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

[RegGuheert]

This photo was posted some time ago by one of the HW Engineers that works at Enphase. It is a photo of an Enphase Microinverter model M215, serial no. 1, so I believe it is a prototype design. Four very large (Nichicon?) electrolytic capacitors are quite evident.

Thanks for the picture! Very cool! That does look like an M215 footprint to me and those are, in fact, Nichicon PW-style aluminum electrolytic capacitors. It is also clear that this unit was assembled in 2010 or later, since many of the components have late-2009 and early 2010 date codes on them.

Unfortunately, since we cannot see the capacitance or voltage rating for the electrolytic capacitors shown, we cannot make a direct judgement whether they are low-voltage capacitors that would be used at the input to the inverter or if they are high-voltage capacitors that would be used in between a boost converter and a buck converter. I suspect these capacitors in that picture are, in fact, at the input to the inverter, just like they were in the M190 architecture. Here is what Enphase said in their brief on capacitor life in 2008 prior to the release of the M190:

A capacitor is commonly used to store the energy that must be stored and retrieved by the inverter. This capacitor is usually located on the PV bus, and has to be large enough to control the voltage ripple across the bus. This ripple would be detrimental to MPPT accuracy otherwise.

Electrolytic capacitors are very well suited to control the ripple because of their low Equivalent Series Resistance (ESR) and high capacitance per volume.

But I still wonder if Enphase shipped M215s to customers containing aluminum electrolytic capacitors. As mentioned, it was rumored that they developed a new two-stage inverter architecture which eliminated the need for electrolytic capacitors. In fact, Enphase provides a link to US Patent 8,659,922, Photovoltaic module-mounted AC inverter, which details such an inverter. This patent application is very clear that part of the intent of the claimed architecture was to eliminate electrolytic capacitors and use only solid-dielectric capacitors:

A photovoltaic module-mounted AC inverter circuit uses one or more integrated circuits, several switches, solid dielectric capacitors for filtering and energy storage, inductors for power conversion and ancillary components to support the above elements in operation.

Previous attempts at development and marketing of AC modules have met with little or no success. The primary reason has been that the sales volume was too low to achieve any kind of economy of scale. The components used in the associated inverters were off-the-shelf, and in many cases were not optimum for the application. The inverter lifetime was limited by many of these components, especially the electrolytic capacitors. The reliability levels of existing off-the-shelf components used in the inverters has limited their lifetime to between five and ten years.

What is desired, therefore, is a circuit that uses a minimum number of components, uses no limited reliability components and has been optimized for large-scale manufacturing.

Note how different the above verbiage is regarding electrolytic capacitors when compared with what was written in 2008.

According to the present invention, a PV module-mounted AC inverter circuit uses one or more integrated circuits, several power transistors configured as switches, several solid-dielectric capacitors for filtering and energy storage, several inductors for power conversion and ancillary components to support the above elements in operation.

6. The apparatus of claim 1 wherein the power conversion module is a buck converter, and wherein the inverter further comprises a boost converter and a charge storage.

13. The apparatus of claim 6, wherein the boost converter further comprises two parallel connected inductors operating in quadrature phase so that neither inductor is simultaneously discharged.

This patent even proposes the possibility of implementing the energy storage element as an inductor rather than a capacitor:

Charge storage 204 is normally implemented as a capacitor, but may also be implemented as flux storage by utilizing an inductor and changing DC bus 205 to a current mode of operation.

This patent also focuses on the reduction of the number of components in the inverter:

The prior art PV module-mounted inverter 101 utilizes between 200 and 1000 standard components to perform its functions. The large number of components requires a large dimension for the enclosure of the inverter and significantly reduces the reliability and maximum lifetime of the device.

The assembly of the monitoring system 211, communications system 212 and control system 213 into integrated circuit 210 results in a total component count for the present invention PV module-mounted inverter 201 of between 20 and 80 components. This is significantly less than that for the prior art PV module-mounted inverter 101 and leads to a significant reduction of present invention PV module-mounted inverter 201 cost and a tremendous improvement in reliability. Reliability is further enhanced since a custom integrated circuit 210 can be designed to achieve a lifetime of 25 years that well exceeds the lifetime of 10 years for typical off-the-shelf commercial integrated circuits. The integrated circuit 210 may also include active and passive components of boost converter 203 and buck converter 206.

