Blog

Fail Of The Week: Solid State Relay Fails Spectacularly | Hackaday

A lot of times these days, it seems like we hackers are a little like kids in a candy store. With so many cool devices available for pennies at the click of a mouse, it’s temptingly easy to order first and ask questions about quality later. Most of the time that works out just fine, with the main risk of sourcing a dodgy component being a ruined afternoon of hacking when a part fails.

The stakes are much higher when you’re connecting your project to the house mains, though, as [Mattias Wandel] recently learned when the solid-state relay controlling his water heater failed, with nearly tragic results. With aplomb that defies the fact that he just discovered that he nearly burned his house down, [Mattias] tours the scene of the crime and delivers a postmortem of the victim, a Fotek SSR-25DA. It appears that he mounted it well and gave it a decent heatsink, but the thing immolated itself just the same. The only remnant of the relay’s PCB left intact was the triac mounted to the rear plate. [Mattias] suspects the PCB traces heated up when he returned from vacation and the water heater it was controlling came on; with a tank full of cold water, both elements were needed and enough current was drawn to melt the solder build-up on the high-voltage traces. With the solder gone, the traces cooked off, and the rest is history. It’s a scary scenario that’s worth looking at if you’ve got any SSRs controlling loads anywhere near their rated limit. China Cj19 Contactor Companies

Fail Of The Week: Solid State Relay Fails Spectacularly | Hackaday

The morals of the story: buy quality components and test them if possible; when in doubt, derate; and make sure a flaming component can’t light anything else on fire. And you’ll want to review the basics of fire protection while you’re at it.

Or you could just buy a decent brand and not have your house burned down autonomously.

afaiu Fotek is a decent brand, but just like all other “good” brands there are plenty of fakes

Whose court system can you sue them in?

99% of the time you can’t even begin to sue because fakes were often dumped and mixed in throughout the supply lines so it’s nearly impossible to track it back to the factory that made the fake. And even if you could find the original factory, international lawsuit rarely works out.

The best defense is to make sure you buy from reputable sources, not from some backyard hardware store or eBay.

the fakers go for brand name recognition, so that put omron or fotek stickers on, not onehunglow

I’ll just leave this here:

https://www.ul.com/newsroom/publicnotices/ul-warns-of-solid-state-relay-with-counterfeit-ul-recognition-mark-release-13pn-52/

As far as litigation goes? I know enough about Chinese law to know that it’s really up to Fotek and whether Fotek applied for trademarks in China or not, and whether they did the rest of the intellectual property legwork in China (which is really tricky because China doesn’t use common law like Western countries do, so it’s a harder system to navigate). If they did, then yea, the counterfeiters could probably be sued although it’s a whack-a-mole game, the counterfeiters come up faster than they can be whacked down. But if Fotek didn’t go through the China IP process to protect their brand, then the counterfeiters may have actually applied for the trademark themselves in China, gotten the right to use the name, took the likeness and everything, legally (at least in China). Best Fotek could do at that point is try to keep them from being imported into the USA (similar to the Sparkfun fluke-colored DMM’s a few years back) but again that’s a whack-a-mole game that they wouldn’t win.

I first thought we were about to see the actual moment when it was about to blow. But this is post mortem autopsy.

https://www.youtube.com/watch?v=1BUciKwRfuc

A REAL Fotek is a good brand ….. however some dishonest shops in China are turning out fakes by the thousands ,,,,, I got burnt by a fake Fotek 25A DC SSR for a printer a couple of years ago and the 2nd or 3rd time I used it the MOSFET went up in smoke although nowhere near this spectacularly …. I ended up replacing it wtih a 100A Fotek that I could ascertain was genuine and it’s been working ever since

What kind of a heat sink did you have on it?

I bloke I know swore black and blue “Solid state relays always fail ON”

then I showed him a few like this!!

Well yeah, that’s what I’ve seen, they fail on and stay on until either they or the piece of kit they are running catch fire.

all of the ones in my little box of horrors had craters blown in them

He did specifically mount/build that setup around the possibility of failure. He talks about it in the first few minutes of the video. Kinda a hyped up headline.

It is not uncommon to see lower than specified rating triacs on these cheap Fotek SSRs. Look up on youtube, it is probably a fake one.

Watch the video. If you doubt this gentleman’s conclusion, oh well, your problem.

I’m constantly amazed by the use of a giant blob of solder to increase the cross section of PCB traces meant to increase their current carrying capability.

That’s a common solution used my mainstream manufacturing.

