Beware of the SSRs

1

Ok here is what happened to me today
I have a 3kW water heater in the basement, controlled with Tasmota ( Sonoff basic), tied to 80A SSR with a big chunk of heatsink and an activ cooler.
Now ,the system worked for 4 years without any problems, and today the SSR caught fire.
No matter what I will use some industrial relay for the switching part.

Luckily I was at home and stopped the spreading fire .I just wanted to document this and to draw your attention that critical parts must be carefuly selected.
Thanks

1 Like
2

What brand/model was the SSR?

was it an off brand that wasn’t UL (etc.) rated?

3

it was an oversized Crydom,well beyond my needed specs. Now is a piece a coal

4

that’s a well known brand so I doubt it is shoddy workmanship or even a reflection on SSR’s in general.

It’s either just the way you were using it (not likely) or it decided to fail.

Sometimes those things just happen. Or I wouldn’t have a job as an industrial electrician. :wink: Things fail. Mostly for unknown reasons.

Glad you were there to prevent a disaster tho.

1 Like
5

Did you have a smoke detector that could have alerted you had you not been home?

Might be something to consider.

6

I don’t think that a smoke detector is a solution, tho it’s not a bad thing but till I’ve got home it’te. I’ve to come up with some better solution . An industrial relay in a metal fireproof box , this is a route that I’ll may take!

7

Yes, when I look back my only fault was not putting this SSR in some fireproof case.
It wasn’t abused at all, no PWM , it had a hysteresis of 10 C. Only operated about 4 times in one day for 1 hour per session. So probably some material failed in the process.

Funny thing I’ve had a pressure valve on the water heater,a standard mechanical temp controller in series with the SSR.
Lesson learned! Murphy was visiting me, I guess :grin:
Best regards !

1 Like
8

Look into how they calculate MTBF ratings, and how they use that to calculate annualized failure rates, they basically say that if something like if you have a mean time between failures of 10 years, expect 1/10 of the samples to fail in 1 year

9

failure can be catastrophic.

10

I wonder if the thermal paste dried out?