Finally, the following introduction video for the M215 microinverter talks about the architecture of the new generation:

Enphase was able to achieve the higher-power output through a completely new inverter architecture which uses a variety of custom-developed components with an ultra-low profile.

[youtube]http://www.youtube.com/watch?v=P8Q9YE6XdYk[/youtube]

 

[dsinned]

Typically, the MTBF of many different forms of commercially available SMPS (Switch Mode Power Supplies) is from 50,000 to 250,000 hours. Some of these are merely reliability predictions based on Mil-Std 217 and often boost such low failure rates based on ideal operating conditions (25C ambient). Enphase claims life expectancies of on par with >300 years. The highest grade of commercially available electrolytic capacitor is around 10,000 hours. That is usually specified at 105C, but what kills them other than excessive heat stress is high ripple current.

 

[RegGuheert]

Typically, the MTBF of many different forms of commercially available SMPS (Switch Mode Power Supplies) is from 50,000 to 250,000 hours. Some of these are merely reliability predictions based on Mil-Std 217 and often boost such low failure rates based on ideal operating conditions (25C ambient).

As you can see from the spreadsheet linked in the OP of this thread, the MEASURED MTBF in the customer application for Enphase M190 inverters averages 50 years and the range seen between different installations is between 5 years and 337 YEARS. MEASURED MTBF for M215 inverters is over 1300 YEARS.

Enphase claims life expectancies of on par with >300 years.

No, they don't. They claimed a calculated MTBF of 331 years for the M190s.

The highest grade of commercially available electrolytic capacitor is around 10,000 hours. That is usually specified at 105C, but what kills them other than excessive heat stress is high ripple current.

The Nichicon capacitors used in the Enphase M190s are rated for an 8000 hour life at 105C. Calculated lifetime for those capacitors in the M190 application is about 30 years. However, as mentioned, Nichicon does not stand behind any calculated lifetimes beyond 15 years. That is why the warranty for the M190s is only 15 years.

 

[Weatherman]

Another m190 death reported to Enphase, today. This is death number seven this year, and one of those was a replacement for a previous death in October, last year.

...

Out of a 30-panel array, I'm at replacement number 8, and this doesn't even include the six other ones which are dropping power due to grid gone events.

Will be calling Enphase, this morning, to report m190 death number 9, only five days after reporting death number 8. They are happening so fast, now that the replacement for death 7 is still in shipping transit, and I haven't receive the shipping confirmation for death 8, yet.

The latest death marks a 50% failure rate in a little over two years. I'll be surprised if I have more than a handful of the original micros left, after five years. I don't think any of them will survive the full warranty period.

 

[RegGuheert]

I finally got around to making some updates to the Enphase Microinverter Field MTBF Estimate spreadsheet. The major change this time is that I've changed the column that contained the links to the old Enlighten websites to contain the power rating of the PV modules connected to the inverters.

Interestingly, most of the M190 microinverters are grouped around about 230Wp, which was the recommended found on the Enphase M190 datasheet. I had thought that perhaps I would find a correlation to power rating, but I don't really see it. Weatherman has 240 Wp panels connected to his M190s and has a high failure rate, but there is an array in PHOENIX with 235 Wp panels (the same panels I have) attached that was sporting an MTBF of over 224 years when I lost the ability to track failures last September. The critical factor in that case seems to be that the array is ground-mounted rather than roof mounted. Now that I have moved 12 of my M190s to a ground-mount array, it will be interesting to see if their failure rate diverges from the other 30 that are still on the roof.

