Triac failure modes

[tooki]

<back story>
A few months ago, we picked up a Xytronic IR610 quartz preheater at work. Recently, I populated a board and put it on it to soak at 100C. I stepped out of the room for a few mins (with someone else there to check on it), and a few mins later, they came to me and said there was a problem. Well, the board was toast, burnt to a crisp, with the now-black epoxy having bubbled out from within the board...

It didn’t take me any time at all to determine what happened: the preheater, despite being set to 100C, was heating at full blast (800W) continuously. </back story>

I opened the controller (which is basically a lamp dimmer with digital controls; there’s no temperature feedback) and quickly my suspicions were confirmed: the triac had failed short. No visible damage on it or any other component. I ordered a replacement triac, installed it, and the problem was fixed. I turned it back off to redo the wiring properly (to add ferrules, zip ties, etc), did that, and then turned it on again, only to discover the fault was back. The triac had failed again.

Clearly, there’s a bigger problem here. Can anybody give me some hints on to what to look for? The triac is mounted onto a heatsink which itself is bolted onto the steel housing, and it did not appear to get hot (neither before nor after failure). The triac is a type (BTA12-600CW) with an insulated tab, so no unintentional shorts to ground. The 800W of the heater is under 4A at 230V, so well below the 12A rating of the triac. I don’t think ESD damage is to blame, since the room is a fully ESD-safe lab, and I took all the usual precautions (not to mention that the original triac failed while within the device). What are things that cause triac failure?

I’m hoping there’s something simpler I can fix so we don’t have to ship the thing away for repair.

Thanks everyone!

[2N3055]

Take a look at driving circuit. Gate is sensitive to overvoltage and overcurrent. Those dimmers usually have horrible driving circuits..
It might be easier to just replace dimmer circuit with another one that works better.
Or to just dump in an analog input SSR, an whip up a 10V regulator with potentiometer for input.
Or of you feel like doing it right, to drop in a PID temp regulator. There are ready made temp reg.... Maybe no need for PID at all, a simple on/off reg would do the job ..

[tooki]

Take a look at driving circuit. Gate is sensitive to overvoltage and overcurrent. Those dimmers usually have horrible driving circuits..
It might be easier to just replace dimmer circuit with another one that works better.
Or to just dump in an analog input SSR, an whip up a 10V regulator with potentiometer for input.
Or of you feel like doing it right, to drop in a PID temp regulator. There are ready made temp reg.... Maybe no need for PID at all, a simple on/off reg would do the job ..

”Those dimmers” is an awfully broad statement, dontcha think? It’s not an actual light dimmer, you know...

I’ll take a closer look next week, but the control part of the driver is a small microcontroller powered by a 7805 regulator. It hasn’t fried, so I doubt the regulator is over voltage. Some part of the circuit is probably doing zero-crossing detection, since the board has two optocouplers. (So I assume one going out to the triac and one coming in with ZCD.)

[2N3055]

Well, you did say light dimmer ..
If it uses optodriver it will probably be fine then.
Since load is not inductive, snubber is probably not a problem either. One thing that comes to mind is ceramic heaters have positive tempco.. so they have lower resistance when cold. Did you try to measure heater resistance?
And to make sure check circuitry arround optodriver and opto too..

[Gyro]

I know that large area puck SCRs can suffer problems with localised hot-spots if their gate drive rise times are too slow. I don't know if this could apply to smaller parts - I suspect not.

[james_s]

Maybe it's a counterfeit triac that is not actually the rating it says it is? Even so, 4A is below what most TO220 triacs can tolerate. Intermittent short in the wiring to the heater? Large spikes on the line? Wrong value resistor somewhere in the drive circuit? Triacs are usually very reliable.

[2N3055]

I know that large area puck SCRs can suffer problems with localised hot-spots if their gate drive rise times are too slow. I don't know if this could apply to smaller parts - I suspect not.

Yeah, you're right, small ones can have same problems with slow gate drive, but it is not hard to drive them properly, and they are not big enough and currents are not big enough to see that often, unless you drive them right at the edge of specs ... or drive circuit is damaged...

[floobydust]

You have to be methodical troubleshooting repetitive failures. Is it overcurrent, overvoltage, over temperature, bad design etc.
Guessing is OK but it's best to know what is going on. Maybe post pics of the circuit and we can try figure it out.

A heating element is a crappy resistor - it has inductance due to the coil, and it has huge inrush current because the resistive element goes through a several 100's C rise. Not as much as an incandescent light bulb (10:1) but still several seconds of high current. You can measure the heater's cold resistance and I've seen some heating elements have shorted turns when hot due to the wire's expansion and the adjacent coils touching, or causing a ground-fault and the heater current surges high.
Inductance usually 10's-100's uH (measure it) not as much as a motor but enough to bite. For this reason I find a snubber is necessary despite the EE notion that a heating element is a pure resistor and ST calling their triac "snubberless" for the dV/dt aspect.

Sometimes manufacturers use the same circuit for both 120V and 240V versions of their product, so the component values are wrong for the gate drive and snubber.

I've had many crappy chinese triacs fail, even way under current and voltage specs they get senile hog gate current and eventually short.
Your first triac failing might have damaged some of the gate drive circuit.

As far as heatsinking, the triac should dissipate around 4W but the steel housing might not be flat or working good enough.

[tooki]

Maybe it's a counterfeit triac that is not actually the rating it says it is? Even so, 4A is below what most TO220 triacs can tolerate. Intermittent short in the wiring to the heater? Large spikes on the line? Wrong value resistor somewhere in the drive circuit? Triacs are usually very reliable.

