. differences, convective losses amount to ~ 60% - 70% of the thermal losses with thermal radiation losses making up most of the rest.

The most important temp. diff. is that between a panel's surface temp. and the surrounding ambient conditions of the air temp. Under a clear sky and relatively high summer POA irradiance levels and at midday incidence angles, depending mostly on wind, that temp. diff. between the ambient air and the panel(s) is usually between about 25 C. to 30 C. depending mostly on wind vector as determined from the formula Q(convective) = h(convective)*A* delta T.

where: Q(convective) = heat transfer from the panel(s) to the surrounding air via convection, delta T = temp. diff. between the panel and the surroundings, A = panel area (both sides, top and bottom), and h(convective) is what usually amounts to an estimate of the convective heat transfer coeff. per unit area (m^2) between the surrounding air and the panel)s) from known or estimated heat transfer correlations for panels sitting on a roof or other configuration. Most inverters do not measure such panel temps. When they do, the results tend to be little more than SWAGS.

So, and very roughly, if the solar input is, say 1,000 w/m^2 in the plane of array and the panel is 20% efficient, 200 W of electrical power will be produced. Almost all of he other 800 W of solar input is dissipated as heat via the above mechanisms of convection and thermal radiation. Of that dissipation, ~~ 0.65*800 = 520 W is via convection to the surrounding air, and the rest of the rejected energy (the 800 W - 520 W) is rejected via thermal radiation to the surroundings (including the sky).

[Steeler.Fan]

Thanks for the very nice basic thermodynamics refresher.
I would like to see measured power optimizer (PO) temperature (T) presuming that the panel T ~ PO T. l know that efficiency is decreased by the Panel's T coefficient. I specifically chose panels with a low T coefficient (Silevo -0.27, Panasonic -0.3). I would like to calculate how much less of a decrement that I would expect from my low T coefficient panels vs standard panels at the measured PO T on my roof now and in the hottest months of Aug & Sept.

[J.P.M.]

Thanks for the very nice basic thermodynamics refresher.
I would like to see measured power optimizer (PO) temperature (T) presuming that the panel T ~ PO T. l know that efficiency is decreased by the Panel's T coefficient. I specifically chose panels with a low T coefficient (Silevo -0.27, Panasonic -0.3). I would like to calculate how much less of a decrement that I would expect from my low T coefficient panels vs standard panels at the measured PO T on my roof now and in the hottest months of Aug & Sept.

You're welcome, but it's mostly about heat transfer rather than about Thermodynamics. The two both deal with thermal engineering, but in different ways.

Bottom line: Panels with lower (less negative) power loss coeff. will tend to operate slightly more efficiently than panels with higher (more negative) power loss coeff. So, "lower" coeff. panels will probably run slightly cooler at higher temperatures than panels with "higher" power loss coefficients.

Aside from that, in practice and to my limited experience, it's very difficult to measure and compare such things as coeff. of power, one panel to the next with the required accuracy.

Reality is, most equal (electrical) size arrays in the same location, and performing the same duty will produce just about equal annual output pretty much regardless of physical size. Seems to me physically smaller (but more electrically efficient) arrays operate cooler because they intercept less radiant energy but put out as much electrical power. But if so, will only be doing so by rejecting less heat through a physically smaller area. That would seem to make sense since the convective film coeff./m^2 is about the same for different panels but in the same location and orientation.

FWIW, there's a pretty strong, reliable and quite linear correlation between string voltages and average panel temperatures that's quite close to the temp. coeff. of voltage correlation as published by the mfg. of most panels that's reasonably accurate or more so. Takes a little calibration, but the results are quite accurate and reliable, at least I've found them to be so.

[Steeler.Fan]

In another string on this forum, I just learned that the SolarEdge auto uploader in PVOutput.org will upload optimizer temperature data as long as one makes a minimum donation to their site. Yesterday was a mostly cloudy day and my peak optimizer temp was 51C ~ 124F.

[JSchnee21]

Hi Steeler.Fan,

I'm not sure that's true. You can (and I do) upload main inverter temperature. But I'm not sure if one can get at optimizer temperature. PVOutput doesn't have module level data that I am aware of (and I make an annual donation). Are you sure you want the temperature of every panel's DC optimizers? Or are you referring to the temperature of the DC to AC inverter mounted on the wall?

See if you can access this link for my system -- my peak inverter temp was 137.7F that day, for example.
https://pvoutput.org/intraday.jsp?id...80626&gs=3&m=0

Since you mentioned the roof, I assume you do want the temp for each optimizer. I don't even see DC optimizer temperature in SE web portal (Layout, choose a panel, click the "i" for additional device information). I'm trying to remember if optimizer temp is displayed on the inverter LCD as they report it. There's input and output voltage and current for each optimizer. I don't remember if temp is reported or not.

FYI, the newer "HD" series inverters seem to run much hotter (as they do not have fans). I'm not sure that was the best design choice. Sure fan modules fail, but they are easily swapped with replacements.

That said, I'm not sure anyone here has really seen a SolarEdge system shutdown or limit power production due to temperature (save perhaps a non-HD unit with a failed fan). There's been some limited concern with a few overzealous folks building supplemental cooling fans. But no smoking gun.

-Jonathan

[Steeler.Fan]

Good point about what temp that represents. I thought 124F must be on my roof because the inverter doesn't seem to get very hot. If the peak temp gets even hotter on a sunny day that would make optimizer temp more likely and visa versa.

