PV life, MTBF, failure modes and maintenance schedule

[RegGuheert]

Really? I actually used 15 amp breakers on my M215 circuits just because I at max output I wasn't even at 80% of 15 amps. I don't remember it being particularly hard to find Siemens (GE style) breakers though for my subpanel though.

Yeah, I couldn't find a single one locally. I ended up buying a box of 6 on eBay. (Home Depot had them for SquareD.)

 

[RegGuheert]

I've already provided links that explain all the details for long service life, including the independent analysis of the capacitors, but since it is still not clear, I will put the calculation for the capacitor life directly into this thread:

The capacitors used are rated for operation at 63V for 8000 hours at 105C. Enphase says the capacitor will operate at a maximum voltage of 40V. They use a maximum temperature of 65C based on Palm Springs, CA, which means the capacitor will operate 40C lower than the rating. Life doubles for every 10C drop in temperature. That means four doublings of life at 65C or 128,000 hours. But the unit is only energized and at temperature when the sun is out, so you can figure that it spends about half it's life at 0V and below 45C, or 512,000 hours at those temperatures. So, in 20 years, the capacitor in Palm Springs will have spent 87,600 hours at high temperatures and the same number of hours at lower temperatures. So, it will have consumed 87,600/128000 + 87,600/512,000 or 86% of its useful life. Where I live, and where most Enphase installations are located, these capacitors will last much longer than in a hot location like Palm Springs.

Since manufacturers define end-of-life at around 80% of capacitance and Enphase has stated that their inverter will operate fine with less than 50% of the original capacitance, just with more movement around the peak power point, the capacitors should be able to last well beyond 20 years.

That is exactly what design for reliability is all about. You derate your components to achieve the reliability you need to meet the market objectives.

All that said, Nichicon has a general note that says not to estimate a life longer than 15 years. Perhaps that is how long they feel the seals will last. It is likely the reason for the 15-year warranty on the M190s.

As noted, there are other stresses involved over a long life, such as thermal cycling. Those are the types of things that can be wrung out in accelerated life tests. The life of the capacitor as well as the life of the unit as a whole have been verified with accelerated life tests.

Please note that a string inverter and an off-grid inverter are two different things. You seem to use them interchangeably. Note that I do not. As you have noted, and DesertDenisen has confirmed, string inverters can actually achieve relatively long lives and decent MTBFs, but they are unable to derate their input capacitors to the degree that microinverters can due to the high voltages, as already discussed.

Most off-grid inverters and charge controllers continue to use fans for cooling and that fact is reflected in their short warranty periods.

The Sandia report on a working group meeting for 100kW and larger inverters has no bearing on a discussion of inverters for home PV systems. True, they have no sway with component manufactures since their volumes are low. Enphase has lots of sway over their component manufacturers since their volumes are high.

Enphase is unable to hide the failures of inverter with their current public websites. That is why I am able to obtain failure data from there. That won't last too much longer, but I already have enough data to see several things:
1) MTBF over a wide range of climates at 20 locations with over 500 inverters installed is quite high. MTBF outside the desert is significantly higher than Enphase's projections.
2) There are some relatively early failures in Phoenix and Tuscon. We have no way to kow if these are random failures or wear-out. With two failures in a single system with only 10 inverters, I suspect that this installation may experience temperatures much hotter than the product rating and some interconnects are failing, but there is really no way for us to know.

This is much more field operational data than I have seen for any other inverter type. I will note that no such public data has been produced for sting inverters or off-grid inverters. As stated, the manufacturers likely do not have good data for their products due to their inability to track failures.

 

[AndyH]

I've already provided links that explain all the details for long service life, including the independent analysis of the capacitors, but since it is still not clear, I will put the calculation for the capacitor life directly into this thread:

<snip>

That is exactly what design for reliability is all about. You derate your components to achieve the reliability you need to meet the market objectives.

All that said, Nichicon has a general note that says not to estimate a life longer than 15 years. Perhaps that is how long they feel the seals will last. It is likely the reason for the 15-year warranty on the M190s.

As noted, there are other stresses involved over a long life, such as thermal cycling. Those are the types of things that can be wrung out in accelerated life tests. The life of the capacitor as well as the life of the unit as a whole have been verified with accelerated life tests.

Can (or have?) you provided proof of independent life tests for Enphase's inverters?

As for the rest, thanks for taking the time, but you could have saved the typing by simply linking to Enphase's marketing paper on capacitor life again. This is not an independent source of info, however. Yes, I said marketing because that's what all of their public website material is.

"Proprietary and Confidential" - really? This is another ploy to provide the appearance of something special or secret. If the doc really was "proprietary and confidential" it would not be on their public website - it would on an internal server or on WikiLeaks. Sorry. This is a message manipulation technique. Why would a company resort to such a tactic? They tell us why in their paper:

However, for micro-inverters to succeed in practical applications and be accepted by the marketplace, the reliability of micro-inverters must demonstrably exceed that of traditional string inverters.

Emphasis mine. (No, I'm not beating on Enphase here - other manufacturers do it too.)

Enphase MUST address capacitor life in some way because the rest of the industry has "raised the BS flag" about their performance claims. One must ask themselves how Enphase managed to hire the only engineers in the world that can turn design upside down. They must have hired the geniuses because the rest of industry - including engineers that designed some of the most reliable units on the streets - cannot understand what Enphase did. Red flag number one.

The paper includes plenty of numbers and other info that appears to be backed by cap manufacturers design documents. But there's a subject that's missing - electrolyte. Under the best conditions, electrolyte is consumed in the capacitor. Seal degrade and electrolyte leaks. Caps have a pressure release valve and electrolyte and gasses vent if the device overheats or operates out of limits. Nichicon, in their electrolytic cap design paper, after working through temperature, ripple, and other life factors, they include this statement:

Typically, fifteen years is generally considered to be the maximum for the estimated life obtained by the above formula.

Why does the component manufacture say the calcs should still be limited to a 15 year max, yet Enphase says the caps will last 25 years? Red flag two. Hmmm...maybe they don't, because:

Because several electrolytic capacitors are paralleled in the micro-inverter, and because capacitors normally fail open, there is in fact very little impact on the micro-inverter if a capacitor fails. The main impact is that MPPT efficiency is slightly reduced because there is now more ripple voltage on the PV bus.

Ok. How much is 'slightly'? Enphase took six pages to tell us that caps are important, and to suggest that they did their homework, yet finish by telling us that caps aren't really that important, and that, essentially, since the cap should fail open and since it's unlikely the end user will notice the loss of efficiency, accepting the probability that caps will fail after 15 years shouldn't negatively affect warranty return rates. Never mind - the whole capacitor lifespan thing is a diversion.

The paper presents a 'paper exercise' that uses climate data from NREL. The numbers in this paper were NOT generated by instrumenting an inverter and putting it on a roof. Why not? Red flag three.

Marketing.

Please note that a string inverter and an off-grid inverter are two different things. You seem to use them interchangeably.

No. They're still inverters. What brought me in to the inverter discussion from the start is precisely because 'inverters' and inverter reliability numbers are used interchangeably. I came here to highlight that all inverter numbers are NOT the same. Enphase and other microinverter companies MARKETING departments want to lump all the numbers together - the goal of marketing is to provide the worst case info for their competition and provide the best case for their products. It's clear that much of the dialogue here is based on sales-speak not tech fact.

The equipment that interfaces with the panels in an off-grid system is the charge controller. They 'only' have to do DC-DC. They're known to last much longer than their warranty as well - like the 30+ year example referenced earlier.

Note that I do not. As you have noted, and DesertDenisen has confirmed, string inverters can actually achieve relatively long lives and decent MTBFs, but they are unable to derate their input capacitors to the degree that microinverters can due to the high voltages, as already discussed.

Except here as in other places, you do!

By "high voltages" I assume you mean the devices that accept PV inputs up to 600VDC? That's fine! But not all central inverters run to those levels. Seems to me that again, the brush is a bit wide. Seems we could extract lifetime numbers for the central inverters that keep a wider voltage margin. But lumping all inverters together does provide better marketing for the microinverter folks, doesn't it?

There are three major inverter types - string, central, and micro. Micros are one per panel, strings handle one string of PV, and central handle multiple strings. But how do we know, since an inverter that can accept inputs from, say, six strings, is likely still combining them internally into a single input?
http://www.pv-system-tech.com/fileadmin/user_upload/material2012/InSyst2012_Screen.pdf

I agree completely that "all" inverter reliability numbers should not be used in blanket statements because the majority of inverters in the field are not grid connected - they're used in cars (EV, hybrid), they're used in ships, trains, RVs and off-grid buildings. None of those are directly connected to PV inputs as grid-tied inverters are. The off-grid analog is the charge controller, not the inverter. And we already know that charge controllers last longer than the commonly used "5-10 years" for inverters...

You agree that central inverters CAN

can actually achieve relatively long lives and decent MTBFs

in SPITE of not being able to derate caps to magic Enphase levels. Good.

But this is all a huge red herring. Why? Because the central inverter world, both off-grid and large scale grid tied, sees capacitors as a maintenance item and designs for repairability. Enphase, on the other hand, is willing to accept the loss of efficiency when the caps fail in their non-serviceable assemblies. And, again, it's NOT about warranty, as Enphase's primary customers - folks like us and our parents - are not at ALL likely to notice the efficiency deterioration from years 12 thru 25 either because they're not looking, or because it's lost in the noise of panel and wiring deterioration and is thus deniable by Enphase.

Most off-grid inverters and charge controllers continue to use fans for cooling and that fact is reflected in their short warranty periods.

Yes, many do use fans. Not all do, however. Outback has VFX...models (vented/fan cooled) as well as FX... (sealed, not fan cooled). They both live long lives in the field. And they both have the same warranty length.

Outback's cast aluminum case and heat transfer process is highlighted as a 'thing that works well' in Sandia and NREL reports - was adopted by microinverter companies. I wonder who invented that lumpy black case?
http://www.midnitesolar.com/pages/frontPage/nwHistory/history.php

Additionally, the commonly used off-grid equipment is more than an inverter - my Outback VFX3524, for example, is an inverter, a battery charger, generator controller, and transfer switch. This equipment is doing more than 'just' invert.

Grid-tied equipment uses a combination of cooling methods as well - convection, forced air, and liquid are all used.

The Sandia report on a working group meeting for 100kW and larger inverters has no bearing on a discussion of inverters for home PV systems. True, they have no sway with component manufactures since their volumes are low. Enphase has lots of sway over their component manufacturers since their volumes are high.

Yes, you're correct, that particular working group does not reflect the entire industry. But that's not the source for my comment. That fact came from a number of other working groups hosted by Sandia and NREL that DO reflect the entire inverter spectrum - including automotive. While your faith in the company is admirable, there is absolutely no way that Enphase's ~12% of residential generation, has more pull with component manufacturers that the companies serving the remaining 88% of the residential market, plus commercial, plus industrial scale, plus mobile inverters, plus the automotive/hybrid/EV sector have.

Enphase is unable to hide the failures of inverter with their current public websites. That is why I am able to obtain failure data from there.

Two points. First, from a tech perspective, they are the ONLY entity that can filter the data they display because they are the only entity that has full control of the data. It would be amazingly easy for them to filter the data. Second, they likely won't have to filter as moving to a pay system and removal of publically-accessible data allows them to fragment/hide/mask any failures that appear as their systems age much better - and they're paid to do it! Brilliant marketing!

That won't last too much longer, but I already have enough data to see several things:
1) MTBF over a wide range of climates at 20 locations with over 500 inverters installed is quite high. MTBF outside the desert is significantly higher than Enphase's projections.
2) There are some relatively early failures in Phoenix and Tuscon. We have no way to kow if these are random failures or wear-out. With two failures in a single system with only 10 inverters, I suspect that this installation may experience temperatures much hotter than the product rating and some interconnects are failing, but there is really no way for us to know.

Sure - when all the data available is good, it can mean all the systems are good, or it could mean only the good systems are represented. Do you know? Do you have ANY way to confirm that you're looking at a full dataset?

This is much more field operational data than I have seen for any other inverter type. I will note that no such public data has been produced for sting inverters or off-grid inverters. As stated, the manufacturers likely do not have good data for their products due to their inability to track failures.

I cannot agree with this from at least two directions. First, for you to suggest that manufacturers don't have failure data, one would have to accept that warranties are not claimed, that vendors are not tracking returns, and that more than 40 years of info just doesn't exist. Sorry, no. Secondly, the data you suggest does not exist does - but it takes more work to find. Much less in the internet age, however - but I agree, it's not as easy as sitting on the couch and looking over an internet feed. How do we find reliability info? With info from manufacturers (limited to what they disclose publicly), from conversations at meetings, conferences, etc. (some public, some not...), from equipment vendors, from installers, and from end-users. The community does a good job of providing the info you seek - here are two examples:

Solar-Guppy and others were working with Xantrex when they released their GT grid-tie inverters. Xantrex didn't have a good rep and wanted to show they'd improved. There was some improvement, but not all was rosy:
http://sgtechnology.dyndns.org/forum/solar-powered-gridtie-inverters-f19.html
Note the topic headings for the GTs: "smoked", "weird issues", "communications problems" etc.
(This was feedback from a small group actively beta-testing an inverter - it wasn't the typical internet enthusiast crowd.)

On the other hand, there are a number of inverter manufacturers that stand head and shoulders above the rest as reflected by this (this is not a sole source for this message by a long shot):

We have been selling Exeltech inverters for over 15 years, and we have never had one single report of any failure of any type. In fact, we have never even heard of one going bad that was not grossly abused (like hooking the battery terminals to AC power..).

http://www.solar-electric.com/exsiwain.html


The information you seek is available, but one has to look for it.

Enough of all this. Enphase has the warranty they have. Time will tell if they or their equipment is in the game for the long haul. But we've been here before.

Our job, seems to me, is to get whatever we want from our equipment. I prefer equipment that can be serviced, because there is nothing created by man that lasts forever. I'm not a fan of disposable tech. It doesn't matter, though, because Enphase doesn't serve my segment of the market anyway.

All equipment has a design regime, zones of acceptable and unacceptable performance, strengths and weaknesses.
http://solarbridgetech.com/wp-content/uploads/2011/05/SLB_E_Design_Reliability.pdf

We have a choice - we can either learn about the solutions and make our choice, or we can work to find an honest sales rep and hope they have the integrity to tell us that their produce is not the best fit for us if that's the right answer. It can be more difficult to find an honest sales rep than to find failure data for a Chinese inverter on eBay...

edit...spelling, typos, quote tags...

 

[RegGuheert]

Can (or have?) you provided proof of independent life tests for Enphase's inverters?

I have. The contact information is included in the documentation for verication.

Have you provided ANY evidence or analysis from the manufacturer or anyone else that the Classic Controller or the OutBack inverters can achieve a long life? No, you haven't. All you have done is you have claimed that you met some unnamed people at some unnamed internet location who had good experiences. All we have at this point is two five-year warranty. Those two companies have chosen not to stand behind their products for longer than that. Nearly everyone understands why the they do not back their products longer. You can say it is all marketing, but what kind of marketing causes a company to relinquish the market over a warranty issue if they have a far superior product?