So, here are some current MTBFs for members here (only counting replacements, not limp-along firmware);

M190s:

Weatherman: 8 years (You get the award for lowest-reliability Enphase installation among LEAF owners here. Sorry.)
ecoobsessive: 10 years (With D380s, I wouldn't be surprised to see you pass Weatherman.)
RegGuheert: 53 years (I'm right at the average for the M190s in the spreadsheet, but many sites have not been updated in over a year.)
drees: over 82 years (No failures and still counting. You get the prize for the oldest Enphase installation here at over 4.5 years! You also get the award for the next-to-the-lowest-power PV modules at 180W.)
Tony Williams: over 114 years (No failures and still counting. BTW, do you have the 225Wp Sunpower panels as I indicated in the spreadsheet?)
Synertx in Phoenix, AZ: 337 years (Only ONE failure after nearly four years of operation with 90 M190s in Phoenix! That was a year ago with no way to update. It is interesting to see the highest reliability M190 array in the hottest area.)

M215s

dsinned: 28 years (Only M215 failure of anyone here, but not in the spreadsheet because I do not have a URL for the array. Also, I will note that this appears to be an infantile failure, so may not actually count for MTBF.)
QueenBee: over 128 years (No failures and still counting. You get the prize for highest-relibility Enphase installation here.)
Jall Investments in Phoenix, AZ: over 405 years (No failures and still counting...and that was back in September 3, 2013. If there have still been no failures, then the MTBF is over 600 years today. Unfortunately, I now have no way of knowing whether this is true or not. As of today, this system has already produced over 277 MWh of electricity! Methinks Jall makes good investments!)

I find it interesting that both of the systems (M190 and M215) with the highest demonstrated MTBFs are in Phoenix, AZ. It is not what I expected, as the original title of this thread will attest to.

I also think it is ironic that Enphase took away the ability to easily monitor MTBF of their products just as they started to ramp up to new heights with the M215s.

 

[Nekota]

You can update line 131 with 20 M215's to zero failures as of today. I do have a watch on two of my panels/inverters that have lost some output compared to the others.

 

[RegGuheert]

Based on this little factoid, I am going to make the following predictions:

- The next inverter to fail in our PV array will be 121045414066. This inverter is in the garage array, top row, second from the left.
- The second inverter to fail in the future in our PV array will be 121045415714. This inverter is in the house array, middle row, third from the left.
- The third inverter to fail in the future in our PV array will be 121045414823. This inverter is in the garage array, top row, fifth from the left.
- The fourth inverter to fail in the future in our PV array will be 121045415782. This inverter is in the house array, bottom row, second from the left.

Well, after four months, my prediction of failures was not perfect, but also not too far off. I predicted four inverters to fail next and of those four, one was replaced and one got new firmware. OTOH, I had an inverter which failed and was replaced which was nowhere on my radar screen.

I would like to update my prediction based on the same criteria (overly high production numbers), but using the most recent production information. Yesterday was the perfect solar day (something very rare around here!) and it allows me to easily compare production from the various inverters. Here is a picture showing yesterday's numbers:

Based on that image, I will be keeping the first inverter from my previous prediction, but dropping the last one. Here is my prediction of the next four failures, in numerical order:

- 121045414066 - Garage, Top, Second from the left
- 121045414688 - Garage, Top, Fourth from the left
- 121045414885 - Garage, Middle, Third from the left
- 121045414916 - Garage, Middle, Fifth from the left

These are the inverters which produced either 1.43 or 1.42 kWh yesterday. Interestingly, they are all located on the garage. (Note that the inverters in the field array ALL produced more energy than these four, but that array is pointed more suitably for this time of the year. As such, I do not see any indications of inverter problems there.)

Also notice the production of the two inverters which Enphase has put on limp-along firmware are underproducing their neighbors:

- 121045414867 - Garage, Middle, Second from the left (Underproduced neighbors by 3%. This was the first of my microinverters to get the limp-along firmware and it no longer has any obvious drop outs, yet still lags production slightly for some reason. I suspect it is using a fixed duty cycle for operation of its power stage, but I have no way to confirm that.)
- 121045415714 - House, Middle, Third from the left (Underproduced neighbors by 8%. It had a bad day yesterday with long dropouts. I still do not know what prompted Enphase to change the firmware on this unit. To my knowledge, it did not indicate any malfunction PRIOR to the change in firmware, even though it was on my watch list.)