Yes and it is a bad solution. Solder has much lower conductivity than copper (~ 1/10).

You are correct and if you design this type of solution you already know this. With PCB design, the alternatives are heavier copper, wider traces, plating up copper in the areas that carry heavy current or a combo of any of the three. But you can only go so far with these solutions. Hack apart a known good SSR and I bet they use the exact same solution.

I’m an old wire-wrap guy and I know better!

Wire wrap introduces it’s own list of issues, most centered around technique.

True …. but it’s also 1/10 the price of thick copper plating ……

As an engineer over 20 years ago I coined the term Beancounter Engineering this kind of stuff where some Idiot MBA sets the specs based on saving a few pennies rather than a Electrical Engineer that understands basic (and complex) science. If you are too damned stupid to understand the basic science behind Global Warming you have no business being an engineer or telling engineers what to do

It’s as simple as that

Haven’t seen such an accurate comment in many years. I am a 50yr engineer from the old school, still studying data science and at the coal face now. Not many left any more. Seen many “MBA Engineers” passing themselves off as “certified engineer consultants” Even engineer association support incompetent engineers, and incompetent test labs, as long as they pay their membership, and those used by their mates in government as experts. Specs written for algorithms used world wide that have anomalies that are hidden by manufacturers. And so on. Govt labs with untrained low cost “agricultural engineers from the east ‘ with minimal qualifications ands skill, as they are cheap. Low price is what people want to day, and thats why it ‘s accepted practice.

The 40 Gallon Hot water tank takes 22 Amps with both element on. And you used a 25 Amp relay. What really happened is you under rated your relay. Solid state or Mechanical does not matter. At the Max for a load it should never use more then 80% of your ratted load of your switch/relay/controller.

One more thing. If you are not a electrician you should not be playing with Live power. Like you said you almost killed your family, And Possibly more people.

I think he deserves at least a kmol of +1s: 6.022E26

BTW, a couple I’m acquainted with chose 10/23 as their wedding date.

1 to the power of anything is still 1 :)

80%? When dealing with high current loads I top at 70% to allow for wobble of line voltage. For something like a Water heater I’d drop that to 50% ratings to make sure the surge doesn’t cause failure.

What kind of surge would you expect from a water heater (basically a giant resistor)?

The concern isn’t a surge from the element, but a surge from the line itself.

Do you really expect your line voltages to vary by more than a % or so?

@Carios (Hit report instead of reply yada yada)

Sure I expect my line voltages to vary especially given that the actual spec for my line voltage (UK) is 230V +10%/-6% – my nominal voltage where I’m sitting is around 242V and I’ve seen it go as high as 252 and as low as 220.

Even in the US the spec is for 120V +-5%, so even in-spec allows for a 10% variance.

Heater might be pulling 22 A when hot, but on power up … likely much more. Think about incandescent lightbulb or electric motor, for example.

Temperature swing in the water heater may be about 50K, incandescent bulb filament temperature swing is closer to 3000K. I doubt there is much variation in the resistance of the water heater’s heating element

If your water heater is emitting photons in the visible spectrum you are doing it wrong.

There is a huge difference between heater and electric motor when it comes to inrush current. But anyway haven’t seen 600A contactor for switching on 60A motor.

Oversizing switching gear is good to some point but instead of relay withstanding surges I would prefer breaker that would protect everything.

Some heaters are a coil of Alloy wire cast in ceramic and then enclosed in a stainless steel tube. Basically a bad inductor, but still having the surge properties of one. Same with big boy halogen bulbs.

In my college days, designing stage lighting systems for customers, I had a test rig that I used during a sales presentation. The reason I built this is that a customer had his electrician contracted to do the install work (we didn’t do the electrical install, just specification). When I specified 20 Amp breakers and told the customer that they’d be limited unless they upgraded their service, they told me that their guy had taken care of it.

I walked in on instrument installation day and began the burn in test of running the halogen fixtures at 100%, then bringing it down over the course of 30 minutes to determine if we had a bad SCR dimmer.

Snap! Didn’t even get the first instrument lit. I began to investigate. Wiring size, check. Dimmer output side breaker, still on. Go further down the line and find that the electrician had indeed taken care of it. He re-did my calculations, which at full nominal load looked like a 15 amp breaker on the input side of the dimmer would work just fine. What they didn’t realize is that when you bring an instrument or instruments (depending on the circuit) to full on, you go beyond even 20 amps for just a moment, then it goes down to about 13 amps.