As for the rest, thanks for taking the time, but you could have saved the typing by simply linking to Enphase's marketing paper on capacitor life again. This is not an independent source of info, however. Yes, I said marketing because that's what all of their public website material is.

Actually, the approach Enphase used is standar industry practice for determining the life of electrolytic capacitors. It is verified by an independent evaluator that also did accelerated life testing. You apparently didn't read that.

Enphase MUST address capacitor life in some way because the rest of the industry has "raised the BS flag" about their performance claims.

Check the dates on the Enphase and independent verification documents. They predate any claims of BS that I know about.

One must ask themselves how Enphase managed to hire the only engineers in the world that can turn design upside down.

Let's see, they designed an inverter application with the capacitor application having a 30-year wear-out, limited to 15 years by the capacitor manufacturer. The product gets a 15-year warranty. It's all very reasonable.

They must have hired the geniuses because the rest of industry - including engineers that designed some of the most reliable units on the streets - cannot understand what Enphase did.

They understand exactly what Enphase did and that Enphase is currently eating their lunch. They are simply spreading FUD. It is a common marketing approach when your business is knocked on its heels by a competitor. This is business as usual.

The paper includes plenty of numbers and other info that appears to be backed by cap manufacturers design documents. But there's a subject that's missing - electrolyte. Under the best conditions, electrolyte is consumed in the capacitor. Seal degrade and electrolyte leaks.

Actually, the Nichicon paper that Enphase references covers that subject in some detail. The electrolyte is apparently also consumed in the process of "healing". More healing occurs under high-voltage operation, which is why the string inverters have more difficulty achieving a very long life.

Caps have a pressure release valve and electrolyte and gasses vent if the device overheats or operates out of limits.

As shown in the analyses, the electrolytic caps in the Enphase inverter are VERY far from the point where the vents would open.

Nichicon, in their electrolytic cap design paper, after working through temperature, ripple, and other life factors, they include this statement:

Typically, fifteen years is generally considered to be the maximum for the estimated life obtained by the above formula.

Why does the component manufacture say the calcs should still be limited to a 15 year max, yet Enphase says the caps will last 25 years?

But they only warranty the M190 product for 15 years. The newer M215 product has the 25-year warranty. I think it was after this product came out that there have been a lot of voices crying "BS". I don't have visibility of what they have done, so I cannot say. But my guess is that they have gone to their capacitor manufacturers and asked what could be done to get their components to a 25-year life. It's clear that the application supports the long life, but there seems to be something that Nichicon feels limits these caps to 15 years. The bottom line is that it seems clear that a 15-year life is very well-established. As far as getting from 15 years to 25 years, I suspect it is one of three possibilities:
1) It is all marketing BS as AndyH claims. If so, they have less than 20 years before the bill comes due.
2) Enphase sat down with Nichicon and said "We need a 25-year life on your capacitors, but we are held back by this vague limit in your application guide." Given Enphase's status as a large and growing customer buying literally millions of these caps, Nichicon looked carefully at the Enphase application and determined, together with Enphase, that the capacitors actually WILL reliably last for 25 years in their design and application
3) Same as 2) except Nichicon came back and redesigned the 15-year failure mode out of their product just for the Enphase application.

There is no way I can know which one it is, but I will say that if they did the first one, it was a suicidal mistake given they actually warranty the products for that long. I suspect it is either 2) or 3).

The paper presents a 'paper exercise' that uses climate data from NREL. The numbers in this paper were NOT generated by instrumenting an inverter and putting it on a roof. Why not?

Clearly you didn't read the independent analysis paper. The author clearly stated that Enphase had provided measured data from instrumentation inside inverters in Palm Springs to verify the temperature assumptions.

The equipment that interfaces with the panels in an off-grid system is the charge controller. They 'only' have to do DC-DC.

It's true. But the Classic Controller the you (and I) have bought have fans which I think we both agree need replacement every five years to avoid a catastrophic failure. But they also have electrolytics capacitors inside. I see five of them. If they have been well-designed into the application, they can possibly reach the 15-year design limit that Nichicon sees. If they are not well-designed in the application, they should fail before that point. Since Midnite only warrants five years of operation, not analysis is needed to convince the customer that these will last for five years.

But AndyH, you have come here and claimed the units would last for 20 years if you replace the fans. I see no such statement from Midnite anywhere. i see no discussion of how these capacitors are used and how an insependent reliability expert has done additonal anaylysis and testing to verify a long life. Instead, they provide a 5-year warranty.

They're known to last much longer than their warranty as well - like the 30+ year example referenced earlier.

I must have missed the refernce. No matter. There are no thirty-year-old MPPT charge controllers out there. Those didn't start to show up in the market until the 1990s. I have a whole pile of dead ones in my attic from one of the earlier attempts.

Don't get me wrong, I think the Classic is the best of the breed and it is very well-designed. That's why I chose it for an off-grid application. But I'll be extremely surprised if it outlives most of the microinverters on my roof. Time will tell.

Note that I do not. As you have noted, and DesertDenisen has confirmed, string inverters can actually achieve relatively long lives and decent MTBFs, but they are unable to derate their input capacitors to the degree that microinverters can due to the high voltages, as already discussed.

Except here as in other places, you do!

No, I didn't. I have only ever seen you conflate the terms string inverter and off-grid inverter. I use the term central inverter to refer to both types because they both share the unfortunate fact of having a central system impact.

By "high voltages" I assume you mean the devices that accept PV inputs up to 600VDC?

I mean string inverters, just as I stated.

That's fine! But not all central inverters run to those levels. Seems to me that again, the brush is a bit wide. Seems we could extract lifetime numbers for the central inverters that keep a wider voltage margin. But lumping all inverters together does provide better marketing for the microinverter folks, doesn't it?

Fair enough. Your inverter likely has a lower input voltage than the Enphase microinverters. But it also is unlikely to live nearly as long as the microinverters. If you do the analysis on the electrolytic capacitors in that unit, you will find that they at on all the time and they likely operate at similar, if not higher, temperatures than those found in Enphase inverters on the rooftop. In any case, they are subject to the same 15-year list which is in the Nichicon paper.

You agree that central inverters CAN

can actually achieve relatively long lives and decent MTBFs

Sting inverts. Here is my actual quote:

As you have noted, and DesertDenison has confirmed, string inverters can actually achieve relatively long lives and decent MTBFs, except they are unable to derate their input capacitors to the same degree as microinverters due to the high voltages, as previously noted.

But this is all a huge red herring. Why? Because the central inverter world, both off-grid and large scale grid tied, sees capacitors as a maintenance item and designs for repairability.

Agreed.

Enphase, on the other hand, is willing to accept the loss of efficiency when the caps fail in their non-serviceable assemblies. And, again, it's NOT about warranty, as Enphase's primary customers - folks like us and our parents - are not at ALL likely to notice the efficiency deterioration from years 12 thru 25 either because they're not looking, or because it's lost in the noise of panel and wiring deterioration and is thus deniable by Enphase.

Let's see...Enphase inverters offer an efficiency of around 95% while your Outback inverter peaks around 90%. Considering a grid-tied system harvests well over 90% of the electricity coming from the panels while an off-grid system might achieve 25% to 50%, I don't think a drop of a couple of percent at the end of the life of the inverter is a terrible thing.

Most off-grid inverters and charge controllers continue to use fans for cooling and that fact is reflected in their short warranty periods.