 

[Weatherman]

In my case, they, apparently, were so desperate to stop the bleeding that they updated all of my micros to the latest software (the one that cuts power on grid gone events). They didn't tell me they were going to do this, they just went ahead and did it. I have about ten of the original ones that drop power. The new ones seems to be ok. And about a dozen of the originals (only 40% of the total), are still running properly.

Didn't help much with the bleeding, though. Another micro died last week, and I'm waiting for a replacement.

 

[wwhitney]

I would like to update my prediction based on the same criteria (overly high production numbers), but using the most recent production information. Yesterday was the perfect solar day (something very rare around here!) and it allows me to easily compare production from the various inverters.

Hi Reg,

Just curious--how do you know that variation in inverter production is primarily attributable to the inverters, rather than the panels?

Cheers, Wayne

 

[RegGuheert]

Hi Reg,

Just curious--how do you know that variation in inverter production is primarily attributable to the inverters, rather than the panels?

Cheers, Wayne

Hi Wayne,

Good question!

It's really just a theory I have based on the second failure I saw in the array (the first one died very early on). That inverter had reported the highest energy harvest of any of the inverters in the system. In my mind, it didn't matter whether the higher production was due to the inverter or the panel, just that it was higher than the others. I wanted to see if I could predict other failures based on that criterion, and some of the predictions were accurate.

But, based on your question, I went back to October 15, 2011, when I had another perfect solar day and looked at production. Here it is:

That image is a little bit confusing because back then, there were only 42 inverters in the system. Since that time, the 12 inverters that are seen above on the house as grey rectangles have moved out into the field array. But the important thing is that the inverters on the garage have not changed, except for the two there which have been replaced since then.

As you can see from this image, the four inverters which are currently "overproducing" as is shown in the recent image upthread were previously producing very much the same power as their neighbors. This could still be due to the PV modules, but I doubt it. The reason I say this is that when previously-overproducing inverters have been replaced, their production then falls in line with their neighbors.

Again, it is just a theory. I'm just trying to see if I can spot the failures before they occur. We'll see how it goes...

 

[RegGuheert]

Here is my prediction of the next four failures, in numerical order:

- 121045414066 - Garage, Top, Second from the left
- 121045414688 - Garage, Top, Fourth from the left
- 121045414885 - Garage, Middle, Third from the left
- 121045414916 - Garage, Middle, Fifth from the left

I am going to update my "watch list" based on a plot from a recent day in which it was sunny in the afternoon, but cloudy in the morning. That eliminates the difference in production between the few panels that normally get shaded in the early morning and the rest which do not. Here is the image from that day:

Based on that image, I will arbitrarily create TWO watch lists for the inverters in my array. The first will be for all inverters on the house or the garage which produced at least 1.00kWh that day (the inverters in the field produce more this time of year due to better pointing):

Primary Watch List:
- 121045414066 - Garage, Top, Second from the left
- 121045414856 - Garage, Top, Seventh from the left (Far right)
- 121045414885 - Garage, Middle, Third from the left
- 121045415731 - House, Middle, Sixth from the left

Those inverters on the Primary Watch List are reporting the highest production anomolies and I consider them to be the most likely inverters to fail next.

Next, I will create a Secondary Watch list, for the remaining M190 inverters on the house and garage that reported production above 985Wh on that day. They are:

Secondary Watch List:
- 121045414688 - Garage, Top, Fourth from the left
- 121045414916 - Garage, Middle, Fifth from the left
- 121045415781 - House, Bottom, Far left
- 121045415782 - House, Bottom, Second from the left

The inverters on the Secondary Watch List are reporting production higher than the other inverters on the house and the garage, but are not as high as those on the Primary Watch list. All other inverters are reporting normally except for the two which are currently on Enphase's "limp-along" firmware. They are producing relatively normally, although they produce "Grid Gone" events daily or multiple times each day:

Limp-Along Firmware:
- 121045414867 - Garage, Middle, Second from the left
- 121045415714 - House, Middle, Third from the left

As such, I currently have 10% of the original M190s on the Primary Watch List, 10% on the Secondary Watch List and 5% on Enphase's Limp-Along Firmware. We'll see if the next failures come from these groups or from elsewhere in the array.

