Solar micro-inverter addresses capacitor reliability issues
Nov 3 2009 11:19AM | Permalink |Comments (21) |
The lead-in to this post is here: Micro-inverters offer one solution for optimizing solar efficiency.
Probably the most well-known micro-inverter company is Enphase, which sells a 200W inverter for about $200 or about $1/W. This compares with a 3kW string inverter for about $2,000 or about $0.66/W. Enphase suggests that the additional $.33/W is compensated for by the lowered installation labor and investment costs.
(From yesterday’s post: A 200W micro-inverter that only has to deal with 30Vdc input and no special installation investment becomes very attractive for small installations. Safety is also an issue: A solar panel string or array’s input to a central inverter can be as high as 600Vdc in the US and 1000Vdc in Europe -- hazardous voltage levels for installers, maintenance personnel, and emergency responders. Output from modules attached to a micro-inverter scheme will be at much lower levels between 200 -- 300Vac.)
Enphase’s leading position in the micro-inverter market is because, as far as I know, it’s the only micro-inverter company actually delivering product. Enphase’s co-founder Raghu Belur, who I met with at Solar Power International last week, told me that they have so far delivered 100,000 devices in their first year.
Common inverter topologies use electrolytic capacitors on their output filters, and electrolytic capacitors have poor reputations for reliability especially when subjected to the elevated operating temperatures of solar installations. When you go from a single central inverter to 10-20 micro-inverters, the likelihood of a failure due to an electrolytic capacitor increases likewise. Most solar panels have a life of 25-30 years, and operators want a similar lifetime from their inverter circuits.
Enphase has posted several white papers on its site dealing with capacitor reliability and lifetime, such as (pdf) Reliability Study of Electrolytic Capacitors in a Micro-Inverter by Enphase's CTO, Martin Fornage. My take-away from the white paper is that Enphase uses higher-reliability electrolytics than those normally found in power supplies, and the paper shows how their higher-reliability translates to a longer life in real-world temperatures encountered in solar installations.
More detail from the paper: For traditional power converters, an acceptable useful life of capacitors is as low as 2000h at 85°C. Enphase micro-inverters use Nichicon (pdf) capacitors rated from 4000 to 10000h at 105°C. Capacitor lifetime is very sensitive to temperature and follows the Arrhenius equation that states that useful life doubles for every 10°C temperature drop. Temperatures from the National Solar Radiation Data Base (NREL) for the California desert town of Palm Springs in the summer show a maximum ambient temperature of 46°C, resulting in a core temperature for the capacitor of 65°C. Thus, capacitors rated at 4000h at 105°C have a useful life of 50 years when operated in the Palm Springs climate of 46°C ambient temperature. Other papers (pdf) on the website refer to its micro-inverters being “designed for a service life of 20 years,” but the website lists the warranteed lifetime for its products at 15 years.
Enphase’s micro-inverters are stand-alone units, sold separately from the panel, and can work with a wide range of solar panel, and by definition require no central inverter. MPPT (defined previously) is performed at the panel, so each panel provides its individual maximum power so there’s no fear of a central inverter reaching only a local MPPT for a whole array, increasing the efficiency of each module. Central inverters have a higher efficiency, exceeding 98%, while Enphase micro-inverters are currently pushing about 95%.
Next up: External micro-inverters vs integrated
Related entries in: Power supplies | Solar/Photovoltaics |
Nov 3 2009 11:19AM | Permalink |Comments (21) |
The lead-in to this post is here: Micro-inverters offer one solution for optimizing solar efficiency.
Probably the most well-known micro-inverter company is Enphase, which sells a 200W inverter for about $200 or about $1/W. This compares with a 3kW string inverter for about $2,000 or about $0.66/W. Enphase suggests that the additional $.33/W is compensated for by the lowered installation labor and investment costs.
(From yesterday’s post: A 200W micro-inverter that only has to deal with 30Vdc input and no special installation investment becomes very attractive for small installations. Safety is also an issue: A solar panel string or array’s input to a central inverter can be as high as 600Vdc in the US and 1000Vdc in Europe -- hazardous voltage levels for installers, maintenance personnel, and emergency responders. Output from modules attached to a micro-inverter scheme will be at much lower levels between 200 -- 300Vac.)
Enphase’s leading position in the micro-inverter market is because, as far as I know, it’s the only micro-inverter company actually delivering product. Enphase’s co-founder Raghu Belur, who I met with at Solar Power International last week, told me that they have so far delivered 100,000 devices in their first year.
Common inverter topologies use electrolytic capacitors on their output filters, and electrolytic capacitors have poor reputations for reliability especially when subjected to the elevated operating temperatures of solar installations. When you go from a single central inverter to 10-20 micro-inverters, the likelihood of a failure due to an electrolytic capacitor increases likewise. Most solar panels have a life of 25-30 years, and operators want a similar lifetime from their inverter circuits.
Enphase has posted several white papers on its site dealing with capacitor reliability and lifetime, such as (pdf) Reliability Study of Electrolytic Capacitors in a Micro-Inverter by Enphase's CTO, Martin Fornage. My take-away from the white paper is that Enphase uses higher-reliability electrolytics than those normally found in power supplies, and the paper shows how their higher-reliability translates to a longer life in real-world temperatures encountered in solar installations.
More detail from the paper: For traditional power converters, an acceptable useful life of capacitors is as low as 2000h at 85°C. Enphase micro-inverters use Nichicon (pdf) capacitors rated from 4000 to 10000h at 105°C. Capacitor lifetime is very sensitive to temperature and follows the Arrhenius equation that states that useful life doubles for every 10°C temperature drop. Temperatures from the National Solar Radiation Data Base (NREL) for the California desert town of Palm Springs in the summer show a maximum ambient temperature of 46°C, resulting in a core temperature for the capacitor of 65°C. Thus, capacitors rated at 4000h at 105°C have a useful life of 50 years when operated in the Palm Springs climate of 46°C ambient temperature. Other papers (pdf) on the website refer to its micro-inverters being “designed for a service life of 20 years,” but the website lists the warranteed lifetime for its products at 15 years.
Enphase’s micro-inverters are stand-alone units, sold separately from the panel, and can work with a wide range of solar panel, and by definition require no central inverter. MPPT (defined previously) is performed at the panel, so each panel provides its individual maximum power so there’s no fear of a central inverter reaching only a local MPPT for a whole array, increasing the efficiency of each module. Central inverters have a higher efficiency, exceeding 98%, while Enphase micro-inverters are currently pushing about 95%.
Next up: External micro-inverters vs integrated
Related entries in: Power supplies | Solar/Photovoltaics |
Reader Comments
at 11/3/2009 1:29:08 PM, Simpleton said:
Having just received quotes from Akeena, a Solar company pushing the Enphase micro-inverter, for both a central inverter equipped system vs. a micro-inverter equipped system, I am having trouble with the math.
The micro-inverter equipped system costs $2,745 more (after incentives, even more before) and produces 51kWh less electricity per year (due to lower efficiency).
at 11/3/2009 1:33:32 PM, SunShine said:
Microinverters have the same limitations standard inverters always have had with long term reliability. And now they want us to put one on every panel! Not a good idea =\ Other power optimizer technologies (ie DC-DC converters) using more reliable architectures would be a much better suited for per panel deployment and especially for panel integration.
at 11/3/2009 2:57:31 PM, seacrow said:
Obviously peak efficiency and fault tolerance and identification are the advantages of microinverters. Using cheap organic caps, commodity-grade semis and packaging, and topologies based on lowest recurrent cost are probably not gonna get us there.
Keep at it anyway!
at 11/3/2009 2:59:20 PM, Andy T said:
I don't buy the math. One grid or distribution surge event, pinholing the dielectric, and the caps are compromised since I doubt they are self-healing. I think the answer to all this 60Hz AC stuff from solar in terms of reliability is to use a sinusoidally-shaded grey scale "color wheel" above the panel and just a simple H-bridge to flip polarity every half cycle.....plus, it'll shed ice in cold climates.
at 11/3/2009 3:04:42 PM, seacrow2 said:
...besides, the cost per watt processed should be *less* or *mich less* than the raw peak power provided by the panel. Enphase's pricing model is basically flawed and pointy-haired, providing negative incentive to budget-makers and installers.
at 11/3/2009 3:12:38 PM, seacrow3 said:
... and finally, parallel DC-AC microinverters is way stupid, since a full 1/2 AC cycle of energy storage is needed per module. Sorry, Andy T, but wasting a huge fraction by shading or spinning the panel is lame too, just to get AC out. Solar-Areal efficiency is important too since real estate ain't free.
The conversion should be high-kHz MPPT DC-DC with a HV link to a central inverter with hockey-puck IGBTs. Each DC-DC need store very little energy and MTBF goes thru the roof. Fallback and redundancy should be at the line interface, not at each panel.
at 11/3/2009 4:16:04 PM, SolarBozo said:
The math does work out (barely), but only if the reliability numbers can be trusted. The slightly lower inverter efficiency is outweighed by the increase in lifetime energy harvest with a per-panel approach. Enough even to counteract Enphase's (temporary, I hope) very high cost/unit. Many in the industry are very skeptical about the reliability claims though, not just because of the low temperature assumptions and the hopeful take on e-caps, but also because of the complexity of the design.
To seacrow, it seems you might have been fooled by Andy T's joke about the rotating color wheel, but maybe not - your suggestion of spinning the whole panel is right up there!
at 11/3/2009 6:55:29 PM, Andy T said:
I love it - direct-to-grid 3600 RPM solar panel spinners; much better than the color wheel as it keeps the cells cool which boosts conversion efficiency. We can also harvest additional power from the spinning panels with downstream wind turbines a la Wile E Coyote.
at 11/3/2009 7:42:47 PM, DVanditmars said:
I find the one inverter per panel very intriguing. 1) You get redundancy, on panle/inverter fails you only loose 200W.
2) You do not have to go for the whole 3.5KW install all at once. You could start out with 3 or 4 panels/inverters and work your way up over the years to more panels/inverters.
at 11/3/2009 11:08:01 PM, Robert Godes said:
I think the best Micro-Inverter on the market is at
www.xetenergy.com/solar/solutions/
at 11/4/2009 12:18:50 AM, Undermind said:
Andy T, you are definitely on to something there. Since the panels have current running through them and we are rotating the panels in the earth's magnetic field we should also be able to generate excess current in the panels.
at 11/4/2009 7:06:42 AM, Rich Lee said:
The Enphase units have been hanging under panels around the USA for some time now.
And having MPPT on each panel allows them to perform very well in real-world conditions.
In cases where comparisons (to old tech inverters and the same model panels) have been done, the Enphase units have out performed the old stuff.
If there was a problem with the Nichicon capacitors,
wouldn't it have shown up by now?
Why do I get the feeling that some of the people posting here, work for a old tech company?
I think everyone who has looked at the Enphase installation info (and video), knows this system is a DIY job. No professionals needed.
No need to pay double anymore.. :)
at 11/4/2009 1:55:05 PM, Roger said:
I can't see how the math works at all -- taking the extreme upside (10,000 hours) estimate and an average day length of 12 hours, you should expect seeing failures after 833 days. Which would be completely unacceptable -- a working life of 2.3 years for an inverter just doesn't fly, especially given the aggravations of climbing up on the roof and lifting up the panel to swap it out.
That inverter on the side of your house is warranted for 10 years, and doesn't involve climbing up a ladder to swap it out when it fails.
Frankly, I'd be quite surprised if they lasted this long, keep in mind that they're getting (electrically) hammered 60 times per second in a very hot environment.
As for MPPT gain -- sure, when they're all working. In the real world, what'll happen is one will fail, then another, and finally there'll be enough dead ones that it's worth climbing up on the roof to replace the failed units, by which time any gain from per-panel MPPT optimization will more than have been sucked up by months of having 1/4 of the array down.
Bah Humbug to microinverters and the turnip truck their owners rode into town on.
at 11/5/2009 4:53:32 PM, john L said:
Roger..
I believe your math is incomplete.
10,000 rating was based on the higher temp (105C)
and it was explained the expected increase in life was based on operating in Palm Springs measured temp of 46C.
Obviously .. you can dis-agree...
but your math isn't making your point.
also..
It has been stated in prior postings... large or small solar arrays ... both need regular cleaning...
No getting around it... you will have to be on the roof on a regular basis.