Yes, many do use fans. Not all do, however. Outback has VFX...models (vented/fan cooled) as well as FX... (sealed, not fan cooled). They both live long lives in the field. And they both have the same warranty length.

They've only been in the field for ten years and they contain the same electrolytic capacitors that you have villified Enphase for using. By Nichicon's definition, they will not live beyond 15 years no matter what. But beyond that, they are on 24/7. Pleas show me the analysis that indicates the many, many electrolytic capacitors in the Outback inverters will last for the 20 years that you have claimed, even though the manufacturer only back you up for five.

They plan to make good money selling parts. It's a poor business model, IMO, which has run its course.

While your faith in the company is admirable, there is absolutely no way that Enphase's ~12% of residential generation, has pull with component manufacturers that the companies serving the remaining 88% of residential, plus commercial plus industrial scale plus mobile inverters plus the automotive/hybrid/EV sector has.

You're kidding, right? Enphase sells an inverter for each PV panel. They have literally sold MILLIONS of them, each one with four of the Nichicon capacitors. Yes, anyone who purchases 10 million components within a four-year timeframe will have sway with their suppliers. Can you name any other inverter company who has used 10 million components of a single type in their products? How about 100,000?

Enphase is unable to hide the failures of inverter with their current public websites. That is why I am able to obtain failure data from there.

Two points. First, from a tech perspective, they are the ONLY entity that can filter the data they display because they are the only entity that has full control of the data. It would be amazingly easy for them to filter the data.

If they could, they would, but they can't, so they don't. Trust me, I've asked them to filter the data on my array to show the production from the failed inverter together with the new inverter, but they cannot.

Second, they likely won't have to filter as moving to a pay system and removal of publically-accessible data allows them to fragment/hide/mask any failures that appear as their systems age much better - and they're paid to do it! Brilliant marketing!

The history is that they moved from a pay system to a free (pay up-front) system because the business model was flawed Because it did not match PV buyers' goals.

They are not changing the business model, just the website. I link both the old and new websites in my spreadsheet. Notice that there is no per-module data in the new website. Individual owners can still track everything on their secured site.

Sure - when all the data available is good, it can mean all the systems are good, or it could mean only the good systems are represented. Do you know? Do you have ANY way to confirm that you're looking at a full dataset?

My confidence level is very high that the data is accurate. Note that many of these systems are owned by posters here. The point is that there is NOT a full dataset after a failure, since Enphase ONLY shows data per inverter serial number. The failed inverter is removed from the system to make room for the new inverter's data.

First, for you to suggest that manufacturers don't have failure data, one would have to accept that warranties are not claimed, that vendors are not tracking returns, and that more than 40 years of info just doesn't exist. Sorry, no.

Five-year warranty. Once the warranty ends, the manufacturer loses ALL visibility of failures. Sorry, but the five-year warranty makes it impossible for the manufacturer to track long-term issues with their own products. SMA has finally been forced to offer a 20-year warranty (for a fee), but since they do not have online tracking of all characteristics of their fielded products, they are still way behind Enphase.

Secondly, the data you suggest does not exist does - but it takes more work to find. Much less in the internet age, however - but I agree, it's not as easy as sitting on the couch and looking over an internet feed. How do we find reliability info? With info from manufacturers (limited to what they disclose publicly), from conversations at meetings, conferences, etc. (some public, some not...), from equipment vendors, from installers, and from end-users. The community does a good job of providing the info you seek - here are two examples:

-snip-

The information you seek is available, but one has to look for it.

A bunch of anecdotal information from a self-selected set of users is hardly useful for determining things like product wear-out and MTBF.

Enough of all this. Enphase has the warranty they have. Time will tell if they or their equipment is in the game for the long haul. But we've been here before.

And Outback and Midnite still offer the same 5-year warranties that were commonplace in this industry for over a decade. SMA is moving up the reliability curve and they continue to compete effectively.

Our job, seems to me, is to get whatever we want from our equipment. I prefer equipment that can be serviced, because there is nothing created by man that lasts forever. I'm not a fan of disposable tech. It doesn't matter, though, because Enphase doesn't serve my segment of the market anyway.

Fair enough.

We have a choice - we can either learn about the solutions and make our choice, or we can work to find an honest sales rep and hope they have the integrity to tell us that their produce is not the best fit for us if that's the right answer. It can be more difficult to find an honest sales rep than to find failure data for a Chinese inverter on eBay...

True enough. I've had my share of troubles with off-grid inverters and charge controllers to bias my opinion and you have likewise dealt with microinverter problems. Both have gotten better with improved design and components. The PV market is maturing and growing, and that is the main goal.

Here's hoping that all the PV systems out there last for at least 20 years!

 

[AndyH]

Can (or have?) you provided proof of independent life tests for Enphase's inverters?

I have. The contact information is included in the documentation for verication.

I'm sorry to report that I missed your grand summary post on page 2. While I did appreciate reading the report by the gents hired by the venture capitalists, I hope you're not disappointed when I suggest that's not independent. In my experience (such as it is) when an electronics, PV, battery, or other energy-related manufacturer wants independent validation of their products, they send them to a national lab or one of the independent commercial test facilities. I've reports from Sandia on the lithium cells I used to import, for example. I hoped for something that confirms the total performance the devices from a truly independent source. I'd like to see results from accelerated testing on all inverters, but that's not likely to happen.

Note that the #1 criticism about the devices from the balance of the industry is on the capacitors. Note that the VC team didn't want an analysis of semiconductor life or heat dissipation, or optoelectronics, they wanted to know if the caps could survive. I'm not saying that the device is bad, or that the cap choices are bad - I'm only saying that Enphase was forced to provide these papers because they addressed specific criticisms from industry and potential customers. That's also why they had to push the warranty to 25 years - they'd certainly much rather have the warranty tail the industry standard 5 years long. It's a business decision forced on them, not a show of altruism. The engineering was driven by the warranty, not the other way around, because products are engineered to be just good enough and not a penny more. That's basic business.

Have you provided ANY evidence or analysis from the manufacturer or anyone else that the Classic Controller or the OutBack inverters can achieve a long life? No, you haven't. All you have done is you have claimed that you met some unnamed people at some unnamed internet location who had good experiences.

I'll answer for me, thanks. No Reg - this is not correct. What I have said a number of times is that in order to estimate system life I 1. went back to the designers so that I could evaluate their design goals and how well they think they achieved them, and 2. sought folks that are successfully doing what I intend to, in order to learn of their experiences. This is required because some of the equipment is too new - are are no 20+ year field studies available for much of what I've chosen. The only reason I brought that into the conversation was to point out that we are looking at similar problem from different directions.

The paper includes plenty of numbers and other info that appears to be backed by cap manufacturers design documents. But there's a subject that's missing - electrolyte. Under the best conditions, electrolyte is consumed in the capacitor. Seal degrade and electrolyte leaks.

Actually, the Nichicon paper that Enphase references covers that subject in some detail. The electrolyte is apparently also consumed in the process of "healing". More healing occurs under high-voltage operation, which is why the string inverters have more difficulty achieving a very long life.

As for "string inverters" (and central inverters), I agree completely! Apparently utility-scale central inverter manufacturers do as well - that's why many of them do not use electrolytic capacitors at all. There's been a two-pronged push in the inverter industry - one to bring prices down (Sunshot program), and the other to increase reliability. Manufacturers understand that optical devices, e-caps, and IGBTs are wear items.

I asked you about Enphase, not any other product. In order to achieve a long capacitor life, at least some of the electrolyte should stay inside and active. The fact remains, as Enphase has clearly stated, that they do not expect the caps to live beyond 15 years even in their products with a 25 year warranty. As you've already stated, you'll accept an efficiency hit later in the life of the inverter.