 

[RegGuheert]

It has been a bit over four months since my last prediction of which microinverters are next to fail but none have failed since that prediction.

All other inverters are reporting normally except for the two which are currently on Enphase's "limp-along" firmware. They are producing relatively normally, although they produce "Grid Gone" events daily or multiple times each day:

Limp-Along Firmware:
- 121045414867 - Garage, Middle, Second from the left
- 121045415714 - House, Middle, Third from the left

On January 19 this year, Enphase updated ALL of the M190s in our system to the same version of firmware as these two. It seems my characterization of this as "limp-along" firmware was not accurate since only those two microinverters regularly produce the "Grid Gone" events, even after the firmware update. All other microinverters continue to produce very normally with this new firmware.

Anyway, it has been over six months since the last failure in this system. The microinverters don't seem to failing all at once as has been seen with Weatherman's system and a few others.

Any updates from others?

 

[Weatherman]

Called Enphase on Monday of this week to report another failure. That would make two this year and thirteen total failures for my 30-microinverter array.

I've got another half dozen running on reduced power due to grid-gone events. I fully expect to reach the 20 failure mark before the end of the year.


A couple of positive things, however... The replacements haven't cost me anything but time, since both parts and labor are under warranty. And, my array is still running at 100% of what PVWatts predicted when it was first set up. In a little less than three years, it has produced 29.5 MWh of energy. At 11.5 cents/kWh, that's $3,392.50 that I didn't have to hand off to FPL.

 

[dhanson865]

Next, I will create a Secondary Watch list, for the remaining M190 inverters on the house and garage that reported production above 9.85kWh on that

might want to pull that k and the decimal out of the 985 Wh or move the decimal and make it 0.985 kWh

I know you and I know what you meant but for the newcomers trying to figure out the terminology.

 

[DaveEV]

Any updates from others?

All is well, here. No issues to report and system still performing like new.

 

[RegGuheert]

Called Enphase on Monday of this week to report another failure. That would make two this year and thirteen total failures for my 30-microinverter array.

Updated. Your MTBF is now down to 6 years. I'll update the range of MTBFs seen from members (and a few non-members) in the next post.

I've got another half dozen running on reduced power due to grid-gone events.

It's interesting that my two "Grid Gone" whiners behave completely differently from each other. The one that got the firmware because I reported its production had gone practically to zero still complains "Grid Gone" multiple times per day but produces just as much electricity as its neighbors. The other one, which Enphase out of the blue chose to update for no apparent reason has daily periods of low production following its "Grid Gone" reports and therefore underproduces its neighbors. But even this inverter achieves about 95% of the production of nearby inverters.

Clearly, in the case of the first inverter, the new firmware has been quite the magical elixir. That second inverter, however, never underproduced *until* it after got the new firmware. Strange. It sounds from Weatherman's experiences that the new firmware does not prevent failures, however. Rather it seems to just delay the requirement for replacement.

I fully expect to reach the 20 failure mark before the end of the year.

Now I'm not the only one making predictions in this thread. So far I've found these M190s quite difficult to predict!

A couple of positive things, however... The replacements haven't cost me anything but time, since both parts and labor are under warranty.

That's good to hear! Others have reported that Enphase has stopped paying for labor.

At this point I suppose you have about six or seven of the new, black, M215IG-style M190s. Hopefully those will bring your failure rate back down to (or below) original expectations! I still have just the one, but I also now have 12 M215IGs installed. No failures so far! MTBF is >9 years for my M215IGs so far. If the MTBF of my M190s continues to hover around 60 years, the demonstrated reliability of the M215IGs should surpass them in about four more years.

And, my array is still running at 100% of what PVWatts predicted when it was first set up. In a little less than three years, it has produced 29.5 MWh of energy. At 11.5 cents/kWh, that's $3,392.50 that I didn't have to hand off to FPL.

At 55 MWh, my system has produced about $6600 worth of electricity so far and currently produces about $2200 worth each year. I find it quite amazing, really!

might want to pull that k and the decimal out of the 985 Wh or move the decimal and make it 0.985 kWh

Good catch! Fixed!

