Out of curiosity what sort fo mount to you have?. A tight to the roof mount with little clearance to the roof and side panels to make them look pretty can lead to much higher panel temps than the ambient temp. Ideally you would take the air temps under the panel in several locations and average them, I expect quite a big increase in temp over outdoor temp.

You will never, or almost never, see STC output from you array under most any operating condition.

For starters, how do you measure your cell's temperature ? Under clear skies and high P.O.A. (Plane Of Array) irradiance, and when the cos (angle of beam irradiance) on my array is greater than 0.99, my array's cell temps. run somewhere between 25 C. and 32 C. above the roof ambient air temp. which itself runs between 2 C and about 8 C. above ground level ambient air temp. both variations mostly dependent on wind vector and irradiance for the trials and measurements I've done.

But, I've got a lot of clearance under my array which increases air circulation under the panels and keeps things cooler.
As a 1st approx., for roof mounted arrays up to about 15 cm (~6") of free space under an array, I estimate every inch of clearance between an array and a roof, the average difference in temp. between the array's cells and the immediately surrounding air temp. will increase by maybe about 0.75 to 1.0 C., higher in windy situations, lower for still air.

Also, that 0.5% you write of is 0.5% of the STC rating. So, if your panels are, say, 300 STC W, it's probably more accurate to say your panels lose (300W)*(.005) = 1.5 W per deg. C. of cell temp. increase above 25 C.

In addition, is your array clean ? The rate of array fouling is highly site dependent on location and other things like array tilt and other local conditions. Having written that, my array seems to foul at a rate that decreases array output by somewhere around 0.75% /week if it doesn't rain. Without cleaning (or rain) my rate of array fouling also seems to become asymptotic after about 6 to 8 weeks or so. But again, fouling is highly site dependent. just know that arrays get dirty and dirt decreases array efficiency and thus output. Your array may have similar or different rates of fouling. Whatever your rate of fouling is, if you don't clean the array, it'll affect performance.

But the biggest somewhat easily quantifiable contributor to why you're not seeing STC performance is that your array is not seeing STC irradiance for several reasons, a few of which might include:

1.) The atmosphere is usually not as clear as it could be.
2.) Even if the normal irradiance (that is, irradiance on a surface pointed directly at the sun) approaches 1,000 W/m^2, the direction of the irradiance is not usually normal to the plane of the array. That means the P.O.A irradiance must be adjusted (downward) so that the beam normal irradiance on the array is reduced by the cos (solar incidence angle), and the diffuse portion of the irradiance (which is, depending on cloud cover, somewhere between maybe 20 % and 100% of the total G.H.I. (Global Horizontal Irradiance)) must be reduced in some more complicated way(s).
3.) The anti-reflection coating on a panel's glazing works best at normal incidence angles which are rarely encountered. There can still be significant reflection losses that can occur at off normal incidence angles. Under a lot of conditions, expect a reflection loss of maybe 3% - 10% depending on solar angle of incidence on the array.
4.) An Inverter's temp. related efficiency is pretty much a constant and probably not much of a consideration with respect to changes in system output as f(temp.) but its overall efficiency will affect the overall output.

All that works to reduce the efficiency of a panel or an array or a system..

Hard to say, but back of the envelope stuff, if you have some clearance under your array, at close to normal (perpendicular) incidence angle cosines (that is, cosines close to unity) under sunny skies I'd suggest your array's running maybe 35 C +/- some above ground level ambient temp. with solar noon incidence angles of maybe between 10 and 30 degrees, depending on array orientation.
The less under array clearance you have, the greater will be that temp. difference.

So, at that 35 C ambient air temp. you may have a cell temp. near 35 + 35 = 70 C. .
70 C - 25 C = 45 C.
45 C*0.005 = 0.225 . (1-0.225) * 6.2 kW = 4.81 kW.
Add to that your array probably isn't clean, the G.H.I. irradiance probably isn't 1,000 W/m^2 and the solar incidence angle probably isn't 0 degrees.

Looks to me like your array is probably running just fine.

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

This time of year in Phoenix we hit 38-41C daily and my system is quite similar to yours (but 8 years old soon). Typical power collected the last few weeks under these conditions is around 38kWh each cloud-free day. My best collection time here is between noon-1 PM which under these same weather conditions, nets me about 4.8 kWh in that time period.

I don't notice any obvious issues with your numbers. My data also compares quite well with what others report in my location.

I am measuring the surface temp of my panels with a an IR sensor gun and they are typically around 45C. The microinverters are mounted on the top of the rail directly under the panels. This does not allow for the manufacturers suggested 3/4 inch of air space between the panel back and the inverter itself. I asked the solar install company if perhaps they should have used the offset “L” brackets to mount the inverters to the sides of the rail instead and they said it was not an issue and they don’t use the “L” brackets. I only have 10 inverters and I am going to remount them using the “L” brackets to provide the required 3/4 inch of air space on all sides of the inverter for proper cooling.
I am wondering if anyone knows about micro inverter de-rating at approx 45C and where I could find specs for this?

Also I see the array put out the max of 6.2kw when it has been shaded by a cloud for more than 1/2 hour and then the sun suddenly comes out. It will spike up to 6.2kw then slowly fall back to 4.8kw as everything heats up

I don't know about derating, or measuring the efficiency of a microinverter or (output/input) as f(micro temperature), but I'd bet it's not more than a string inverter which isn't much. The mfg. spec. sheets may have such information, but I'd bet a spec sheet will only list the input/output efficiency.

I really would be more concerned about micro inverter service life with respect to temperature however.
Micros don't have a sterling reputation for robustness. and electronics don't like elevated temperatures much. That doesn't seem like a good combination to me.
Maybe it's not an issue to the vendor, but from what you write it sounds as if the micro company may have a different opinion that they'd express as a warranty claim denial.
Before remounting a micro I'd also consider that a vendor might consider doing so a reason to deny responsibility in case of a problem, whether or not it was a good idea.

Besides being contrary to what the micro mfg. seems to be saying, as for why it's not a good idea to mount a micro on or closer to a panel: If a micro is in more "intimate" thermal contact with, or closer to, a (hot) panel, the micro is going to get a lot warmer in that location than if it's on a standoff bracket for two reasons. First is the issue of less air circulation around the micro and, being closer to the panel also means decreased available micro surface area for convective heat transfer. Second, if strapped directly to the back of a panel, there will probably be more thermal conduction from the panel to the body of the micro.
A smaller issue might be that the portion of the panel directly above the micro will be at a different temp. than the rest of the panel. Hit the panel with your R thermometer both at ate micro and away from it to get an estimeate of the temp. effect of a micro strapped to the back of a panel.
Also, If you do remount the micros, I'd suggest doing so in a way that minimizes exposure of the micro to rain/snow and sunlight.

This is a bit involved, but on temp. measurement: The cell temp. is different (and higher) than either the front side or the back side of a panel with the front side lower than the backside.

I designed my rooftop array so that I can get under it. And, I too have /use an IR thermometer to measure panel temps.
To determine (actually estimate) cell temps. or panel temps., or array average temp. as f(array voltage), I did the following. About 8 years ago or so, I measured all 16 of my panel temps. at 3 random places on each panel from the underside, twice over 4 minutes each side of the minute of min. solar incidence angle (lots of asses and elbows over 9 minutes, with one minute to catch my breath at the min. solar incidence angle time, and do the whole 4 minute process again for symmetry). I did that 56 times over 56 separate days, 30 days around the winter solstice and 26 around the summer solstice, always under bright sun and cloudless skies. I then plotted the average of the panel temps. against the array voltage for that minute for all 56 trials. Doing so I confirmed the voltage coefficient of temperature for my panels so I didn't need to keep measuring temps. while my array fouling measurements were being done (which amount to about 500 or so similar measurements but using the array voltage found from the temp. measurements). Along the way, I also learned a lot about other temps such as roof tile temps. under the array and not under the array, as well as front and backside panel temps. under bright sun and so also their temp. differences, and stuff like an empirical correlation for array temp. difference as f(irradiance, wind vector) and other stuff.
The reasons for the glazing temp. being less than the cell temp. have to do with several things. One is the most likely greater convective heat transfer from the front side due to better air flow over the topside of a panel. The reason your temp. est. may be lower may be because of your IR thermometer and/or how you use it. IR thermometers are more than point and shoot devices. I had to correct for the difference in surface emissivity between the front glazing (~ 0.89) and the back sheet (~ 0.95). In addition, the glazing acts as a better insulator than the panel back sheet by about an order of magnitude. That lowers the temp. of the glazing vs. the cell temp.

Unless your IR meter has a correction that accounts for the surface emissivity you're measuring, (which is referenced to absolute temperature), or you account for it manually, your frontside temp. measurements will be lower than the actual temps. by maybe 10 - 20 C depending mostly but not entirely on the glazing emissivity.
Also, the cell temp. needs to be increased above that corrected temp. because of the insulating effect of the glazing which is adding to the temp. difference between the cell and the top of the glazing.
I believe I solved most of that problem as others have done by measuring the back panel temps., correcting for the back sheet emissivity and assuming the back sheet insulating value to be zero. Then, using info and data from papers I have or read, NREL information and text books, I added 3 C to that average back sheet temp. to arrive at an estimated average cell temp.

All that and about 500 additional measurements similar to those described above but using array voltages converted to cell temps. is how I arrived at my statement about your average cell temp. being about 35 C above your ambient roof temp.

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

You didn't state whose solar panel you have but isn't the frame cross section a squared C ? The glass and cells mounted to the top of the C and the frame bottom of the C mounted to the rails giving you 1"- 1-1/2" air space under the panel, and between the inverter?

Thanks for all the replies, but it seems that there is little to no information regarding the de-rating of the micro inverters and at what temps the de-rating occurs. I'll try contacting the manufacturer and asking the same questions. If I get a an informative response I'll post it here

I mentioned that if a micro's efficiency as f(temp.) is similar a sting inverter's, it probably doesn't change much as a function of it'd own temp. I also mentioned that the mfg. may have such information.

FWIW, I'll restate that given what info you've provided: IMO, your system seems to be running just fine and about as expected.
You're just a victim of your own ignorance as to how PV works.

Seems hardly worth the effort but it's your system to mess with with all you want. I would not bother. What might you gain? A tiny drop in micro operating temps when the wind is blowing?

If you lose a micro a month after you re-mount them, will the installer replace it for free if they notice you've modified their install? It doesn't take much to void a warranty these days. All they need is one excuse.

In answer to the first question: There is lots to gain, de-rating inverters are losing 10% of the total power or 620 watts for every minute they are over 45C. This is not insignificant. In my neck of the woods we easily get a full month where the inverters would be de-rated. If the fix is to simply adjust the placement of the micro-inverters then there is no good excuse for throwing that power away. As far as "You're just a victim of your own ignorance as to how PV works." goes. This is not a helpful comment and I am in fact trying to diminish the ignorance by sourcing the problem and finding the solution.
In answer to the second comment. I am the installer. I attended a one week long course and passed and obtained the required permits and did the install myself, which the inspector passed with flying colours. When I was installing the micro-inverters I noticed that with the standard brackets being used by the company who supplied the materials the inverters did not meet the minimum air clearance requirements between the bottom of the panels and the inverters. I brought this up with the supply company and they said this was the way they installed all of the micro-inverters and it should not be an issue. One of my APSytems YC600 micro-invferters failed within the first 3 months and APSystems replaced it free of charge since they have a 10 year warranty. During the recent hot spell of high 30's weather I noticed the large drop in system output and began to ask the questions and found out about the the "L" mounting bracket to allow the inverter to be mounted farther away from the back of the panel and thereby meeting the 3/4 inch air clearance on all the sides.
I will post more once I make the modifications and see some results.

Here is a good article


The main points:

1. Install inverters in cool locations (shaded wall rather than the roof).
2. Choose locations with sufficient air exchange. Ensure additional ventilation when necessary.
3. Do not expose inverters to direct sunlight. For outdoor installations, use existing shadows or covers for inverters.
4. Maintain the minimum clearance to neighboring inverters or other objects given in the installation guide.
5. Increase the clearance when it is foreseeable that higher temperatures could occur at the installation location.
6. Arrange multiple inverters so that they do not draw in the warm air of other inverters. Offset passively cooled inverters to allow the heat from the heat sinks to escape upward.

I also installed my own system I just don't see how you think you have less than 3/4" space. See the attached cut sheet for a typical aluminum framed solar module, section view AA will give you the dimensions.

just an FYI that article speaks about string inverters not micro inverters under each solar module.