That's fine - I'm glad we've gotten here! As we've agreed earlier, no device is perfect and designers must design around the strengths and weaknesses of the parts with which they're working. It appears that, at least as of Feb 2010, the folks at Enphase are also satisfied with a loss of efficiency and solar harvest when a customer uses their inverters:

For grid-scale applications:

You don't have to go out there and fix a broken module or fix a broken inverter until you're on [the next scheduled service visit...]

Home-scale systems:

In the unlikely event you have a failure, you don't have to service it for a year because the loss of energy is so small...you can choose to fix it whenever it's convenient.

Note this comment from a utility-scale central inverter manufacturer from the same interview. It gives insight into their concerns and focus as well as their reliability goals:

Reliability ultimately is about wear-out if things are well made. It's about components, it's not about voltages and it's not semiconductors, it's about components. Satcon has not sold a product in our history that used a single electrolytic capacitor. We've not sold a product in at least 10 years that's used any optical components or optical isolators. It's well known that if you use those sort of components in hot environments and in environments that have temperature cycling that you have a fundamental limit to the lifetime of the product...Satcon is very comfortable giving a 99% up-time guarantees...Extensive studies have shown that inverters today, big inverters, exceed 99% uptimes, this is by independent people.

Feb 4, 2010, number 54 on the I-Tunes list.
http://www.renewableenergyworld.com...-vs-central-inverters-is-there-a-clear-winner
https://itunes.apple.com/us/podcast/inside-renewable-energy-mp3/id135363982" onclick="window.open(this.href);return false;

The paper presents a 'paper exercise' that uses climate data from NREL. The numbers in this paper were NOT generated by instrumenting an inverter and putting it on a roof. Why not?

Clearly you didn't read the independent analysis paper. The author clearly stated that Enphase had provided measured data from instrumentation inside inverters in Palm Springs to verify the temperature assumptions.

This is incorrect, Reg. Neither the paper you linked, nor Enphase's "Proprietary and Confidential" version use live inverter data.

http://enphase.com/wp-uploads/enphase.com/2011/11/EnphaseElectrolyticCapacitorLife092908.pdf
http://enphase.com/downloads/ElectolyticCapacitorLife092908.pdf

Both clearly state that while they did use climactic data for Palm Springs, it came from:

The climatic and irradiance data used in this test was obtained from the National Solar Radiation Data Base (NREL), which provides hourly meteorological data for locations throughout the United States.

While I'm not suggesting that the calculated result will be statistically insignificant when compared with actual performance data for an array in Palm Springs, it is clear that the capacitor evaluation exercise was nothing more or less than a paperwork exercise. No actual capacitor temperatures were used.

The equipment that interfaces with the panels in an off-grid system is the charge controller. They 'only' have to do DC-DC.

It's true. But the Classic Controller the you (and I) have bought have fans which I think we both agree need replacement every five years to avoid a catastrophic failure. But they also have electrolytics capacitors inside. I see five of them. If they have been well-designed into the application, they can possibly reach the 15-year design limit that Nichicon sees. If they are not well-designed in the application, they should fail before that point. Since Midnite only warrants five years of operation, not analysis is needed to convince the customer that these will last for five years.

But AndyH, you have come here and claimed the units would last for 20 years if you replace the fans. I see no such statement from Midnite anywhere. i see no discussion of how these capacitors are used and how an insependent reliability expert has done additonal anaylysis and testing to verify a long life. Instead, they provide a 5-year warranty.

Incorrect again, Reg. I NEVER claimed that my charge controller would last for "20 years if I replace the" fan. What I have said a number of times is that I will keep a spare fan and a set of replacement capacitors in a maintenance box in case they should be needed later. I did say, however, that I intend to maintain the device so that it provides a 20 year real-world lifespan. Besides, the Classic can be run in a sealed configuration - fanless - for only a slight loss of power (For a 24 volt battery, max current drops from 94A to 80A). Since I am already derating the entire system to 80% max (45% for the charge controller), the fan is not a factor, dead or alive.

Maybe, as with Enphase, I'll decide in 12 years that there is an acceptable amount of ripple on the PV line and simply leave the device on-line. It's good to have options, yes?

They're known to last much longer than their warranty as well - like the 30+ year example referenced earlier.

I must have missed the refernce. No matter. There are no thirty-year-old MPPT charge controllers out there. Those didn't start to show up in the market until the 1990s. I have a whole pile of dead ones in my attic from one of the earlier attempts.

I didn't say MPPT charge controllers, did I? That's a significant part of the problem here, Reg, because some in the microcontroller camp, likely only those with a marketing bent, feel it's ok to use reliability or warranty information from very old tech in order to make their devices look better. The fact remains that some old tech charge controllers, off-grid inverter/chargers, and grid tied devices - including microinverters - had horrible records! But not all - and this is important! There are plenty of devices that have 30+ years of real-world fault-free service.

But this is all a huge red herring. Why? Because the central inverter world, both off-grid and large scale grid tied, sees capacitors as a maintenance item and designs for repairability.

Agreed.

Enphase, on the other hand, is willing to accept the loss of efficiency when the caps fail in their non-serviceable assemblies. And, again, it's NOT about warranty, as Enphase's primary customers - folks like us and our parents - are not at ALL likely to notice the efficiency deterioration from years 12 thru 25 either because they're not looking, or because it's lost in the noise of panel and wiring deterioration and is thus deniable by Enphase.

Let's see...Enphase inverters offer an efficiency of around 95% while your Outback inverter peaks around 90%. Considering a grid-tied system harvests well over 90% of the electricity coming from the panels while an off-grid system might achieve 25% to 50%, I don't think a drop of a couple of percent at the end of the life of the inverter is a terrible thing.

Off track again already? Haven't you been telling me it's wrong to bring off-grid inverters into the mix? I'm confused! :lol:
Peak for the VFX3524 is 92% VS the 96.3% peak for an M215. This is a bogus comparison though, Reg, and I suspect you know that.

Firstly, if we want to compare PV harvest efficiency, we'll need to put an array with micros up against an array with a charge controller. But even that's not a great comparison, because while they both perform MPPT and some level of DC/DC conversion, the micro is also inverting.

Secondly, the goals and real-world experience for grid-tied and off-grid folks are different. Many grid-tied operators want to put as much sun on the grid as possible. Others want enough to net-zero every 12 months. Off-grid folks, on the other hand, want full batteries daily. Once the battery is full, any energy not used is wasted. On good sun days when the battery's full, we go hog-wild and run the waffle iron and the washing machine and the welder. :lol:
http://www.homepower.com/articles/toast-pancakes-and-waffles

Third, when I'm pulling power from the inverter when the sun's out, the path is PV-charge controller-inverter-outlet and in that regime, I expect the inverter's 92% efficiency to combine with the charge controller's ~96% to give me about 88%. I completely agree that efficiency is lower when the sun's down. Flooded lead acid is close to 100% (edit..charge efficiency - discharge is less and depends on current) when below 80% state of charge, so at best nighttime efficiency is about 87%. Off grid efficiency is not compared with grid-tied, however - it's compared with the grid it's severed from. And an off grid system on a bad day is more efficient than the grid on a perfect day.

Most off-grid inverters and charge controllers continue to use fans for cooling and that fact is reflected in their short warranty periods.

Yes, many do use fans. Not all do, however. Outback has VFX...models (vented/fan cooled) as well as FX... (sealed, not fan cooled). They both live long lives in the field. And they both have the same warranty length.