All is well, here. No issues to report and system still performing like new.

Your MTBF is now up to 91 years. Another six months without a failure will bring you above 100 years.

You can update line 131 with 20 M215's to zero failures as of today. I do have a watch on two of my panels/inverters that have lost some output compared to the others.

Any updates, Nekota? Two M215s out of 20 misbehaving after less than two years is certainly not the norm. I'm interested to hear the current status of your array.

 

[RegGuheert]

Here are updated MTBFs for members here and a couple of other notable systems (only counting replacements, not anomalous function);

M190s:

- Weatherman: 6 years (Your MTBF is still dropping, securing your "honor" of lowest-reliability Enphase installation among LEAF owners here.)
- ecoobsessive: 11 years (This may be wrong, since I do not have a recent update. With D380s, I wouldn't be surprised to see you pass Weatherman.)
- RegGuheert: 59 years (I'm now doing slightly better than the average for the M190s in the spreadsheet, but many sites have not been updated in over a year.)
- drees: over 91 years (No failures and still counting. You get the prize for the oldest Enphase installation here at over 5 years! You also get the award for the next-to-the-lowest-power PV modules at 180W.)
- Tony Williams: over 131 years (Assuming no failures since no report. This is the highest-reliability M190 installation here. BTW, do you have the 225Wp Sunpower panels as I indicated in the spreadsheet?)
- Synertx in Phoenix, AZ: 337 years (Only ONE failure after nearly four years of operation with 90 M190s in Phoenix! That was a year and a half ago with no way to update. It is interesting to see the highest reliability M190 array in the hottest area.)

M190IGs:

I believe these are firmware-derated M215IGs. Weatherman and I have each started receiving these as replacements for failed M190s. I haven't added them to the spreadsheet since it will be hard to track, but I thought I would give them a mention.)

- RegGuheert: over 6 months (Installed September 18, 2014. Still working.)
- Weatherman: over ~2 years (~seven installed to date)

M215s:

- RegGuheert: over 9 years (My 12 M215IGs were installed just over nine months ago. No failures so far!)
- dsinned: 28 years? (No recent updates. Only M215 failure of anyone here, but not in the spreadsheet because I do not have a URL for the array. Also, I will note that this appears to be an infantile failure, so may not actually count for MTBF.)
- QueenBee: over 161 years (Assuming no failures and still counting since you have not reported any. You still get the prize for highest-relibility Enphase installation here.)
- Jall Investments in Phoenix, AZ: over 405 years (No failures and still counting...and that was back in September 3, 2013. If there have still been no failures, then the MTBF is over 730 years today. Unfortunately, I now have no way of knowing whether this is true or not. As of today, this system has already produced over 316 MWh of electricity! Methinks Jall makes good investments!)

I still find it interesting that both of the systems (M190 and M215) with the highest demonstrated MTBFs are in Phoenix, AZ. It is not what I expected, as the original title of this thread will attest to.

 

[Weatherman]

Here's a pic of my array, showing each panel's lifetime energy production:



The ten replacements are very obvious. They are the ones with the significantly lower lifetime production (since they have been in production for a shorter time). The two micros under the panels with the red X's are the ones due to be replaced. One of the others was replaced twice, giving the total replacement count of 13.

The micros in good shape are the once producing from 988 kWh to 1.01 MWh lifetime energy. That's about 12, total, out of the 30. All the rest in the 900s without the "X" mark (lifetime 960 to 976 kWh) are defective but not defective enough for Enphase to agree to replace them.


With the exception of the first replacement, which died in six months, all the other replacements are doing fine. I think about seven out of ten of those are the new micros with the IG.

 

[QueenBee]

You'll be the first to know if I have a failure have friends with about 80 M215s/M250s with no failures as well.

 

[dsinned]

No issues since my last post. All 17 M215s in my 4.25kW array are still in great shape! Only had the one M215 replacement about 1-1/2 years ago due to a slowly deteriorating AC output (or internally degrading, DC to AC conversion efficiency). My system's current lifetime energy production is 101% of the estimated prediction. :mrgreen: