what is the longest a grid tie central inverter has lasted?

Re: what is the longest a grid tie central inverter has lasted?

Did not Enphase redesign their newer units to remove the electrolytic capacitors?

Pretty much what Marc/Cariboocoot said. You have aging from heat (for every 10C/18F increase in temperature, live is cut by 1/2. Run something at 36F or ~106F over 70F ambient, live is cut by 1/2*1/2=1/4 the life). And you have the thermal cycling (for example, jet engines have the number of on/off cycles as a major predictor of the next maintenance cycle).

Back in the day when 3 1/2 inch 40-100 MByte disc drives were coming out. I would just setup an environmental chamber to cycle from 40F to 112F (or what ever the spec was) and cycle from cold to hot to cold about 4x per day. After about two weeks, 80% of the vendor drives would fail (usually some component would delaminate from the PCB), and 20% would (appear) to work "forever". As the industry improved, I had almost zero failures of hard disks from thermal cycling.

A lot of people like to say that the computer in their car will last 15 years in the engine compartment. In reality the number of hours that a car operates is not that high. For example, say:

• 15 years * 15,000 miles per year = 225,000 miles
• 225,000 miles / 35 MPH typical average speed for a car = 6,428 hours
• 24 hours per day * 365.25 days per year = 8,766 hours per year

That is less than a year of operating time for a 24x7 off grid inverter. Or possibly two years of life for a GT inverter or solar charge controller

In the end, 10 years is a pretty long time for always on electronics and add the daily thermal cycling for high power electronics/switching components, it is difficult (and expensive) to design and manufacture something that will last longer.

And, it is not that the failures would not be easily repaired or "failure prone"/life limited components cannot be repaired--It is the fact that electronics have been evolving so fast. Miniaturization for anything with a transistor in it means the average production life for a computer/memory component is around 1-2 years. It is simply difficult/expensive/to get spares 10 years later.

Back in the early 1980's, the top of the line 9" Fujistu Eagle disk drive weighted 150 lbs, cost ~$20,000 in volume, and held ~500 MBytes of data. I was at the office products store yesterday -- a 4 GB USB memory stick for $4.99 or so.

Moore's Law of doubling of transistor density every 2 year, 2⁵=32x more transistors every 10 years has just made repairing of old electronics, or even providing 20-30 year life components just not worth the time and money.

Miniaturization has actually really helped to improve reliability. Smaller devices have less physical size which means that thermal cycling has less of an effect. And smaller devices generate less heat.

-Bill