They've only been in the field for ten years and they contain the same electrolytic capacitors that you have villified Enphase for using. By Nichicon's definition, they will not live beyond 15 years no matter what. But beyond that, they are on 24/7. Pleas show me the analysis that indicates the many, many electrolytic capacitors in the Outback inverters will last for the 20 years that you have claimed, even though the manufacturer only back you up for five.

They plan to make good money selling parts. It's a poor business model, IMO, which has run its course.

I haven't villified anyone for anything, just trying to find the connection between a 15 year part and the 25 year warranty that you claim is equivalent to service life.

Off-grid inverters are generally not pushing power 24/7, by the way, Even hard-core homeowners do have to sleep. More efficiency-oriented off-gridders use a small inverter to handle the small 24/7 loads so that the primary unit can be sure to sleep as well. Off-grid inverters get some of the capacitor-extending beauty sleep that grid-tied inverters are reported to get. :lol:

Hang on - they plan to make good money selling parts? Parts for what - their inverters that you say will die soon after their fifth year in operation? But - didn't you say that the string/central/off-grid inverter manufacturers don't have any failure visibility after the 5th year? Seems to me it would be a pretty brain-dead corporation that was not tracking owners of their junk if they planned an income stream from parts sales! Seriously - get on the phone - maybe they'll give you a reward for your suggestion! :lol:

While your faith in the company is admirable, there is absolutely no way that Enphase's ~12% of residential generation, has pull with component manufacturers that the companies serving the remaining 88% of residential, plus commercial plus industrial scale plus mobile inverters plus the automotive/hybrid/EV sector has.

You're kidding, right? Enphase sells an inverter for each PV panel. They have literally sold MILLIONS of them, each one with four of the Nichicon capacitors. Yes, anyone who purchases 10 million components within a four-year timeframe will have sway with their suppliers. Can you name any other inverter company who has used 10 million components of a single type in their products? How about 100,000?

Seriously? Another course change? What I reported is that the ENTIRE INVERTER INDUSTRY is not large enough to sway capacitor manufacturers. It doesn't matter what any single manufacturer does in this instance as it's still a relatively small overall industry.

As far as I'm concerned, Reg, my desired mission is complete. I think we've agreed that off-grid inverters are not performing an equivalent task with any type of grid-tied inverter, whether micro, string, or central. Because of that, I don't think that pulling off-grid stats into a microinverter discussion is useful. By extension, and more to the point, I think it's very useful to understand that it's not useful to point to inverters with fans as there are plenty of devices on the streets that do not use fans and also ship with a 99% uptime guarantee.

Additionally, I think we agree that electrolytic capacitors are a limiting component in power electronics and that devices that do not use them have the potential of being more reliable than equipment that incorporates them - even if they provide redundant parts.

Finally, I think we agree that one should evaluate an entire design - including pros and cons of alternate architectures - before making the final decision about which tech to select.

 

[RegGuheert]

The fact remains, as Enphase has clearly stated, that they do not expect the caps to live beyond 15 years even in their products with a 25 year warranty.

Where in the world did you get that from??? Please provide links to the actual Enphase statements if you want to make claims like this.

- Enphase has always stated that they expect their caps to live much longer than 20 years.
- As I said in my previous post, all of this discussion pertains to the M190, which has a 15-year warranty.
- We have NO IDEA if the M215 even contains a single electrolytic capacitor or not. That is the only product Enphase sells with a 25-year warranty. I have tried to determine if it does, but I cannot find an answer.

No actual capacitor temperatures were used.

From the independent analysis of the capacitor by a reliability expert:

Enphase has made some measurements external to the capacitor during actual application that indicated the surface temperature to be a maximum of 65oC.

What I have said a number of times is that I will keep a spare fan and a set of replacement capacitors in a maintenance box in case they should be needed later.

Electrolytic capacitors die faster if they are operated at 0V than in an actual application. So if you purchase some, put them in a box and pull them out when capacitors in the unit have died the spares will likely have been long dead.

Maybe, as with Enphase, I'll decide in 12 years that there is an acceptable amount of ripple on the PV line and simply leave the device on-line. It's good to have options, yes?

??? Do you have something that tells you that the electrolytic capacitors in your Classic Controller will last for 12 years? Something that tells the operating temperatures and shows a detailed life calculation for the application? Do you have something that says that a failure of any one of these capacitors will only lead to lower operating efficiencies? More likely, it will result in failure of the unit, possibly even cascading failures.

Off-grid folks, on the other hand, want full batteries daily. Once the battery is full, any energy not used is wasted.

That's my point. Energy harvest of off-grid systems is extremely low compared to grid-tied systems.

But - didn't you say that the string/central/off-grid inverter manufacturers don't have any failure visibility after the 5th year?

Yep. Even if they sell repair parts, it does not mean that they will get a good picture of their product reliability. Most failures after the warranty period are likely junked since they are outdated by that point. That's my experience, anyway.

Because of that, I don't think that pulling off-grid stats into a microinverter discussion is useful.

Microinverter discussion? This is not a thread about microinverters.

Additionally, I think we agree that electrolytic capacitors are a limiting component in power electronics and that devices that do not use them have the potential of being more reliable than equipment that incorporates them - even if they provide redundant parts.

Component selection is an important part of reliability, but it all depends critically on the details of the design and the manufacturing. Just because an Enphase competitor offers inverters without electrolytic capacitors does NOT mean that it will be a more reliable product. Enphase has achieved a very high MTBF on FIELDED units. And they have done the analysis and life testing to show a very long life on their capacitors.

I will predict that many of the inverter solutions coming out today which claim to have long life because they eschew electrolytic capacitors will be unreliable compared with Enphase inverters.

Finally, I think we agree that one should evaluate an entire design - including pros and cons of alternate architectures - before making the final decision about which tech to select.

That's what this thread is about. I have compared and contrasted two different architectures to help people see the trade-offs they are making with their system choices. DesertDenisen has added in real, long-term experience with a third type of system showing outstanding results.

 

[AndyH]

The fact remains, as Enphase has clearly stated, that they do not expect the caps to live beyond 15 years even in their products with a 25 year warranty.

Where in the world did you get that from??? Please provide links to the actual Enphase statements if you want to make claims like this.

- Enphase has always stated that they expect their caps to live much longer than 20 years.

I agree that you're reflecting the Enphase marketing message. I must also point out that their same marketing documents direct the reader to vendor guidelines that show that calendar life is ALSO a factor. Marketing is powerful, but the laws of physics are still in play even if not specifically noted in the advertisement.

- As I said in my previous post, all of this discussion pertains to the M190, which has a 15-year warranty.
- We have NO IDEA if the M215 even contains a single electrolytic capacitor or not. That is the only product Enphase sells with a 25-year warranty. I have tried to determine if it does, but I cannot find an answer.

I don't expect anyone considering a new install will specify old tech devices. I'll see if I can find an answer.

No actual capacitor temperatures were used.

From the independent analysis of the capacitor by a reliability expert:

Enphase has made some measurements external to the capacitor during actual application that indicated the surface temperature to be a maximum of 65oC.

Two points. First, no actual measurements were of any type were made for this 'study' - that makes it a paperwork exercise and nothing more.*** Second, electrolytic capacitors exposed to ripple heat from the inside, and the dielectric inside the cap is a very effective thermal insulator. Third and last, under what conditions were the 'some measurements' collected? With or without potting? With or without a case? Under what loading and in what ambient conditions? Garbage in, garbage out.