[ButchDeal]

In another string on this forum, I just learned that the SolarEdge auto uploader in PVOutput.org will upload optimizer temperature data as long as one makes a minimum donation to their site. Yesterday was a mostly cloudy day and my peak optimizer temp was 51C ~ 124F.

that is inverter internal temperature not optimizer temperature. I don't think it is possible to get to the optimizer temp.

[ButchDeal]

FYI, the newer "HD" series inverters seem to run much hotter (as they do not have fans). I'm not sure that was the best design choice. Sure fan modules fail, but they are easily swapped with replacements.

That said, I'm not sure anyone here has really seen a SolarEdge system shutdown or limit power production due to temperature (save perhaps a non-HD unit with a failed fan). There's been some limited concern with a few overzealous folks building supplemental cooling fans. But no smoking gun.

-Jonathan

The non-HD solaredge inverters below 7.6kW do not have fans either. and the fans on the larger ones are thermal coupled (they only come on at high temperatures).

[J.P.M.]

that is inverter internal temperature not optimizer temperature. I don't think it is possible to get to the optimizer temp.

If the optimizer is somehow attached to the racking and not exposed to direct sunlight, it'll probably be at a temp. somewhere between the ambient air temp. on the roof and the panel temp.

At high (~ > 950 W/m^2 or so) POA irradiance levels, panel temps. will run ~ 25 C - 30 C above ambient air temps. on the roof. Depending on attachment details, I'd add ~ 10 C. to the ambient air temp. on the roof and call it a 1st approx. to the optimizer temp. until I could get a better number.

[TAZ427]

One thing to note on the power optimizer is that the heat gain (internally) is going to be from the efficiency loss. When you have a 99.5% efficient PO, and 330W going through it, that's about 1.65W, that's tiny, and given the metal casing if there's any thermally conductive material used to bridge to whatever component(s) that has that heat loss, then as an educated guesser for someone who's done a lot of thermal analysis in electronics, you're probably talking < 10 deg C of internal high temp to external. If it were a very small component, w/o heat bridging and had an air gap, then it wouldn't really matter the outside temp, but would it's temp would be dictated by the power through as the air temp wouldn't be able to compensate - meaning poor design and it would burn up, like trying to put 1W of power through a 1/4W resister. Basic design principles, and these being used in hot desert environments w/o issue, would be a good indicator that this holds true, and their design is not boneheaded.

That said, I did find a study on the Enphase M215, which has a 96.5% efficiency rating (that's 7x the same loss for the same size panel or about 4x for a 200W panel vs 330W.) The study showed that there was a 20 deg C to delta between ambient temp around the microinverter and the hottest of the internal components (and that's w/ 4x the amount of waste heat going into it and similar sized metal casing.) Unless you're getting 75C (167F) ambient air temps around your PO's, you're not going to be hitting or likely coming close to hitting the 85C derating point. I think you'll find that hard to do even in the Mojave, without seriously sealing off the airflow beneath the panels.

[Steeler.Fan]

I spoke to SolarEdge support today. They said that while optimizers measure temp and shut off power when greater than 85C, SE does not track or record optimizer temperature, only inverter temperature. Thanks for all of the interesting information!

[J.P.M.]

One thing to note on the power optimizer is that the heat gain (internally) is going to be from the efficiency loss. When you have a 99.5% efficient PO, and 330W going through it, that's about 1.65W, that's tiny, and given the metal casing if there's any thermally conductive material used to bridge to whatever component(s) that has that heat loss, then as an educated guesser for someone who's done a lot of thermal analysis in electronics, you're probably talking < 10 deg C of internal high temp to external. If it were a very small component, w/o heat bridging and had an air gap, then it wouldn't really matter the outside temp, but would it's temp would be dictated by the power through as the air temp wouldn't be able to compensate - meaning poor design and it would burn up, like trying to put 1W of power through a 1/4W resister. Basic design principles, and these being used in hot desert environments w/o issue, would be a good indicator that this holds true, and their design is not boneheaded.

That said, I did find a study on the Enphase M215, which has a 96.5% efficiency rating (that's 7x the same loss for the same size panel or about 4x for a 200W panel vs 330W.) The study showed that there was a 20 deg C to delta between ambient temp around the microinverter and the hottest of the internal components (and that's w/ 4x the amount of waste heat going into it and similar sized metal casing.) Unless you're getting 75C (167F) ambient air temps around your PO's, you're not going to be hitting or likely coming close to hitting the 85C derating point. I think you'll find that hard to do even in the Mojave, without seriously sealing off the airflow beneath the panels.

FWIW, the numbers you found on the M215 seem reasonable to me. Depending on the location of the optimizers, or other things with internal heat generation with respect to other (external) sources of heating (or cooling), I'd agree that the internal heat generation will probably be a lesser concern, especially assuming reasonable thermal contact between internal heat source and what look like cooling fins on one face of the SolarEdge optimizers. When operating, a device that has internal heat generation that's dissipated to the environment will undergo a temp. increase until the internal to external temp. satisfies the relation Q = h*A*delta T., where Q = total heat rejection/transfer and delta T is the temp. diff. between the device and a heat sink (in the case of micros/optimizers, the sink usually being the temp. of the surroundings), and A being the area available for heat transfer adjusted for any fins and their efficiency and configuration.. Poor "h" means higher delta T's will be required and so higher internal temp. for a device. Heat transfer 101, etc.

Assuming a decent thermal design, the device location comes into consideration. A device such as an optimizer bolted to racking under an array and in a mostly vertical orientation for example, will probably run quite a bit cooler than an optimizer affixed flat and parallel to the underside on a panel. I'd also wonder what an optimizer affixed to the back of a panel might do to the temp. distribution over/around the location of the optimizer location. Probably make that location warmer than the rest of the panel. Different conversation.