The customer wasn’t pleased. I had obviously done something wrong. I showed them the paperwork with the specifications showing that the input to the dimmers was to be 20 amps. They insisted that their electrician knew better than I did.

i built the rig that day. I put an ammeter with max hold on a circuit, plugged the circuit into a known 20 amp connection, and put a load on the circuit that simulated the instrument. Turn it on, 22 amps max hold, 13 amps nominal.

The customer changed their tune. My boss was impressed (though not enough to give me some coinage for ingenuity) and the electrician was told to change the breakers and add the new feeders to support the load. Got a lot of recommendations from that customer in the years after that.

So a 20 amp breaker is designed to take the load shift on startup. A 15 amp breaker will carry the nominal load, but that surge on startup will kick it every time. In the days before switching supplies the input voltage would vary from 108 to 115VAC where I did work. This occasionally caused breaker trips on the unprepared circuit. I learned to spec high enough to minimize that issue and keep a safety factor in play.

The electrician’ s first duty is to size the breakers to protect the wiring system so depending on what he found the electrician may have been right. It’s not the overall rating number of a breaker that determines whether it will take the inrush load but the characteristic curve. A B-type breaker will take an inrush current of 3 to 5 times the rated current for several seconds; a C-type 5-10x; a D-type 10-20x so I smell a whiff of garbage in your story my friend. I don’t know what your true inrush current was and a max-hold meter might not tell you either. You use a scope or a very specialist instrument to get the inrush current, or ask the manufacturer. One way or another or yet another your story says you were pretty underqualified to be designing that system, you were lucky to get away with it and your brash confidence suggests you didn’t go on to be an electrical designer and rack up the experience to tell your own story right.

Agreed, the component is under rated from the start. If you look at a datasheet from a 25A SSR from a better company you will find something like de-rating curves which really show that as you move from the ideal conditions (20C temp, infinite heatsink) you should only use a smaller load. It is pretty easy to reach a limit of half the max load in normal conditions.

Then again, like other counterfeit assemblies…. they might rely on the installer using a 25A relay for a 15A load which is why they put a counterfeit 16A triac inside which the manufacturer actually put a die of a 12A because he knows his user will de-rate the max current anyway.

I seem to remember one test video I saw online, possibly BigClive, finding that the actual rating on the faux-Fotek SSR that he was dissecting was rated at around 10A, in a 25A labeled SSR. I believe it was also a MOSFET instead of the SCR or TRIAC that it was supposed to have, as well.

I’m a bit spoiled as far as AC load SSRs go, since I get IDEC and Omron puck SSRs from obsolete equipment at work, but always exercise due caution when I have to go shopping for DC load SSRs. Far too much fake stuff out there, masquerading as reputable brands, some to the point of ruining the reputation of the brand. (Fotek… For the longest time I thought they were only a Chinese brand, like the various Harbor Freight “brands” that they slap on their junk tooling offerings.)

And now you know why companies react the way they do. It would be like someone going around deliberately ruining your reputation, and there’s little you could do about it.

I couldn’t even find a de-rating chart or graph online for this relay. Also, many SSRs (and other devices) do not recommend the “bottom” orientation for heat sinks.

The leads on that TO220 case SCR and that solder trace are not going to handle 20 amps continuously without heating up. A SCR that’s up to the task of carrying 20 amps for hours on end would have a TO-218 case or even be stud mount. If all else fails you could just use a contactor relay which is what they normally use for this application.

Where I do agree that it is under rated, as far as I know, Mathias is an electrical engineer. So he should be able to work with mains, and he should also know better.

Both elements weren’t on….elements in electric HW heaters are cascading…they cannot come on simultaneously. Bottom element is powered through normally closed side of upper element relay. Therefore, if upper element is on, lower element is off.

A 40G HWH uses 3000W elements. Constant draw is 12.5A@240V and requires a 20A breaker (max draw / 80% safety margin = 15.6A so round up to 20A). 60G uses 4500W or 18.75A@240V and requires a 30A breaker (23.4A @ 80% safety margin so round up).

The height of this tank tells me it’s 60G.

All in all, the current draw through this device was below it’s “rated” value. So either it’s a fake/poorly produced or was simply defective.

Probably best to derate to 1.5x or 2x. But even then, are the “solder” conductors inside really going to be any different? Looking at then they appear to crate a conductor way bigger than 25A requires.

4500 watt elements are pretty common too.

OK, it’s called a Water Heater(WN) -not a Hot Water Heater (HWH) as the water entering the tank is cold.