*** Let's 'go there'.
Source document: VC review: http://enphase.com/wp-uploads/enphase.com/2011/03/Electrolytic_Capacitor_Expert_Report.pdf
Page 2, last paragraph: "Enphase has made some measurements external to the capacitor during actual application that indicated the surface temperature to be a maximum of 65°C."
- What does 'external to the capacitor" mean to them? It could be on the 'surface of the capacitor' but that should be specified. It could also mean on the circuit board near the caps, in the potting near the caps, on the 'surface' of the potting, or on the 'surface' of the case. We have no idea.
- "...during actual application..." If we assume it's the actual application of a microinverter, then what were the conditions? Ambient temperature, duration, input DC parameters - hopefully they don't consider 'on the bench with no input or output connected' to be an 'application'...

Page 3 paragraph 2: "To audit these results, and to present an even more conservative evaluation, a stress test was applied to the calculations."
Audit - paperwork
(If one were going to validate the results, they would repeat the experiment independently of Enphase. This clearly was not done.)
A stress test wasn't applied to the device - it was applied to the calculations - paperwork

Page 3 last paragraph through page 5: Note the section heading "Lab Work"

In this section, the possibility of capacitor degradation from exposure to chemicals in the potting was evaluated. Caps were tested electronically and then physically dissected.

Please note this is the ONLY section of the paper identified as 'lab work' and note that electronic measurements were made of actual capacitors. The authors bios suggest they are well educated and have experience with the scientific method, research, and testing. These gentlemen should know the difference between an actual measurement collected during real-world testing and work derived from models.

They did not model potential halide damage, and did not lab test actual capacitor temperatures for this paper.

What I have said a number of times is that I will keep a spare fan and a set of replacement capacitors in a maintenance box in case they should be needed later.

Electrolytic capacitors die faster if they are operated at 0V than in an actual application. So if you purchase some, put them in a box and pull them out when capacitors in the unit have died the spares will likely have been long dead.

I'm not sure storing them in a package is considered 'operating' them, but we'll let that slide for now. I'm aware that e-caps may need to be re-formed if they sit too long -two years on the shelf seems to be enough. That's why one would test at least capacitance and ESR before using the caps. I have the equipment required for the evaluation and subsequent re-forming should it be necessary.

Please excuse the inaccurate speech. I'll see if I can better state the plan: I am maintaining appropriate tools and materials with which to maintain the inverter and charge controller. Consumable products may be stored on site if conditions warrant. The ability to acquire the necessary replacement parts from the supply chain in a timely fashion is a factor in this decision. Should conditions in the world deteriorate to the extent that the ability to acquire parts is in question, then stocking levels and requirements will be reevaluated and adjustments made. Service materials may include individual components up through replacement boards if determined to be warranted.

I hope that eases your concerns.

Maybe, as with Enphase, I'll decide in 12 years that there is an acceptable amount of ripple on the PV line and simply leave the device on-line. It's good to have options, yes?

???

Tongue in cheek, Reg - tongue in cheek.

Do you have something that tells you that the electrolytic capacitors in your Classic Controller will last for 12 years? Something that tells the operating temperatures and shows a detailed life calculation for the application? Do you have something that says that a failure of any one of these capacitors will only lead to lower operating efficiencies? More likely, it will result in failure of the unit, possibly even cascading failures.

We've covered all this Reg - do we need to do it again? Yes, I have validated my plans using the best information available, including through contact with the designers, by evaluating performance of current and prior tech, by evaluating info from the parts manufacturers, and by designing my installation to provide both the energy and service life I desire. I accept full responsibility for my analysis and mission plan. And no - I'm NOT willing to accept impaired operation. See "preventive maintenance" if you desire more on THAT topic...

Off-grid folks, on the other hand, want full batteries daily. Once the battery is full, any energy not used is wasted.

That's my point. Energy harvest of off-grid systems is extremely low compared to grid-tied systems.

Using what metric? If it's based on the possibility that a low-consumption day will coincide with a high-sun day etc. etc, then sure - knock yourself out if it makes you happy. But - this is a BIG but - that's a grid-tied mentality - NOT an off-grid point of view. It's simply not an appropriate way to evaluate these systems.

But - didn't you say that the string/central/off-grid inverter manufacturers don't have any failure visibility after the 5th year?

Yep. Even if they sell repair parts, it does not mean that they will get a good picture of their product reliability. Most failures after the warranty period are likely junked since they are outdated by that point. That's my experience, anyway.

We've covered this as well. I can absolutely understand why someone would consider a $200 sealed/unrepairable device would be considered disposable should it fail. The same decision is much less likely when the device costs an order of magnitude more. While I agree that some visibility is lost once the warranty is over, it's clear that parts sold are not going to be hanging from the ceiling of any museum of modern art - they're very likely to be used to actually repair a device. I suggest that the visibility number is 50% to near 100% depending on the industry sector. Off gridders need power, and inverter vendors generally provide good support - on and off warranty. Consider Exeltech - $100 flat repair fee for any malfunction from any cause anytime after the warranty expires. Commercial and industrial end-users have the same 5 year warranty, but generally also have service contracts and 99% up-guarantees that provide a longer tail than a simple 5 years.

Additionally, I think we agree that electrolytic capacitors are a limiting component in power electronics and that devices that do not use them have the potential of being more reliable than equipment that incorporates them - even if they provide redundant parts.

Component selection is an important part of reliability, but it all depends critically on the details of the design and the manufacturing. Just because an Enphase competitor offers inverters without electrolytic capacitors does NOT mean that it will be a more reliable product.

That's correct. What's shown the products to be more reliable is performance in the lab and in the field.

Enphase has achieved a very high MTBF on FIELDED units. And they have done the analysis and life testing to show a very long life on their capacitors.

On at least SOME fielded units...according to you (no offense). Again - a company with a desire to show how well their devices function will contract independent testing and publish the results. That's not the case here. If Enphase thought their caps would last for 25 or 30 years without failing in the real world, their summaries would not have any need to confirm that device performance won't be significantly impaired if the caps fail. We can revisit this in 10 years if you'd like.

I will predict that many of the inverter solutions coming out today which claim to have long life because they eschew electrolytic capacitors will be unreliable compared with Enphase inverters.

Noted. The same company has 10 years under their belt for their equipment without optoelectronics or e-caps, and longer without e-caps. They are good to maintain their 99% up-time guarantee. Seems you might be behind the power curve here from the start...

Finally, I think we agree that one should evaluate an entire design - including pros and cons of alternate architectures - before making the final decision about which tech to select.

That's what this thread is about. I have compared and contrasted two different architectures to help people see the trade-offs they are making with their system choices. DesertDenisen has added in real, long-term experience with a third type of system showing outstanding results.

I agree with your assessment of our friend from Tucson's system, though I'd love to know what inverters (s)he's using! And yes, I agree that you've presented a look at the two systems with which you've been involved. I do not agree with your trade-offs, however, and do not think that the off-grid installation you've installed is a typical off-grid package. I will work on fleshing out the examples.

 

[AndyH]

Found it! This is an excellent summary paper for anyone that wants a fairly easy to digest overview of where inverters are used, the development history since the 1980s, where improvement efforts are focused, and which companies are (and are not...) working openly to improve the entire industry.

Summary Report on the DOE High-tech Inverter Workshop
January, 2005

http://www1.eere.energy.gov/solar/pdfs/inverter_II_workshop.pdf

Thumbnail sketch of (primarily grid-tied) inverter development history:

In the early 1980s industry was focused on getting inverters to operate connected to the grid. The main problems were with the requirements for low harmonics and maintaining reasonable efficiency. In the late 1980s to early 1990s the focus turned toward maximizing efficiency, followed by adhering to codes and standards in the mid 1990s. That phase was followed by a focus on “getting to the bottom of the reliability bathtub,” which refers to the classic chart that shows high infant mortality (1st year failures) declining quickly to a much lower level of failures in the field (the bottom of the bathtub) then another sharp incline in failures at the end of their projected lives (the other end of the bathtub). The focus now is on stretching the bathtub, extending lifetime, by controlling temperatures resulting from power semiconductors (point sources) and magnetics (diffuse).

