Double the lifetime of an inverter?

**J.P.M.** · 05-09-2016, 11:30 AM

Originally posted by peakbagger

IMHO, air flow is important but air temperature and loading of the inverter are major factors in inverter life. I have my inverters attached to a concrete wall in a below grade basement. On the hottest day in the summer the incoming air temp is no more than 60 deg F. I also try to oversize inverters slightly with the goal of keeping the output below the maximum rating of the unit (no clipping). I find that the various solar installers all tend to oversize their arrays for the inverter and rarely do I see a microinverter sized to match up with a panel rating. Enphase even came out with technical paper insisting that a microinverter should intentionally be undersized so that it clips as that is best economics. I prefer long term reliability and probably would not even install a microinverter due to the environment it is located but I sure wouldn't overpanel it.

Knock on wood I have a 14 year old Advanced Energy inverter that is still running. These units were an early grid tied only inverter and most installers over paneled them. They had lot of warranty failures and it drove the company out of business. At least one former employee blamed a lot of the issues on thermal issues. It is passively cooled (no fan).

I see PR shots of PV installations and see inverters mounted on exterior walls siting right in the sun. Might be good for a PR shot but lousy for the electronics. The best place is indoors preferably in air conditioned space or in a cool basement.

Air (or other coolant) flow is not only important, it's essential, and what makes heat exchangers work to haul the waste heat away.

Q = M* c*delta T. where, Q = heat removed. M = mass flow rate, in this case air. c = The specific heat of air. Delta T = (to a 1st approx. here) the temp. diff between the heat exchanger and the air.

An additional and complicating consideration for nat. convection devices, at least in a gravity field : The M is largely controlled by the delta T, with design considerations to minimize pressure drop, make maint. as easy as possible, and of course, safety and cost considerations.

A thought experiment: Put a nat. convection cooled inverter in a sealed, insulated box while its operating and imagine what will happen. Without (cooler) air flowing through the heat dissipation apparatus on the inverter (cooling fins, etc.), things will heat up quite quickly and (hopefully) the protective devices will shut the unit down. The unit needs a cool air source.

Devices dependent on nat. convection for cooling will run cooler in cooler environments. An inverter cooled via nat. convection in a 60 deg. F. environment will run cooler than in, say, an 80 deg. F. environment by probably something like 20 deg. F. That's probably part of the reason why you're inverters are still functional. If they were in an outside shed with a commonn air temp. of, say, 90 F., they'd often be running something like 30 deg. F. warmer.

Your 60 F. basement temp. has probably contributed to the inverters' longevity.

Adding forced convection (fan assist) to any nat. convection cooled inverter in the same environment will decrease an inverter's delta T and thus lower the inverter operating temp. How effective the scheme is, and how cost effective it will be in terms of efficiency improvements, longevity and operating/maint. costs are separate questions and considerations that will change with the application and goals of the user.

When planning inverter location during system design, a stable environment with somewhat controllable conditions is often deemed best. That's usually indoors in a controlled environment with reasonable access. Under a panel in the middle of an array where access is limited and an annual temp. swings of, say 100 F. are possible, if not likely, as in the case of microinverter location, does not seem to meet the stable, accessible requirements part as well as does a garage or basement location for a sting inverter. However, the design process is always a set of tradeoffs and priorities.

**DaveDE2** · 06-09-2016, 05:17 PM

My system is fired up now, two each 6kw inverters, each putting out about 4kw just now. I put a small low volume fan (5x5", don't know the cfm, maybe 15cfm?) close under one without any ducting and blowing upward toward the heatsink. The heatsink temp (60C) dropped by 18C compared to the other one. The front face of the inverter dropped by 6C compared to the other one.

I know now that the MOSFETs (mounted in the heatsink) should last at least 10 years longer with said airflow. My next en devour will be to place temperature sensors inside the box, on the electrolytic capacitors. They are not intimately mounted to the heatsink but are non the less running cooler with the fan. More data later...