A 25A relay seems perfect for a 22A load. If the relay is really built for 25A and not only marked as one. What use had a maximum rating when you could never really use it?

It’s the peak rating. For constant loads, or anything that could be temporarily constant you need to de-rate.

Technically, the electrical code calls anything on for 3 hours straight “constant”

Your advice / admonition overlooks the reasonable likelihood that the part wasn’t *really* rated for what it *claimed* to be rated for, nor was it constructed in a manner to keep it safe in the event of an actual failure– as in the article’s example. Such that even if he were a licensed and well-experienced “electrician” (or anything else), it STILL would (could) have failed and caused the same problem. So what you’re saying IS good advice in a general way, but only partially relevant to the actual incident from the article, IMO of course! ;-)

In the case of Chinese relays and SSRs I double the amps and add 10% …. Ditto for Chinese switching regulators which put out everything but DC after they reach about 50-60% of their rated load

Water heaters never use both elements at the same time. You would have to rewire it for this to happen. Look at the circuit.

Here I have always wondered. If the power bill is such a concern, then why not buy a heat pump for water heating instead? (It drawn typically around half to a third of the power to give the same heat output.)

Though, how ever, even with a heat pump, one could lower the power bill a bit more if one takes the power when it is cheapest.

Heat pumps probably does not work that we’ll in the Canadian winters.

Coastal BC might be viable for them, here in the “temperate” bit of Ontario they’d be good in April and October.

True. If air is the source for the “warmth” then a cold winter isn’t really of help.

But there is other sources for the “warmth”, all though, if one lives on permafrost, then many of those sources are off limits. Since it is typically either boring a hole into the ground (Not always an option due to geological reasons), putting a hose that is coiled under the surface of the ground, typically a meter or so down. (the ground temperature a meter under ground is typically rather close to the average year round temperature (though dependent on what soil one lives on). We would find permafrost if the average temperature is below freezing.) Or sinking a coiled hose to the bottom of a lake or river. Since the temperature at the bottom of lakes is typically around 4 degrees C. (This though requires that one has a lake on one’s property that doesn’t completely freeze to the bottom… (So a depths of at least a few meters is typically required for cold climate.))

Then if one is only looking at getting warm water, then one can also just have the heat pump use ambient indoor temperature to make the warm water. It will efficiently give warm water, but heating the house would be just as efficient as using a large resistor. (But at least the warm water got cheaper. (and it cools down the house during summer if one empties the warm water out doors or down the sink.))

The controller board for the heat pump in our “3 Season porch” fried last January during a cold spell. All of the house plants we were keeping in there froze to death. SE Minnesota

Considering the existence of deep subzero freezers, it would still perform better than a resistive element. Or have it draw the heat from the indoors and run some crypto miners to make up for it.

Are natural gas powered water heaters unheard of in the US?

Nope, all three of mine are propane, NG would cost about $20,000 to have a line run.

Or those solar panels made of black pipes that heat water?

Solar water heaters are nice. Sadly, when one gets far north, the sun isn’t a common sight during the cold parts of the year.

Depends how far north you are – direct sunlight isn’t required for them to produce useful heat. If you’ve got a months long night then they’re no use though.

Our heat pump works in conjunction with a gas furnace for the days when it drops below maybe 35f outside

Heat pumps are good as pre-warmers but it’s hard to get really hot water from one without wasting a lot of electricity in the process as you exceed the “point of diminishing returns”. Natural Gas is still the cheapest most efficient way to heat water because it’s fairly easy to make a 95% or better efficiency burner. In Canada you’d likely have to go geothermal which in most cases is nothing but a heat pump hooked to an underground water loop

Of course in many parts of the world that’s not possible like it is here in the US where better than 60% of homes have access to NG

can’t be said too many times: foteks are known to roll over, fart and let the smoke out. doesn’t hurt to overspec by 3-5. not when playing with fire.

It also doesn’t hurt to spend a bit more. I found a 40A SSR, well-known Japanese brand, from a reputable distributor, for $30. That said, I might easily have made this same mistake; thankfully I won’t have to learn the hard way (this time!).

I have been led to believe (largely by bigclive) that Fotek are OK but there are lots of fake Foteks out there.

I have one SSR marked Fotek with 40A on the label and a 16A device inside. I feel a bit sorry for the real Fotek if this is the case.

could be that the fakers just put new sticker on with higher rating to get higher price

there are Omron fakes too

Maybe they’re at the same kind of place Rubicon were during the capacitor plague, actually made by Rubicon caps were okay, exact visual copy fakes and Rulicons, Rubicows Rubycoms etc etc were not

Youtube helpfully suggested this teardown of (apparently) the same relay. Who knows if any of these are fake. [youtube https://www.youtube.com/watch?v=DxEhxjvifyY&w=560&h=315%5D

I design and we manufacture controllers for homebrewing beer and in 2015 we were bit hard by counterfeit FOTEK SSRs, and this was after many years of having no issues at all with the brand. Our problems started when switching from buying direct from a Chinese supplier to buying from a Amazon supplier and the mistake cost us $7,000 in one month in warranty returned products. The issue was rated product built with substandard parts – in other words 40 Amp SSRs were built with 20 Amp TRIACs!

As a result I wrote this article in 2015 outlining the issue, what I did to ID the problem and the solution. https://www.instructables.com/id/The-inner-workings-of-Counterfeit-FOTEK-SSRs/

But don’t believe brand names are any better, especially if you buy them from eBay or Amazon! There is too much profit to be made remarking parts to trust any of these type parts from anyone but a major distributor like DIGIKEY or MOUSER Electronics.

In the old days, waaay back in the 20th Century, you’d buy something from a vendor that had a good reputation. It was in their interest to keep their reputation, because that’s where customers came from, so they’d make sure to only buy their materials and parts from reputable suppliers. And they’d do the same, forming a chain of reliability and trust built up through experience and everyone’s mutual need to keep up a reputation.

Buying from people you’ve never heard of, to get stuff cheaper, which is the only reason you’d use Ebay or Amazon, bypasses this whole chain of trust. And, fuck me! it turns out buying stuff from anonymous suppliers, in a market where Chinese scammers are an utter epidemic, sometimes means people don’t get what they expected to.

There’s a simple answer to all this.

Don’t blame the Chinese for this one! They are more than happy to sell us any quality we want. For example, you are probably reading this on a very well built, Chinese made, smart phone.

The issue is we are more than willing to make unknown compromises for price. And we don’t just make these compromises with electronic components. We make the same compromises in the vehicles we drive, the foods we eat, the houses we buy…..

Tell that to the parents buying baby formula adulterated with melamine. The Chinese have long standing traditions that support cutting as many corners as possible. Basically, cheat until caught. And their government seems to actively turn a blind eye to all manner of this nonsense.

>Basically, cheat until caught. I’m a high schooler in America, and my “fellow peers” pull the same bullshit- and then when they get caught, they get mad at the teachers. >And their government seems to actively turn a blind eye to all manner of this nonsense. Aaand so do peers at my school.

Yes, I get that there are societal differences between Eastern ideals and Western ideals. Please understand that Americans also build cheap crap, too- but being across an ocean makes buying and selling a little different, especially with a language barrier and different societal norms. I’ve been bitten by a couple of misrepresented products before on AliExpress, (RGB mechanical switches that were actually just RGB LED’s), but that was my fault- I didn’t look very carefully, and didn’t do my due diligence.

If you had the choice between two identical-looking boards, with the same pictures, and review ratings, would you spend an extra 20¢ on the more expensive one? That twenty cents really adds up, you know- if I was buying 1,000 ESP-12 boards, for example, that would be two hundred dollars saved, and you would be pretty happy with yourself. That is, if it works. Otherwise, you feel really dumb.

And for the record, If I’m able to, I will breastfeed my children, because nutritional value, probiotics, and antibodies.

I think there may have been executions over the melamine thing.

“The Chinese have long standing traditions that support cutting as many corners as possible.”

You just described Capitalism in general ….. Every corporation I worked as an engineer for did those exact same things all the time and that was over 20 years ago before China even became a player …. American automobiles in the early 80’s are an even better example

The Chinese aren’t any more corrupt or corner cutting than any other manufacturing country ….. Since they are 3 1/2 times larger than the US then I’d expect to see 3 1/2 times more corruption and corner cutting …. It’s like CO2 output, yes China puts out more as a nation but per person the US puts out over 3 times more CO2 plus we had a 100 year head start which is why we are more responsible for the Global Warming problem (along with Europe) than China is by a considerable margin …..

Still, most capitalists, when you order something, don’t ask “what do you want it for?”, so they can substitute something else without telling you. Generally in contracting with capitalists, mysterious “problems” don’t regularly as clockwork pop up, increasing the cost above what was agreed.

Sure you can say it’s cultural and you have to take this into account. It’s still bloody dishonest though! Just means their culture is dishonest. Same place where people skim fat out of the damn SEWER to cook food, and it ends up contaminated with all the industrial chemicals that were illegally dumped down the sewer.