I would expect the micro inverters to be smart enough to report their status / performance.
not quite the same thing as a redundant indicator light going out... ( no one noticing until all lights are out).
at 11/5/2009 4:58:24 PM, William Ketel said:
How about instead of all the little inverters with the less efficient way of doing things, a variable switching system so that the sections that don't provide enough current get re-arranged into a parallel hookup and the whole mess is in series. Yes, now you may have 600 or so volts on the line to the main inverter, that is what safety procedures are all about. I have worked on large arrays of series gell-cells, where not only was there a "quite high" DC voltage present, but no way to shut it off without unbolting a connection or a few. Gell-cells don't switch off, you see. And please note that I never got killed, even once. REstructuring an array would have much fewer sources of loss, and the ability to disconnect failed sections completely. And note that switching relays are a bit cheaper than inverters.
at 11/5/2009 11:28:22 PM, Joel Ligocki said:
Could any of you comment on bypass diode? Any of you experienced any quality or reliability issues? From what I can tell, many bypass diodes on market seem to have very poor quality.
at 11/7/2009 9:48:08 PM, diode answer guy said:
Joel,
The incidence of bypass diodes failures in solar modules is an issue. Originally, the currents in solar strings was on the order of a couple of amps and P/N diodes were adequate. For each amp, these would dissipate 0.7-1.0 watts. But, as the cell currents got up to 8 amps, this power dissipation became unreasonable and the industry had to switch to "low" drop Schottky diodes (with a 0.4-0.5 volt drop). But, at 10 amps, these still produce excessive heat. The module and J-Box designers have been unwittingly been treating these like consumer electronic diodes whereas they see a particular diode is rated at 150 Celsius or 200 Celsius and assume that means that the product will fit their needs (if you put 10 amps into a Schottky, it will heat up to 150 Celsius quite easily). Unfortunately, they don't realize that the way that semiconductor companies rate this is by seeing if it will survive 1000 hours at that temperature. Not really a good test. Further, these diodes leakage doubles every 10 degrees Celsius in reverse operation and this also introduces an undesirable reliability problem as this "leakage * Vreverse" can be bad.
at 11/7/2009 9:50:43 PM, diode answer guy said:
Joel,
If your sources are public information, could you share where you heard that the bypass diodes are causing reliability problems?
at 11/9/2009 4:57:13 PM, BeamMeUp said:
One of the reasons Europe went away from micro-inverters was when a large manufacturer went out of business because a single grid power surge took out over 10,000 units at once. Ouch. All of the string inverters intermixed in the same area survived. The manufacturers who have the experience with many units in the field for many years will tell you it's a harsh world out there and the larger components tend to survive surges and lightning much better than smaller ones.
My 15+ years in various power industries agrees.
My experience also see's capacitor failure as a relatively low probability.
From my experience the most common failures are due to bad solder joints, bad batches of components which fail through many hot/cold cycles, and AC power surges.
at 11/9/2009 5:06:21 PM, Joel Ligocki said:
diode answer guy,
Recently some one sent me some bypass diodes for analysis due to quality issues/concern. I know very little about PV industry(and was not interested in it until now) but do know a few things about diodes and manufacturing of diodes.
I decided to go beyond the samples sent to me and collected bypass diode samples by two other "manufacturers". I analyze their quality by first opening them up and inspect how well they are put together. All three failed in my analysis in that they all have faulty die soldering. Specifically, I found cold soldering and/or rampant solder voids in all these diodes.
Poor die soldering can cause so many different problems. The most critical yet often overlooked one is the heat dissipation problem as a result of poor die soldering. The heat generated by electrical current flowing through silicon die(and by reverse leakage) will have great difficulty to dissipate to the outside should it has poor die attachments, causing junction temp. to rise rapidly(in turn cause higher reverse leakages which causes junction temp. to go up even more). Eventually, the silicon die is forced to operate at a much higher junction temp. if it ever reaches steady state at all, or worse it creates a thermal runaway situation, causing the diode to short out. Faulty solder joints all have very weak mechanical strength which can often lead to separation of solder joints under mechanical impacts and/or thermal cycling, cause "open" failure mode.
The reason I posted the question is because I wonder if the poor die soldering I discovered in these bypass diodes are acceptable to the PV industry for what the bypass diode is designed to do under harsh working environments? Maybe I am missing something here?
at 11/10/2009 7:26:43 PM, M B said:
From an installer standpoint, our main problem with microinverters is after the warranty period. If the inverters start to go, one at a time, who pays for the labor of climbing up onto the roof, removing a panel (middle of a large array) and reinstalling. Then doing it say another 23-25 times for a 5 kw system. With one inverter, your in and out in 1-2 hours. With cost, if you have 24 inverters and they are $200 ea, you have $4800 worth of inverters. With an SMA 5000, your looking at between $2-$3000. If you get into larger systems than 5-10kw, what a nightmare. You would have some great service contracts though, just bad for the consumer. As for individual panel monitoring, we've never replaced a single panel. GE told us they have a 1% module failure rate, historically.