Thanks for the ideas.

The replacement is definitely not a fake, was bought from a proper distributor.
As I said, 4A is well below the 12A rating of the triac used.
Intermittent short: possible, but doubtful, given that in operation, it’s not popping the breaker or GFCI/RCD.
Spikes: theoretically possible, but unlikely. We’d have tons of fried gear if that were the case. I did the repair on a very quiet day when little was going on.
Resistor values: possible, but I have no idea what they should be. (We don’t get to triacs in school till next year, if not the year after!)

[tooki]

You have to be methodical troubleshooting repetitive failures. Is it overcurrent, overvoltage, over temperature, bad design etc.
Guessing is OK but it's best to know what is going on.

Exactly! That’s why I decided to post, since I simply do not know enough about triacs and triac circuits to know what to look for.

At least we’ve ruled out simple random failure.

Maybe post pics of the circuit and we can try figure it out.

I’ll do that. Hopefully I’ll have time this week to do so.

A heating element is a crappy resistor - it has inductance due to the coil, and it has huge inrush current because the resistive element goes through a several 100's C rise. Not as much as an incandescent light bulb (10:1) but still several seconds of high current. You can measure the heater's cold resistance and I've seen some heating elements have shorted turns when hot due to the wire's expansion and the adjacent coils touching, or causing a ground-fault and the heater current surges high.
Inductance usually 10's-100's uH (measure it) not as much as a motor but enough to bite. For this reason I find a snubber is necessary despite the EE notion that a heating element is a pure resistor and ST calling their triac "snubberless" for the dV/dt aspect.

The original PCB (which I haven’t had a chance to see the copper side of) has a big choke and cap on it, right near the terminals for the triac and heater, so I think they equipped it with a snubber network?

Sometimes manufacturers use the same circuit for both 120V and 240V versions of their product, so the component values are wrong for the gate drive and snubber.

Makes sense. Xytronic is a midrange brand. Definitely not on the level of the real brands, but decidedly a step up from the AliExpress junk. (It’s sold by major distributors, and they’d be in hot water with the authorities if they imported noncompliant crap.) So I expect they’d do it right.

I've had many crappy chinese triacs fail, even way under current and voltage specs they get senile hog gate current and eventually short.
Your first triac failing might have damaged some of the gate drive circuit.

I doubt the original triac was fake. Again, since the heater is a name-brand product, even if decidedly not premium.
But the first one damaging the gate driver certainly sounds like a plausible mechanism. (Assuming it wasn’t spontaneous failure of the gate driver to begin with.)

As far as heatsinking, the triac should dissipate around 4W but the steel housing might not be flat or working good enough.

As I said in the original post, the triac is mounted onto a heatsink. (It’s a proper aluminum heatsink.) It’s the heatsink that’s bolted to the steel housing.

Thanks for the various ideas.

Also, I think this problem will actually be a chance to dig out the old ScopeMeter I found in storage at work, since it’s the only scope we have that’s safe to use on mains.

[james_s]

The replacement is definitely not a fake, was bought from a proper distributor.
As I said, 4A is well below the 12A rating of the triac used.
Intermittent short: possible, but doubtful, given that in operation, it’s not popping the breaker or GFCI/RCD.
Spikes: theoretically possible, but unlikely. We’d have tons of fried gear if that were the case. I did the repair on a very quiet day when little was going on.
Resistor values: possible, but I have no idea what they should be. (We don’t get to triacs in school till next year, if not the year after!)

How is it triggered? Does it use an optocoupler? If so, look up the datasheet for the opto or a similar part, it will have an example circuit which is likely pretty close to what you have in your device there.

[floobydust]

It might not be using phase control, but instead a second on, second off kind of thing.
The heating bed is separate from the controller, so lots of extra cable (inductance) so I'm wondering if a 600V blocking triac is enough with 340Vpk mains.
Pics will say lots.

[tooki]

How is it triggered? Does it use an optocoupler?

Yes, I said so early in the thread:

[…]
I’ll take a closer look next week, but the control part of the driver is a small microcontroller powered by a 7805 regulator. It hasn’t fried, so I doubt the regulator is over voltage. Some part of the circuit is probably doing zero-crossing detection, since the board has two optocouplers. (So I assume one going out to the triac and one coming in with ZCD.)

It might not be using phase control, but instead a second on, second off kind of thing.
The heating bed is separate from the controller, so lots of extra cable (inductance) so I'm wondering if a 600V blocking triac is enough with 340Vpk mains.
Pics will say lots.

It definitely does phase angle control and not stove-like on-off cycling. Well, when it’s not broken, of course! Remember, I did use it before it broke, so I know how it’s supposed to behave.

There’s no way the 2 feet of cord between the controller and device has enough inductance to matter. Lamp dimmers routinely handle cable runs tens of times longer than that.

The triac’s datasheet recommends the 600V part for standard 230V applications. And as I said above, it looks like it has a snubber network. (And the triac is a type that supposedly doesn’t even need the snubber, so it should be plenty capable.)

[bdunham7]

The original PCB (which I haven’t had a chance to see the copper side of) has a big choke and cap on it, right near the terminals for the triac and heater, so I think they equipped it with a snubber network?

Maybe have a close look at how that 'snubber' is set up (post a photo and diagram) as maybe it is actually something to reduce EMI that is ending up causing a different issue.  Is it a snubber or a resonant gate vaporizer?

[Tomorokoshi]

Can you post clear pictures of the board, with details of the parts and routing around the Triac?