China is a shithole. Business-wise it’s like the West was 300 years ago, when your “flour” was cut with chalk dust and all the rest. Except apparently in Chinese culture it’s seen as enterprising and showing initiative, to concoct little wheezes like this for a profit, rather than the thievery our own culture for some reason sees it as.

China needs to sort it’s shit out. Eventually, they might, then that’ll be OK. They may evolve the same way the West did til their business standards are reputable. Of course, that all happened to the West around the 19th and 20th Centuries. Since then, in the West, corruption has increased, and political bodies are bought and paid for by gigantic mega-corps, seeking monopolism and every other bent tactic they can try, for an eternally increasing share price.

So China are stuck before capitalism’s stable period, the West are now past that. China might end up passing through it, or skipping it altogether. Actually the West might end up learning a few tricks from them.

For something like this I would go to a proper electrical supplier and probably would use an AC contactor relay that’s rated twice the current. Sure a contactor needs a 24V transformer but it’s a small price to pay and the current use would be insignificant compared the the hot water heater.

The few I bought on AliExpress all had 50% rated triacs in them vs. description. “Luckily” I contested them all for 50% my money back and could use them for lower power projects later

I use a variety of a/c and dc fostek ssrs. I have yet to have a problem, but i use them in circuits where 50% of their rated current is the absolute maximum they will ever see.

We were pretty close to that 40% rule when our controllers started to fail, which is probably why they only started failing after our customers had use the controllers a few times. We were pushing 22 Amps through 40 Amp rated parts. In other words, using a component at half its rating still won’t cover for substandard parts! Our solution was to find a part that was manufactured to spec and also use the component at half its rating and we’ve not had a verified SSR related failure since.

Ultimately under-rating won’t solve the problem. The Chinese scammers are already putting below-rated parts in, because they assume most users will under-rate anyway, so they’ll get away with it. If, in response to that, you under-rate even more, and this becomes mainstream, what do you think will happen? They’ll just take it into account, same as ever, and the 75A parts you’re putting 40A through, because you want to be careful, will only contain 20A triacs.

I think in the long-term this shameful scam-fest won’t last. Because stuff will keep bursting into flames all around us, for one thing. But hopefully people will realise that Ebay’s “rating” nonsense is only valuable if the people who give the ratings actually know what they’re talking about. The whole point of counterfeit stuff is you mislead the customer so they don’t.

Hopefully the era when people did industrial sourcing from an overgrown Beanie Baby marketplace will be looked back on by future generations, and they’ll laugh.

Haven’t really looked at how the eBay feedback system works these days, but is it still the case that if a buyer gives a seller a negative no matter how well deserved, he can pretty much expect a negative in return?

No, sellers cannot neg a buyer any more (retaliatory feedback)

No, ebay removed the ability for sellers to leave negative feedback years ago. Fun fact – the majority of an ebay sellers score these days (not the % positive feedback, but the score used for determining search rankings etc) is determined by the % of orders that are received on time and without a ebay ‘case’ (return, etc) being opened.

Have to be ohnest I would never ever consider a SSR for such an application even if it was a genuine unit form a reputable supplier. this is a perfect example where a relay should be used, tried, tested and ought to fail somewhat gracefully if it came to it. Not snarking at the OP, but any situation I have used an SSR its been massively over-specced for the load im driving, i dont think I could sleep well at night with an SSR controlling tens of amps at hundreds of volts.

Still can’t bring myself to install SSRs on “high current” applications, either. High quality mechanical relays.

There’s a proper place for a relay and a proper place for a SSR. And I agree this solution should have been a relay.

On our case a combo was the right solution. We use a contactor (a relay) to switch incoming power because it’s either on or off. But we use a SSR to switch elements on and off because they cycle on/off many times a minute.

A couple months back I ran across a note in the datasheet for an inexpensive but reputably sourced power MOSFET that said that, while the die itself was rated for 200A continuous/400A peak the package and its leads were only rated for 125A/250A. So I said to myself “I wonder what the mode of failure looks like?” And built up a circuit to drive it to failure in a controlled manner. It was fairly spectacular: 1) the thermal expansion rate of the die doesn’t match that of the package and once you exceed some wattage tension builds up and (at least in the case of my particular DUT) the chip package literally exploded sending shards of epoxy and/or ceramic flying far and wide. 2) with the die now exposed in several areas and the connection to the heatsink thoroughly hosed by the explosion of the package things devolved rapidly and the magic blue smoke escapes (which smells very bad, FYI, and takes a while to fully purge that smell from the workshop).

This seems to me that your experiment failed due to thermal failure cased by insufficient cooling and not the part itself..You stated yourself that the part failed due to thermal expansion (insufficient cooling). But it’s still a valid experiment with valid results. And you show that it’s up to the designer (or designers) to design the entire product, which includes sufficient cooling. This is one area where home designs fail – a lot.

Being in the controller business I’ve been presented more than one friend’s design with insufficient cooling. One friend even mounted his SSR to a decent heat sink then mounted the entire assembly inside his housing with no air circulation. His response was “it will work” and he’s had two SSRs fail that I know of. At least in his case, the steel housing kept the “magic smoke” contained!

I’m surprised no one mentioned how Mattias Wandel mounted the SSR on a piece of sheet metal (pretending it will be a heat sink), then mounted it on a water _heater_. I know the heater is insulated so it really shouldn’t be radiating any heat, but wouldn’t it have made more sense to mount the SSR on the cold water inlet pipe? You want to heat the water anyway… When I saw his first video showing off the project I said to myself that’s not going to end well. I’m surprised it lasted this long.

I did, read the posts.

I had expected the actual leads or the bond wires to fail.

video or it never happened.

I think every power MOSFET in my box, most of them from reputable sources, have packages rated below the die.

The only way, AFAIK, to know how much current it is actually supposed to be capable of is to calculate it based on the listed thermal resistance from junction to case, the thermal resistance of your heat sink from case to ambient, and the Rds(on).

Everything else on the datasheet are useless observations only true under test conditions. A few really awesome datasheets give junction to ambient, which is a blessing if you don’t need a heatsink.

And for example the max continuous current one is “rated” for is 20A, and when I calculated the values without any heat sink, I should be able to run about 700mA. A lot of people forget to do this step before de-rating.

I finally took the time to watch the video and I see two problems.

1. He mentioned grease – higher current SSRs are sweat soldered to the base plate, they don’t use thermal grease.

2. At the 1.8V voltage drop Mattias Wandel mentioned in the video, his SSR was dissipating about 40 Watts of energy and the metal bast plate he mounted the SSR to was not sufficient for dissipating 40 Watts of heat energy.

The melted solder was probably a secondary failure. When SSRs fail they often turn into heaters as his did and the heat from the base plate itself melted the solder before incinerating the plastic housing.

True, when the solder starts running something has gone wrong already.

If you look at the Fotek datasheet for that relay, you’ll find that the specified heatsink does not resemble a piece of galvanized steel sheet metal, horizontally mounted with a half inch air gap under it, and set on top of a heater.

Comparable relay product datasheets from Omron suggest that without appropriate heatsinking, the current limit should be in the range of 4A to 7A.

I don’t think this was appropriately mounted, regardless of the part’s origin.

I buy Crydom SSRs from Digikey. I pay the $40-$50 bucks, I use big assed heat sinks, and I sleep like a baby.

1. 25A for 22A? Too close for my comfort. 2. A thin sheet of aluminum is not much of a heat sink. 3. Sitting on top of the hot water heater means ambient temp is going to be a bit higher. 4. Summer in a garage is likely to get hot.

Looking at the datasheet for the Fotek SSR-25DA: https://cdn.sparkfun.com/datasheets/Components/General/SSR40DA.pdf

Current starts derating at about 82C (see page 12). But look at the size and configuration of the heat sink required for a 25A SSR, it is 100mmx100mm with lots of fins (see page 13). That will be with lots of free air, preferably mounted so the fins are vertical or with some forced air flow.

Even with a large heat sink like that, shrink it to 50mmx100mm and the datasheet lists a max of 10A. With that aluminum plate, I’d be surprised if it could handle more than 5A for a length of time.

I think this could have failed purely due to the inadequate heat sink, made worse by elevated summer temperatures. Add the possibility of counterfeits in, it was doomed to failure.

There is a reason SSRs come with four bolt holes and a wide, flat aluminum base. And a reason manufacturers use these large heat sinks on them, and don’t forget a thin film of thermal paste: https://sc01.alicdn.com/kf/HTB12s_2KFXXXXbqXXXXq6xXFXXXM/aluminium-led-heat-sink-radiator-heatsink-heat.jpg

What about putting a uC in SSRs, so you can ask the think if it’s an original part or not. Propper encryption needs to be implemented of course.

Anyone “hacking” an appliance that can literally burn down your house, voids any safety certifications the appliance was manufactured with. The insurance company would ‘probably’ have the option of simply denying any claims arising from a fire caused by the hacked appliance. This is speculation since I’m not an attorney, but it would be prudent to consult with a knowledgeable agent before undertaking any such projects. My policy specifically excludes “events” as a result of deliberate modification of appliances that the manufacturer of said appliance has not authorized.

I know some policies that clearly state work needs to be performed by “duly licensed” personnel (ie. you better be a licensed electrician or plumber). That’s one of the reasons I did not want to install my own automated transfer switch. Insurance company would possibly deny coverage in event of a mishap.

I use these kinds of SSRS in high current applications driving kilns. we protect ourselves quite happily by having a contactor in the line. The SSR when it fails fails ON and the contactor is help closed all the time but overtemp or any other kind of alarm will release teh contactor and break teh circuit. My recommendation – and no one wil ever get to the bottom of the comments on this feed – is to add a contactor in series which is help closed and opened in an alarm condition. Then the SSR can fail to its heart’s content and you’re good. The SSR is of course much better at switching on and off than the relay which will fail eventually from on/off transients. As usual – YMMV

Getting reasonably priced SSRs that are trustworthy seems almost impossible. A 40A TRIAC costs $34. I know there’s more than the TRIAC in the SSR, but it’s not like there are that many components either… I end up using relays if at all reasonable and scratch my head when they’re not…

SSR for a water heater :| Get a mechanical relay, problem solved.

It really sounds like the issue here is only anticipating a steady load and not a cold start (figuratively and literally) when choosing componenets

For this application I would have used a contactor from a reputable supplier that breaks the connection on both legs of the 220V circuit vs a cheap ebay SSR. https://en.wikipedia.org/wiki/Contactor

Meanwhile in my house (uk) we have a street lamp time switch with a single pole switch that looks like it was designed to handle 30A. It is beautifully engineered from brass that increases the on period as the days get shorter, I felt bad hacking it so that it would turn on my water heater for an hour twice a day, a job it has been doing for 17 years now without incident. Electromechanical is by far the best way for high current loads at any voltage as fatcas I’m concerned.

Or the other takeaway is that only some piece of esoteric surplus made several decades ago is the right thing for the job and then one can go have a philosophical crisis about whether at this point we’re just plundering the ruins of a failed civilisation or not. ;-)

I didn’t know mathias was a pixie wrangler, i thouhgt he was just a wood fairy.

Relay was severely underrated for the application. Also the real moral of the story is that when you use a ten dollar relay, you get ten dollar performance.

I’m Electronic Design for SSRs, among other sensors. I just wanna give a briefly recommendation for further SSR homemade projects (I know it sound obvious, but then we have cases like in the video): -Measure what’s the maximum current of your load. -Consider max. ambient temperature, if the SSR+HS(HeatSink) are enclosure, consider to add the heat dispirited from the SSR. -Based on datasheet’s thermal derate curves, select an adequate HS. -In this case, the plate use as HS was attached to the heat source, reducing the capabilities to dissipate SSR heat. (do you agree? or the top of the lid of the water tank does not get hot?). -The position of the HS, plays an important key to dissipate the heat, place it vertical for better performance. -SSRs have a variety of technologies embedded, look for the one that best suits to your project, in this case, a timer SSR could reduce the set-up. -Verify your product has agency besides CE (anyone can mark this one), like (UL, TUV, VDE or CSA). -If possible, use a SSR with SCRs instead of TRIACs :)

It would be interesting to see an SSR on a cool plate, where the input cool water helps to lower the temperature of the SSR, and the output hot water goes to water tank.

ssr for an heating resitor? …you must expect this result… why not use an relay that its designed for this current…

That was strange, all the house sized electric water heaters I’ve worked on were 220Volt on a double pole circuit breaker, I have DUAL Solid State Relays that I use for turning 220 volt pumps on that I rate at twice the circuit breaker rating to be safe. How did he wire a single pole solid state relay to control two hot legs?

My explanation is very different: a current/voltage surge (from lightning probably). According to the SSRs datasheet, it is rated to withstand 600 V AC. However, without any surge protection (MOV, GDT, TPSD, etc…), voltage will rise much higher.

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Fail Of The Week: Solid State Relay Fails Spectacularly | Hackaday

High-Quality Contactor Relay Suppliers By using our website and services, you expressly agree to the placement of our performance, functionality and advertising cookies. Learn more