(Please note that off-grid inverters have been in use in the civilian world since at least the 1960s.)
http://en.wikipedia.org/wiki/User:Santosh77/Subpage
http://www.midnitesolar.com/pages/frontPage/nwHistory/history.php

Participants came from a broad range of the power industry:




Three inverter manufacturers have been working as part of the high-tech inverter process and are using accelerated live testing and other advanced testing of actual hardware in realistic environments in order to double time to first failure and reduce inverter cost. They are GE, Satcon, and Xantrex.

Here's a snapshot of Xantrex' progress in this effort as of Oct 2004:
http://www1.eere.energy.gov/solar/pdfs/5_hudson_xantrex.pdf
GE as of Oct 2004:
http://www1.eere.energy.gov/solar/pdfs/6_smolenski_ge.pdf
And SatCon, as of also Oct 2004::
http://www1.eere.energy.gov/solar//pdfs/7_casey.pdf

The top four areas identified as needing significant improvement were capacitors, thermal management, communications and surge/transient suppression from the grid. [Pages 3-5]

When it came to thermal management and packaging, cast aluminum cases were identified as a superior technology, and Outback Power was identified as the only company as of that date that has successfully commercialized this technique. [Page 29] (It's important to note that Outback has been using this case since about 2001 - they are one of the companies known to be leading the industry.)

On the subject of PV industry size and capacitor improvement:

The group acknowledged that with the PV industry being a relatively small player when it comes to using capacitors, it would not have a big influence over capacitor designs and requirements. They discussed other methods of addressing capacitor issues including reducing the requirements for capacitors with innovative circuit topologies. But they concluded that PV most likely could not influence capacitor reliability, because the PV industry does not have a large segment of the market and cannot demand improvements in reliability.

There are additional comments in the paper that show that the group is too small to have the 'tail wag the dog' even if the PV, automotive, and the rest of the inverter users gather.


Fast forward to 2010:
Achieving Low-Cost Solar PV: Industry Workshop Recommendations for Near-Term Balance of System Cost Reductions
Sept 2010
http://www.rmi.org/Content/Files/BOSReport.pdf


One thing's for sure - these and a number of other workshops and projects since have accelerated inverter reliability greatly. Today's inverters are better than 1960, 1970s,1980s, 1990s, and early 2000s-era inverters because the industry is working hard to keep up with the full range of distributed generation, battery storage, and vehicles.


The more we know about the tech side of the industry, the more easily we'll be able to see through (or validate) the marketing message.

 

[AndyH]

Everything breaks. That brings us to a decision point - shall we ignore it, fix it, or replace it?

Some equipment is considered disposable from the start. Even if some don't see it that way, once you've picked your way through potting a tenth of a millimeter at a time like an archaeologist, you might decide at about 2AM that the blasted thing has just become disposable regardless of what the label might say! :lol:

For the equipment that's designed to be fixed, how supportive is the manufacturer? Are parts more expensive than replacing the device? Have components evolved so quickly that it's better and less expensive to get something from this decade? Who'll do the repair work - a local tech, maybe our installer? Will we send the box to the manufacturer? Will they send a loaner to keep up running?

Here's an example of board level maintenance/repair for a leading grid-tied with battery back-up and off-grid inverter. You'll notice that none of this is brain science..er, rocket surgery...whatever.

Written instructions:
http://www.outbackpower.com/docman/1401104040430fx-circuit_board_replacement.pdf

Video:

[youtube]http://www.youtube.com/watch?v=cb2PpRs9Bss[/youtube]

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

As you can see from these videos, a complete tear-down and reassembly is about 20 minutes. These units are available in two variants - vented and sealed. The vented units have a fan in the cover. The fan does not run at all times and is expected to last much longer than a typical computer power supply or other typical consumer fan.** It can be replaced by removing the cover (as seen in video 2).

In the broad inverter world, the second and third tier components in the list of 'most likely to fail' are the electrolytic capacitors and the field effect transistors (FETs). Note that these are on the same board, and that this board can be swapped without removing the computer board or the AC board. (Note that for this specific series of inverters, Outback expects the FETs to fail before the caps. The caps installed are made in Japan and are high-ripple, low ESR, 105°C caps; there are six on the board.)

Notice how modular the system is - a new computer development or more efficient AC section is 20 minutes away - without a ladder or an entirely new unit.

Before buying any equipment, call the manufacturer and pose as a current customer with a broken device and see how you're handled.

** The fan in my VFX3524 is a PN AD1212UB-F51. This is a high-speed aluminum framed fan with ball bearings. It has a minimum 50,000 hour life at 40°C. If the fan ran continuously 24/7, that would equate to a 5.7 year lifespan. The fan does not run continuously, however, it's computer controlled and triggered by a temperature sensor on the electrolytic capacitors (and possibly another on the transformer). This inverter, in its sealed configuration, is a 2500W unit. I expect that the fan won't activate at all if the inverter is kept in a climate controlled space and not used to supply more than 2.5KW. We'll see.

http://www.farnell.com/datasheets/1424505.pdf
http://octopart.com/partsearch#search/requestData&q=ad1212ub-f51

 

[AndyH]

One of the sales tactics used by folks selling microinverters is to point to old industry rules of thumb that say it's wise to plan to replace a central or string inverter after 10 years of service. They then show a pro/con list that seems to prove that it's much less expensive to replace 10% of their micros rather than install two central or string inverters.

The problem is that while microinverters are not repairable, central and string inverters are (as seen above).

Continuing with the Outback example started above, we learn from Outback that the inverter has five major components - a transformer, a fan, a field effect transistor (FET)/capacitor board, a computer board, and an alternating current (AC) board. The two most common failures for the inverter are the fan and the FET/Capacitor board.

The fan is a very inexpensive device that is not necessary for the inverter to function. If or when the fan fails, the inverter signals the failure, and may automatically reduce inverter output power (the sealed unit with no fans is rated for 2500VA continuous while the vented unit is rated for 3500VA continuous). If the failed fan leads to overheating (the inverter has a temperature sensor on the transformer and on the capacitors), the unit shuts down. Contrary to the sales rhetoric, a fan failure will not result a failed inverter. Fun fact number one. http://www.wholesalesolar.com/pdf.folder/inverter pdf folder/outback_MATE3_manual.pdf
http://noaz.blogspot.com/2011/08/warning-light.html
http://hireelectric.com/2011/inside-the-outback/

The next most common component to fail is the FET/Capacitor board. Yes, the same filter capacitors we've talking about in this thread. Sorry - they're a maintenance item and do not live forever. One has a couple of options. The first is to replace the FET/Capacitor board. Retail price is about $510. From power off to power on, replacement will take less than one hour. Sales rhetoric aside, while $510 is not necessarily chump change, it's about the same price as four microinverters - and it's much less money than the cost of a new inverter (about $1700). Another option is to service the board locally. The FET's are not expensive, nor are the capacitors. Labor will likely be the largest portion of the repair cost.

The remaining boards, AC and computer, are less than $250 each and are known to be very reliable, especially if the inverter is not subjected to the electric grid, and protected by surge suppressors that work. I've talked to service reps, parts suppliers, and folks that have been off grid for 40+ years and the message has been very consistent - fans and possibly capacitor boards fail but none of the dozen folks I talked with had ever heard of any other part failing in these inverters.