Power triac circuit latching on

[Joe Voytovich]

I've got a project where we're redesigning a pump controller board and have reached a serious road block. The triac reaches a certain temperature (it seems) and locks on. I can hit it with some freeze spray for a second or 2 and it will shut off.

I've tried all types of circuit changes - added gate pulldowns (240 ohms), different gate bias resistors (220, 1k, 5k), changed the snubber circuit. Adding a larger heat sink helped for a while, but we need a longer run period also, and the board design is locked for now.

The pump draws about 10A, and after 30 seconds switching the gate off with a manual switch doesn't turn off the triac. I removed the MOC3022 opto driver because I wanted to test with it out of the circuit, and it also made no difference.

The triac currently reaches about 60C on the case (it's a D2PAK package) in 30 seconds. I can't see that this would be enough for thermal breakdown, but apparently I'm missing something here somewhere.

Any help or ideas on this would be appreciated, since I've just about used mine all up.

Thanks.

 

[chuckey]

Is there a low impedance path for the gate leakage current?. Try putting a DVM on the gate to D1/2 and monitoring the voltage to see if it drifts upwards, causing the latch on. If the device is bolted to a decent heat sink, I would have thought its thermal time constant would have been longer, but as the case only reaches 60, it would seem to be OK.
Has the pump motor got brushes? if it has there will be a lot of switching transients which might re-trigger the triac when its hot. try a bigger (?) snubber across the motor. The transients can fire the triac by virtue of their High Dv/Dt feeding through the internal capacitance, triggering the triac, less the requirement for a low gate circuit impedance.
Frank

 

[Joe Voytovich]

Is there a low impedance path for the gate leakage current?. Try putting a DVM on the gate to D1/2 and monitoring the voltage to see if it drifts upwards, causing the latch on.

Thanks for your reply.

I did read about gate leakage that might retrigger. I had a 270 ohm from gate to cathode with no change. At one point I had shorted the gate to cathode and it still didn't turn off.

Try putting a DVM on the gate to D1/2 and monitoring the voltage to see if it drifts upwards, causing the latch on.

I did monitor the gate voltage, and it sat at about 600mV until off, when it dropped to about zero while the pump still ran.

If the device is bolted to a decent heat sink, I would have thought its thermal time constant would have been longer, but as the case only reaches 60, it would seem to be OK.

It is not on a real heat sink, other than some aluminum I've put it to cool it slightly, and I think that's the problem, but I can't verify that. I'm trying now to connect a much bigger one to check this theory.

Has the pump motor got brushes? if it has there will be a lot of switching transients which might re-trigger the triac when its hot. try a bigger (?) snubber across the motor. The transients can fire the triac by virtue of their High Dv/Dt feeding through the internal capacitance, triggering the triac, less the requirement for a low gate circuit impedance.

I don't know about brushes (probably not) - but the old board version had the "same" snubber and it worked. BTW - what would be considered "bigger"? I've read that you can increase the cap, but then the resistor value would go down. Do you mean going higher on the cap and resistor at the same time?

 

[chuckey]

No, bigger C lower R. Sounds weird.
Frank

 

[ALERTLINKS]

Check it with resistive load if problem persists.

 

[Joe Voytovich]

Check it with resistive load if problem persists.

I tried that using a power heat gun - set on low. What I didn't foresee was that on low a diode was inserted in line with the heating element. That naturally caused a break in the current every cycle, so it ALWAYS worked!

However, now the boss has said we WILL use a heat sink on the bottom of the board vs using the board copper as a sink. When I mounted the triac on a big piece of aluminum, everything seems to be good. :grin: It must be that a major hot spot on the package caused the triac to lock up. It should be good then if I can get a big enough heat sink in there.

The problem here was the customer changed the run time period from about 40 seconds to around 6 minutes. It worked great at 0:40 - not so hot (or very hot) at 6:00.

 

[ALERTLINKS]

Instead two SCRs, connected back to back is a remedy in these situations.

 

[janosandi]

how 2 use 2 SCRs ?

 

[ALERTLINKS]

Back to back SCRs as the output elements of its ac output solid state relays.
There are performance advantages associated with the back to back SCR configuration compared with triac outputs. The major benefit is dv/dt, triacs have a severe dv/dt limitation when they are turning off, the commutating dv/dt of a triac is normally in the 5 to 10 V/microsecond range. Back to back SCRs do not have this limitation, there is no commutating dv/dt associated with this arrangement, just critical dv/dt, which is greater than 500 V/microsecond. There are thermal benefits to using two output elements (back to back SCR) compared with a single element (triac) as the heatdissipated is spread over a wider area

**broken link removed**
**broken link removed**

 

[janosandi]

Thx for ur reply its very useful for me but Mr

i've used the Circuit in shown fig1 in crydom solid state relays tip
its simple & i can use a normal relay to control it
is tht enough for me or should i build the opto transistor isolating driver?

ur help will be appriciated

 

[ALERTLINKS]

Although relay can be used but it is bulky, slow and need more power. An Zero-Cross Optoisolators Triac Driver such as moc3041 fires on zero crossing and is suitable.

**broken link removed**
See fig.14, 15.

https://www.fairchildsemi.com/an/AN/AN-3008.pdf
Fig.23

 

[Joe Voytovich]

I found hybrid relays (relay and triac in parallel) are being used to cut power dissipation and to eliminate arcing. Relays do use more power for the coils, but you save 10W of heat, which would be a huge heat sink in an enclosed area.

I did find a triac that is rated for high commutation voltage and this didn't suffer the latch-up that others did. It's a lot more expensive than the original part though ($2.40 vs $0.60).

I also found that using 2 SCRs in reverse parallel work fine too - and at cheaper costs, since each go for about $0.40/each. A vendor tech sheet explained that triacs are inherently more fragile than SCRs to latch-up at higher temps. SCRs would work for me, but then I have a heat problem in a contained space and no way to cool a large heat sink.

The hybrid solution looks pretty good about now.

 

[janosandi]

wht is tht the hybrid Solution?

 

[Joe Voytovich]

Paragraph 1 - using a relay and triac in parallel.

 

[janosandi]

i've used moc3041 as shown example in datasheet but i had 2 damaged moc so i've used a simple relay instead
any idea why it got damaged?
5v with 680ohm resistor in series on pin1 of moc Gnd on pic 2
pic 4 & 6 to the SCR module

 

[Joe Voytovich]

You only mention 1 resistor in your circuit; you'd also need a resistor on the output of the MOC circuit. There should be one in the gate loop, or you'll be drawing more current through there than the MOC could hold. That is probably the reason why you're losing parts.

If that's not the case, you'll have to post a drawing of your circuit here so I can take a look at it.

 

[janosandi]

Thx Joe
actually i forgot if i've connected another resistor on the output
i'll try another one tomorrow
thx for reply