Hacker News Reader

Comment by nominated1 | Hacker News Reader

A friend of mine is a union electrician with 15 years in the trade. He’s helped install dozens of solar farms across Northern California. One thing I found interesting/unnerving was that apparently the spacing tolerances between panels is quite low. He said it’s their biggest concern when it comes to fires.

So, maintenance seems like a must. Especially on rooftops or structures that will naturally “settle” over time. If I had to guess, these tolerances weren’t properly maintained or the buildings/roofs themselves weren’t solid enough to begin with.

If anyone can shed more light on this (pun intended)…

View on HN

sarcher's example of JLC. It does look like a faulty optimizer, which is a DC electronic for each panel.

As for Walmart, it's a lot to unpack.

> dozens showing hazardous problems such as loose wiring and “hot spots” on panels, according to court papers filed in New York State Supreme Court.

Loose wiring? A lot of setups are mostly plug and go. If they had to splice one wire, it's only a minute process to do and if they used their eyes for a few seconds (I'm describing the literal process of splicing a panel wire. I'm not demeaning the work. It's minutes. 3 max, if you brought the parts on the roof.) As for the other connections a solar installation may need, the time it takes it one minute for all connections on the roof. For the 244 panel setup of Walmart, 30 minutes tops (20.3 minutes to be closer to realistic because there might be 21 j-boxes, each connecting 12 panels to the main array).

Hot Spots? I have no idea about that one. It's too vague.

At a different location, this is what Walmart said: >One of the fires happened months after the system was de-energized, Walmart said.

I don't understand this one at all. Tesla/SolarCity must be using very cheap electronics because this shouldn't happen. Each panel's optimizer should be equipped for the situation when they aren't producing usable power.

Source: I worked in residential solar for about 6 months and as an electrician for two years. I'm not a bona fide source but I'm better than most because I've paid attention to everything that was taught to me. Plus I'm looking into starting my own Solar company.

The optimizer consists of MOSFETs, an inductor, and a microcontroller. The circuit is pretty similar to a BLDC motor controller.

The circuit relies 100% on the microcontroller outputting the right signals, otherwise a mosfet gets stuck on (causing overcurrent, and burning out). Little boxes like that are hard to waterproof and make resistant to repeated temperature fluctuations. My guess is some water went in and shorted some control signals, or repeated temperature changes caused a solder joint to crack. That caused the MOSFETs not to get the right control input, so they got really hot and caught fire.

It's just as likely to happen on a de-energised panel too - the optimizer is still connected, and the sun is still shining.

> The circuit relies 100% on the microcontroller outputting the right signals

I'm not an E.E. but does that apply for DC optimizers? Older optimizers had no AC conversions in them.

> Little boxes like that are hard to waterproof and make resistant to repeated temperature fluctuations.

Hmm. I never checked to see if the warranty for the optimizers are 20 years. The panels are but I'd be surprised to see the optimizers be less than 20. Most, non-cheap, optimizers do have insulation to prevent moisture from getting in.

> It's just as likely to happen on a de-energised panel too - the optimizer is still connected, and the sun is still shining.

Based on your description, that makes it a possibility. Why aren't the optimizers equipped with some type of regulator to prevent output in situations like this?

DC optimizers seem to be DC to DC converters, you rarely do those with MCU in professional setting, there are dedicated chips for it. Googling pictures of PCBs there are a ton of products with Aluminum electrolytic capacitors, those are rated at up to 10K hours at 105C, thats probably less than 5 years of sun divided by days whole thing gets cooked on the roof. Still that would merely lead to product failing to output power, MLCC and tantalum capacitors are the real problem, those love to short HARD under mechanical stresses, and temp cycling is a perfect environment to induce it.

mlcc fire https://www.youtube.com/watch?v=Q2rvAoO-MIA , causes https://www.youtube.com/watch?v=QgKY5QWehME

> Tesla/SolarCity must be using very cheap electronics because this shouldn't happen. Each panel's optimizer should be equipped for the situation when they aren't producing usable power.

It feels nice to see the truth come out. Cheap parts [1].

[1] https://www.businessinsider.com/tesla-project-titan-replace-...

lack of clearance below caused overheating and fire ?

lack of clearance between panels. if a roof settles a bit after construction it might flex down reducing the space between panels. (i think)

How would that cause a fire?

Once all the clearance is gone, panels can push each other and all sorts of breakage can happen.

At least in my installation they’re all strongly fastened to a cross bar. I’m not quite sure how they would shift in that scenario?

Theorizing...

Reduced gaps can limit/change air circulation. No matter the efficiency, solar panels absorb/create a nontrivial amount of heat that could be trapped underneath. That heat alone could be a problem or it could accelerate wear of other systems/parts.

You could also have breakage, rubbing, pulling, or other mechanical issues that can result in shorts and by extension fires.

What is in the article suggests poor quality control or poor quality installation to my eye.

The solar city acquisition still stinks to me of one Musk company buying another Musk company using investor money and shouting 'synergy' until people refocused on the poor quality of Tesla's car production instead.

Very interesting theory when it comes to the change in air circulation. But I can't imagine a situation where that would occur unless Walmart's claims are valid on 'poor installation'. The brackets, that hold up the rails, that go into the roof should lock in and shouldn't be any cause for concern.

The underbelly of the panels do run warm but they aren't hot, except in extreme heat. Nor is there any thought to the air circulation when designing a system, since the panels themselves don't have holes in the sides to encourage flow between an array.

Not enough air circulation is unlikely to get anything much over 100C, not hot enough to start a fire...

In OP's case, something electrical was almost certainly involved...

Reply on news.ycombinator.com