Enphase field MTBF: M190: ~36 Years M215: ~316 Years M250: >357 Years

My Nissan Leaf Forum

Help Support My Nissan Leaf Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

RegGuheert

Well-known member
Joined
Mar 19, 2012
Messages
6,419
Location
Northern VA
I decided to update the title and first post in this thread since the original premise I had was false. There does not seem to be a Phoenix-heat-related issue with Enphase microinverter failures. For reference, you can still read the original unedited post below, if you like.

For reference, please refer to Enphase Microinverter Field MTBF Estimate spreadsheet. For those that have looked at this before, please have another look, as it has been updated by dividing M190 and M215 calculations and I have added a new column for "Site MTBF" due to the extremely wide variations in MTBF seen with the M190s. I have added notes of anomalies for each of the sites where strange performance was observed.

Please note that with the exception of the one site in Troy, IL, none of these sites were selected BECAUSE they had failures. The rest of the sites were selected because they were close to me or to Phoenix, AZ, or I found out about them here or on the NAWS forum.

Here are some facts from the spreadsheet:

M190 (Also includes M380 models):

Number of sites tracked: ~80
Number of inverters currently tracked: ~2500
Total number of device-years tracked: ~8800
Total number of device failures detected: ~129
Overall Field MTBF: ~60 years
Highest site MTBF (with a failure): 337 years
Highest site MTBF (without a failure): >237 years
Lowest site MTBF: 5 years
Oldest site: 4.78 years

The bottom line is that M190 MTBF seems to vary widely by installation. Several of the worst MTBF sites I have found are near here where heat is not a huge factor. As such, I do not think that heat is the major issue with failures so far. I suspect lightning and/or installation quality may be bigger issues at this early point in their lives. I am also wondering if there may have been some manufacturing issues with the M190s in the middle of 2011 when the M215s were just coming online. We may never know.

M215:

Number of sites tracked: ~40
Number of inverters currently tracked: ~1000
Total number of device-years tracked: ~1300
Total number of device failures detected: 1
Overall Field MTBF: ~1300 years
Highest site MTBF (without a failure): >405 years
Oldest site: 2.18 years

While the M215s are quite young, they have been rock solid so far. Only one failure has been found to date. Tiime will tell if this high reliability holds up until end-of-life sets in.

Unfortunately, it looks like the old public Enlighten sites which allow me to see inverter failures will be taken down on September 3, 2013. (But as of today, September 3, 2013, it is still up!) As such, this little experiment may not last much longer. I added a bunch of Phoenix sites recently to see if there is an issue there. Phoenix looks no worse than anywhere else for M190s or M215s. Today, September 3, 2013, I added sites from Petaluma, CA, and Santa Rosa, CA.

All-in-all, this has been an interesting experiment.

If you have any suggestions before the per-inverter data disappears, please let me know.

Original post:

I have been tracking field failures of Enphase microinverters for a while now by watching the public sites for many of our arrays to try to spot the signature of microinverter replacements. So far, I have accumulated over 1200 device-years of operation and see a total of 4 failures in the field, including one of my own. That gives an MTBF of over 300 years.

For your reference, here is my current spreadsheet showing the MTBF calculation:

https://docs.google.com/spreadsheet/ccc?key=0AqhUVOciAXVhdEJRM2xXSFVZY3Vnak53ZFAyMHhuc3c" onclick="window.open(this.href);return false;

To see failures, click on the link for the OLD website and then click the "Lifetime Energy" button in the top right. Panels with lower production are potential failures. Verify by checking for shading by looking at the "Past 7 Days". This approach is not foolproof, as there may be some false negatives due to failures very early in life, but those could be considered as infantile failures anyway.

Any, I found an array yesterday in Gilbert, AZ that appears to have a failure about 9 months ago and there is currently a second panel which is not producing. Of course this second failure is not confirmed to be a microinverter until it gets replaced, but I will assume for now that it is. If so, then I see three failures in AZ and one here in VA. But my failure was very early in the life of the array and it could have been caused by improper operation, so it is a questionable inclusion.

Anyway, I'm starting to wonder if we will see a higher failure rate for Enphase microinverters in AZ than in other areas. Or even worse, I'm wondering if these failures in AZ might be the leading edge of the back end of the bathtub curve where units reach their end-of-life.

It's too early to answer these questions, but it does make sense that AZ would be harder on microinverters sitting on hot roofs just like it is harder on most other things. Of course the one site in Gilbert with two failures out of ten could be a bad installation, but it could also simply be a site with particularly high roof temperatures.

Unfortunately, I will not be able to track Enphase failures much longer. Enphase has redesigned their public web pages and I do not see any way with the new pages to see microinverter replacements. I suppose this was intentional to hide that type of information.

If you have an Enphase-based array which is not included on this spreadsheet, please post your URL here and I will add it.

Note to QueenBee, I have your array on the spreadsheet, but I have excluded it from the analysis since it has not reported anything since January. Have you disconnected from Enlighten?
 
Hey, thanks for tracking this kind of stuff. I've got M215s BTW. My panels are currently offline because I redid the roof in preperation of tripling the size of my system. Unfortunately it's taken longer to get people back to put them on. I really need to get things moving since spring is coming.

I've yet to have any failures.
 
QueenBee said:
I've yet to have any failures.
Thanks! I've updated the spreadsheet to include your inverters. Please verify it is correct now!

Also, if everyone else that has an array on that spreadsheet could verify that all the data is correct, that would be appreciated! TIA!
 
My Sunpower dealer strongly cautioned against microinverters here, and I'm happy with my production, so it just wasn't worth the risk until they have more years in the area. String inverters have a lot more years on the ground here.
 
RegGuheert said:
Any, I found an array yesterday in Gilbert, AZ that appears to have a failure about 9 months ago and there is currently a second panel which is not producing. Of course this second failure is not confirmed to be a microinverter until it gets replaced, but I will assume for now that it is.
My assumption was correct. The second inverter out of ten total in this array was replaced today. That's a 20% failure rate in three years! I wonder if this trend will continue.
 
Yet another failure in AZ: Zomeworks Trackers.

And this report, complete with pictures, of a system with three failures, the third one looks as if it was blown apart. Unfortunately, I do not know where this poster lives. Hopefully he will add a URL for his system once he reconnects his Envoy unit.

The MTBF calculation on my spreadsheet is dropping. With the new failure in AZ and a system I added nearby me with two failed inverters. I wish I knew the owner. I'd like to learn more about that system.

Anyway, I may be adding a bunch more systems to the calculation soon since someone over on NAWS posted a link showing ALL the Enphase systems, many of which have public URLs: All Enphase Installations. Anyway, this could be interesting for anyone trying to calculate MTBF for systems in your area.
 
Does Enphase list a min/max operating temp? I would love to know how hot it gets under a black panel in the middle of July here.
 
jspearman said:
Does Enphase list a min/max operating temp?
The M190 has an operating temperature range of -40C to +65C. These are the units that have been failing. But it's interesting that Enphase has written the specifications slightly differently for the newer M215 microinverters:
Ambient temperature range: -40C to +65C
Operating temperature range (internal): -40C to +85C
This is interesting, since the calculations done for the M190 capacitor life were done at +65C, but the dissipation of the two units should be about the same since the efficiency of the M215 is higher. So it implies to me that Enphase is seeing internal temperatures in the M190s up to 85C in some installations. I'll have to look at the temperatures in our inverters on a hot day.
jspearman said:
I would love to know how hot it gets under a black panel in the middle of July here.
If you know anyone with these on their roof out there in AZ, they should be able to read the temperatures from the secure Enlighten website for their installation.
 
jspearman said:
Does Enphase list a min/max operating temp? I would love to know how hot it gets under a black panel in the middle of July here.

I have a roof temp sensor, and I believe it hit 170-180F last summer. I don't have Enphase micros (have Honeywell/Exeltec), but Monday will be the seventh one to be replaced out of 20. The one Monday is a replacement for a replacement.
 
Anecdotally, it seems that if a site has one failure, it appears that it is more likely to experience additional failures. It seems that there may have been some bad batches here and there resulting in premature failure. Specifically with the M380. The M215 should be more reliable than the 190W inverters.

Either way, heat will speed up the failure rate of electronics, just as it speeds up capacity loss of lithium batteries.
 
LEAFfan said:
I have a roof temp sensor, and I believe it hit 170-180F last summer.
Wow! :shock:
That would explain why we see so many failures in AZ. It doesn't explain why I see two failures out of 41 inverters on a roof near here.
LEAFfan said:
I don't have Enphase micros (have Honeywell/Exeltec), but Monday will be the seventh one to be replaced out of 20. The one Monday is a replacement for a replacement.
What a pain. At some point this becomes unworkable for both the manufacturer and the customer.

In the case of Enphase, I wonder if they will simply offer an upgrade to M215 if the failure rates get high enough.
drees said:
Anecdotally, it seems that if a site has one failure, it appears that it is more likely to experience additional failures. It seems that there may have been some bad batches here and there resulting in premature failure.
THAT may explain why the nearby system has two failures. Or perhaps it just happened.

I suspect that the roof characteristics and the installation may play a key role in operating temperatures. For instance, a roof with a low pitch and with the mounting rails down close to the roof should have lower airflow from chimney effect on a still day and thus should see higher temperatures. It may be possible to estimate roof pitch based upon seasonality of production. I wonder if a correlation might be seen there.

I have also taken the additional step of mounting the microinverters on our roof so that the outer "can" is upside down so that water could not pool in the inverter. While Enphase has said these are NEMA 6 rated enclosures and that the encapsulant would prevent water incursion, I didn't want to risk any moisture getting inside. Frankly, it was a pain to do, but who knows it it will make a difference. Perhaps it was wasted effort.
drees said:
Specifically with the M380.
I don't think I have any M380 failures showing in my spreadsheet, yet. The signature for that should be two adjacent inverters being replaced simultaneously. That's propably because there weren't as many of those installed. IMO, it wasn't one of Enphase' better ideas.
drees said:
The M215 should be more reliable than the 190W inverters.
Let's hope so. If not, then Enphase may be in serious trouble.
drees said:
Either way, heat will speed up the failure rate of electronics, just as it speeds up capacity loss of lithium batteries.
Agreed. And that brings us back to the title of the thread. So we watch as we see what appears to be a lower-than-predicted (by Enphase) MTBF in Phoenix. We also have roofs with multiple failures in more moderate climates like around here, which I find surprising.

So, like with the LEAf, the rest of us wonder what the implications are for our installations. While we have a much better warranty in place than Nissan offers for the LEAF, if the overall failure rate is high enough, then Enphase is doomed and our warranties will not cover us.

I wish I could keep this spreadsheet running long-term, but I suspect Enphase will turn off the old websites within the next few months. After that, we will be limited to anecdotal, self-reported data, which is not usable for calculating MTBF.
 
RegGuheert said:
I'll have to look at the temperatures in our inverters on a hot day.
Well, I dug around and found the hottest inverter in our array as well as the highest temperature it achieved. Interestingly, the highest internal inverter temperature was NOT recorded on the hottest day of the year, but rather it was on a hot day with very little wind: July 6, 2012. Here's a plot of power and temperature data for the week around that date:
Hottest Inverter 6 July 2012.png
(The new Enlighten website refuses to put Y-axis information on their graphs for anything other than power. But by poking around, I have figured out that the scale for temperature in Fahrenheit is one-half that shown for Power Produced (W). In other words, if you see 250 on the scale, you read the temperature curve as 125F.)

As you can see, the inverter achieved a temperature of 149F, which is 65C, the specified limit for this inverter. It stayed at that temperature for one hour. I'm pretty surprised to find that the internal inverter temperature was 54F higher than the day's high temperature of 95F. Also surprising was the fact that the temperature difference between the hottest and the coolest inverter on our roof that day was 25F.

For reference, our roof has a 7/12 pitch, which is the same as a 30-degree elevation. In other words, it is not overly flat. I would expect a flatter roof to have an even higher temperature rise than ours. Also, if the array were larger or had more restrictive airflow, further rises are conceivable. A 65F rise does not seem unreasonable. In Phoenix where the ambient temperature can reach 118F, that means the internal inverter temperature could reach 183F or 84C. It's interesting that this is roughly the maximum internal operating temperature of 85C specified for the M215.
 
RegGuheert said:
LEAFfan said:
I have a roof temp sensor, and I believe it hit 170-180F last summer.
Wow! :shock:
That would explain why we see so many failures in AZ. It doesn't explain why I see two failures out of 41 inverters on a roof near here.

Exactly! That's what I tried to show in another thread. The problem isn't limited to Arizona - the sun is pretty intense 'down here' across the south. Those 170-180°F roof deck temps run through Texas and Florida as well.

http://www.fsec.ucf.edu/en/publications/pdf/FSEC-CR-1496-05.pdf

And then there's Illinois: :shock:
Peak wood sheathing temperatures during summer were also reduced by about 4°F for south facing sections (173° vs. 177°f).
 
RegGuheert said:
Interestingly, the highest internal inverter temperature was NOT recorded on the hottest day of the year, but rather it was on a hot day with very little wind: July 6, 2012.
I have found the same correlation. Low wind = hot inverters and the hottest inverters are in the middle of rows and if there are multiple rows the higher rows have higher temps.

I believe the panels themselves are seeing the same temperature differences as production tends to be higher on the edges and bottom row, too.

The higher the rails/panels are off the roof the better and having gaps in between rows is good to let the hot air escape. Best case scenario for keeping panels/inverters cool is probably panels tilted up at a higher angle than the roof.
 
AndyH said:
Exactly! That's what I tried to show in another thread. The problem isn't limited to Arizona - the sun is pretty intense 'down here' across the south. Those 170-180°F roof deck temps run through Texas and Florida as well.
But do roof deck temperatures correllate well with microinverter temperatures? It seems there are many factors involved so it would be very hard to know if the climate for the inverter would be hotter, cooler or about the same as the bare roof deck had been. Here are some factors:

- Original color of the roof deck. Of course if the roof is a lite color it will reflect a significant amount of sunlight and will therefore run cooler. A black roof will attain the highest temperatures.
- A producing PV panel (such as one with a microinverter attached) should never attain as high a surface temperature as a black roof would. This is because the PV panel will convert 12 to 18% of the light into electricity rather than heat. OTOH, the PV panel may attain higher surface temperatures than a white roof would.
- The microinverter lives in an environment beneath the PV panel. In cases where the air is mostly trapped around the inverter that environment might resemble a solar cooker, resulting in a large temperature rise in the air around the inverter. In other cases, there will be easy movement of ambient air around around the microinverter perhaps aided by a chimney effect caused by the heating of the panels themselves.

By looking at the peak temperatures in the microinverters on our roof, I can see that there is a fairly wide range of environments found there. Both the hottest and the coldest inverter are found on the garage. The hottest is in the top row, third from the right end while the coolest is in the bottom left corner. The former is at the top of the "chimney" and in and oven, of sorts and the latter is at the bottom of the "chimney" and is very open to ambient air. I will say that the thermal environment would have been significantly better had I run the mounting rails up and down the roof rather than side-to-side. (Of course, the panels would not have fit very well that way!)

In any case, it seems that the M215s should manage to live in an environment for which they are designed in all but the very worst installations/inverter locations. But I'll agree that if the term "Operating temperature range" on the M190 datasheet means the INTERNAL temperature like it does on the M215 datasheet (and I think it does), then those electrolytic capacitors might be spending a few hours during some summer days close to 85C. Is that a problem for the capacitors? I doubt it. According to Nichicon, they should be able to live for 32,000 hours (3.6 years) operating at that temperature. Since they spend the vast majority of their life significantly cooler than that, they should be good for well over a decade. So my conclusion is that none of these capacitors must be worn out yet.

But the high temperatures and daily thermal cycling stress everything, including solder joints. So are we seeing a wide range of random failures with the rate increased by the heat or is there a consistent failure mode in the M190s. While I'm sure Enphase knows the answer since they replace all the failures and pay for return shipping, it's really hard for us to say.

Perhaps someone with microinverters also has a roof sensor. LEAFfan, do you have access to internal temperature data from your inverters? It would be interesting to compare that to your roof sensor.
 
drees said:
I have found the same correlation. Low wind = hot inverters and the hottest inverters are in the middle of rows and if there are multiple rows the higher rows have higher temps.
That's precisely the distribution of temperatures that I see, too.
 
RegGuheert said:
AndyH said:
Exactly! That's what I tried to show in another thread. The problem isn't limited to Arizona - the sun is pretty intense 'down here' across the south. Those 170-180°F roof deck temps run through Texas and Florida as well.
But do roof deck temperatures correllate well with microinverter temperatures?
Are we missing something here? Two roofs at latitude angles with the same brown asphalt shingles - one in central Wisconsin and one in Shreveport -- which will have the higher roof deck temperature? Which will have the hotter solar panel? Which will have the hotter air between the roof and panel?
 
AndyH said:
Are we missing something here? Two roofs at latitude angles with the same brown asphalt shingles - one in central Wisconsin and one in Shreveport -- which will have the higher roof deck temperature? Which will have the hotter solar panel? Which will have the hotter air between the roof and panel?
You are making a non sequitur argument here. Just because a roof deck in Shreveport is hotter than a roof deck in Wisconsin it does not follow that a microinverter-based PV installation on the Shreveport roof will have a high failure rate.
 
RegGuheert said:
AndyH said:
Are we missing something here? Two roofs at latitude angles with the same brown asphalt shingles - one in central Wisconsin and one in Shreveport -- which will have the higher roof deck temperature? Which will have the hotter solar panel? Which will have the hotter air between the roof and panel?
You are making a non sequitur argument here. Just because a roof deck in Shreveport is hotter than a roof deck in Wisconsin it does not follow that a microinverter-based PV installation on the Shreveport roof will have a high failure rate.
No, stop reading into the statement. I didn't say a THING about microinverters - only temperatures.
 
AndyH said:
No, stop reading into the statement. I didn't say a THING about microinverters - only temperatures.
This thread is about microinverters, not roofs. Please start a separate thread if you want to discuss roofs.
 
Back
Top