**sensij** · 06-09-2016, 05:59 PM

Originally posted by DaveDE2

I know now that the MOSFETs (mounted in the heatsink) should last at least 10 years longer with said airflow. My next en devour will be to place temperature sensors inside the box, on the electrolytic capacitors. They are not intimately mounted to the heatsink but are non the less running cooler with the fan. More data later...

Huh? The inverter obviously spends only a small part of the day operating at peak temperature. Where does the 10 years estimate come from?

**DaveDE2** · 06-09-2016, 06:14 PM

It comes from the fact that semiconductor and capacitor lifetime decreases by a factor of two for every 10C increase in operating temperature. The fact that the inverter spends a small part of the day operating at peak temp is irrelevant.

The 10 year estimate comes from typical time to failure of an inverter, which is about 10 years. Those failures are caused by and large by the average temperatures that the components reach over that lifetime.

**jflorey2** · 06-09-2016, 06:35 PM

Originally posted by DaveDE2

It comes from the fact that semiconductor and capacitor lifetime decreases by a factor of two for every 10C increase in operating temperature. The fact that the inverter spends a small part of the day operating at peak temp is irrelevant.

No, it's not.

If the inverter is dissipating almost no power, then the difference between forced air cooling and no forced air cooling is zero. No extension of life.
If the inverter is dissipating a lot of power, then the difference between forced air cooling and no forced air cooling is large - and only then do you see an extension of life.

Thus the time operating at high power matters a lot.

Also, keep in mind that while some failure mechanisms do scale like that (specifically electrolytic capacitors) semiconductors do not. The thermal effects (increased diffusion for example) are very small compared to the other failure modes - and in modern semiconductors, those other non-temperature-dependent failure modes dominate.

**ButchDeal** · 06-09-2016, 07:40 PM

Originally posted by DaveDE2

It comes from the fact that semiconductor and capacitor lifetime decreases by a factor of two for every 10C increase in operating temperature. The fact that the inverter spends a small part of the day operating at peak temp is irrelevant.

The 10 year estimate comes from typical time to failure of an inverter, which is about 10 years. Those failures are caused by and large by the average temperatures that the components reach over that lifetime.

semiconductors do not care that much about temperature.
Peak temp doesn't really matter, it is over operating temperature range. solaredge inverters for example have a peak operating temperature of 140F, most people are not going to have a problem staying under that.
Other inverters are like Outback (mine) have lower temperatures (well outback cut performance above 104F)

Old school data center managers back in the 70s, 80s used to keep data centers around 60F too thinking it would keep the computer working better. Many tests have shown there is no reason for this so most data centers started ramping temperatures up in the 90's , now most are 80F and even 85F in the COLD rows.

**DaveDE2** · 06-09-2016, 08:32 PM

You go Butch, keep pedaling SE and your stories. Ignorance is bliss.

Mod Note: Please avoid personal attacks on other members.

**solarix** · 06-09-2016, 08:38 PM

I totally agree that heat is the enemy of electronic reliability - especially with the new regime of lead-free solder. I'd like to know how the microinverter folks substantiate their 25 year warranties when operating at much higher temperatures than standard string inverters. I know they know have designed out the weak-link of elec. caps but I still do not trust them.

By the way, electrolytic caps are really only suitable when used at lower frequencies. This is not something that capacitor specs identify very well - which the whole electronic industry found out the hard way when the big move to switching power supplies came through and power supply frequencies went from 120hz to 2kHz to 5kHz to 10kHz to 20kHz and on up as advances in transistor switching times improved. Great for reductions in the transformers - but the poor low cost electrolytic's dissipation factors suffered. Just google the subject "bad caps" sometime.

**ButchDeal** · 06-09-2016, 09:11 PM

Originally posted by DaveDE2

You go Butch, keep pedaling SE and your stories. Ignorance is bliss.

ok Dave for a person trying to find out some info from people that know more than you, you sure are a bit of an ass to those that do.
I would suggest that you listen instead of speaking but it is a bit too late for that.

Mod Note: Please avoid personal attacks on other members. (You too Butch)

You still haven't stated where the 10 year estimated lifetime comes from either?

Where are the stats on the 10C temp rise leading to doubling failure rate? This would be simple to prove based on install location latitudes.

**DaveDE2** · 06-09-2016, 09:43 PM

With all due respect, I don't see that lead free solder has anything to do with it. Lead free solder melts at a higher temperature but isn't any less conductive as far as I've heard. Yes, electrolytic caps are best suited for low frequency applications but they are well known to handle jobs in the multi kilohertz region.

Back to the matter at hand. Let's say an inverter manufacturer warrants its inverter at 10 years. The mfg knows that it will only be putting out for say 8 hours out of a 24 hour day, but that's all the electronics are expected to handle. Now say we come along and cool those electronics by 10C - the life will double. Someone cries foul and says "yeah but those electronics are not being cooled by 10C for the whole time because the cooling system is not cooling it by10C the whole time it's operating". Fine, you'd be somewhat right but I actually measured 18C difference at the hottest part of the day, so I guesstimated an average 10C delta average which I think is reasonable. If someone were to nail this they'd have to integrate the temp and then average it over time. Arrhenius equations hold true.

**ButchDeal** · 06-09-2016, 09:56 PM

Originally posted by DaveDE2

With all due respect, I don't see that lead free solder has anything to do with it. Lead free solder melts at a higher temperature but isn't any less conductive as far as I've heard. Yes, electrolytic caps are best suited for low frequency applications but they are well known to handle jobs in the multi kilohertz region.

Back to the matter at hand. Let's say an inverter manufacturer warrants its inverter at 10 years. The mfg knows that it will only be putting out for say 8 hours out of a 24 hour day, but that's all the electronics are expected to handle. Now say we come along and cool those electronics by 10C - the life will double. Someone cries foul and says "yeah but those electronics are not being cooled by 10C for the whole time because the cooling system is not cooling it by10C the whole time it's operating". Fine, you'd be somewhat right but I actually measured 18C difference at the hottest part of the day, so I guesstimated an average 10C delta average which I think is reasonable. If someone were to nail this they'd have to integrate the temp and then average it over time. Arrhenius equations hold true.

You guesstimated and were did you come up with the doubling lifetime.
Arrhenius equations do not state a doubling of life, in fact it is about chemical reactions. This would be true in caps but not silicon chips ( not very reactive are they )

**solarix** · 06-09-2016, 10:15 PM

Lead-free solder is less flexible and does not handle temperature cycling nearly as well. I'm all for getting the lead out, but the electronic industry has struggled for decades to come up with anything nearly as reliable as good old 60/40 solder. And now the parts are soooo much smaller and just surface mounted with really tiny contact areas. Name me any other consumer electronic product that has more than a 2 year manufacturer's warranty. I still maintain that when choosing an inverter brand, base it on reliability, reliability, reliability.....

**azdave** · 06-10-2016, 08:50 AM

We're all shootin' from hip here anyway.

How about half of us put extra cooling on our inverters and the other half leave them as is. Then we'll all check back on this thread in 20 years and see who was right.

**jflorey2** · 06-10-2016, 10:41 AM

Originally posted by DaveDE2

Gosh, you are so totally wrong. I've been in the aerospace business for 30 years

Well, let's all hope you haven't gotten your hands on anything critical, then.

**J.P.M.** · 06-10-2016, 11:38 AM

Originally posted by azdave

We're all shootin' from hip here anyway.

How about half of us put extra cooling on our inverters and the other half leave them as is. Then we'll all check back on this thread in 20 years and see who was right.

Probably a practical observation. Failure rates/MTBF's, etc. are tough to quantify/estimate for this stuff anyway.

Double the lifetime of an inverter?

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment