Triac protection against spurious turn on

[Klaus Kragelund]

Hi

I have a triac control circuit in which I supply gate current all the time to avoid zero crossing noise.

https://electronicsdesign.dk/tmp/TriacSolution.PNG

Apparently, sometimes the circuit spontaneously turns on the triac.
It's probable due to a transient, high dV/dt, turning on via "rate of rise of offstate voltage" limits.

The triac used is BT137S-600:

https://www.mouser.dk/datasheet/2/848/bt137s-600g-1520710.pdf

I am using a snubber to divert energy, and also have a pulldown of 1kohm to shunt energy transients that capacitively couple into the gate.

The unit is at the client, so have not measured on it yet, so trying to guess what I should try to remove the problem.

I could:

Do a more hard snubber
Reduce the shunt resistor
Get a better triac
Add an inductor in series to limit the transient

One thing I though of, since I turn it on all the time, and it is not very critical that the timing is perfect in terms of turning it on in the zero crossing, was to add a big capacitor on the gate in parallel with shunt resistor R543. That will act as low impedance for high speed transients.

Good idea, or better ideas?

Cheers

Klaus

[Wanderer]

Do what I did. Read a lot of application notes.

https://www.st.com/resource/en/application_note/cd00004096-rc-snubber-circuit-design-for-triacs-stmicroelectronics.pdf

In the STMicro app note above, they say.

"An RC snubber circuit must be used when there is a risk of TRIAC spurious triggering, i.e.
when the dI/dtOFF - dV/dtOFF couple, measured in the application, is higher than the TRIAC
datasheet values, (dI/dt)c at a given (dV/dt)c."

Which sounds simple enough, but you don't know what the specs of the killer pulse is. So increase the dv/dt spec of your triac and design your snubber network to roll off pulses above that. Then life test it preferably in an oven.

[John Larkin]

It's a sensitive-gate triac. R542 and 543 look big to me. They could
be smaller and bypassed.

[Klaus Kragelund]

I have read a lot of application notes too. They seldom list alternative ideas from just using a snubber, which I already have. So guess first off, make the snubber bad ass :-)

[Klaus Kragelund]

Yes, it's all the way down to 5mA for trigger. I agree that R543 could be smaller and that beta times R542 is not super small.

I could add a NPN for the turn-off state, so it would be 2 emitter followers (NPN/PNP) driving the gate with a order of magnitude smaller resistors.

[Fred Bloggs]

If there are motors in the vicinity, you want to at least use twisted leads in all feeds of the gate circuit.

[Fred Bloggs]

Snubbers are used to prevent what used to be called self-commutation at turn off. That's really totally different form spurious turn on, which has to do more with poor layout and construction.

[Fred Bloggs]

On Friday, September 1, 2023 at 4:53:24 PM UTC-4, John Larkin wrote:

It wouldn't hurt to shunt the gate resistor with an R + C shunt that attenuates R543 pickup by 20 dB at 100KHz and above. That would be a 100R + 0.15u. The little opto thing should be able to trigger the TRIAC in about a usec with that. This is in addition to twisting leads.

[Jan Panteltje]

On a sunny day (Sat, 2 Sep 2023 11:20:49 -0700 (PDT)) it happened Fred Bloggs
<bloggs.fred...@gmail.com> wrote in
<df747eb5-5463-437d...@googlegroups.com>:

>On Friday, September 1, 2023 at 4:53:24 PM UTC-4, John Larkin wrote=

>:
>> On Fri, 1 Sep 2023 12:56:31 -0700 (PDT), Klaus Kragelund
>> <klaus.k...@gmail.com> wrote:
>>
>> >Hi
>> >

>> >I have a triac control circuit in which I supply gate current all the ti=

>me to avoid zero crossing noise.
>> >
>> >https://electronicsdesign.dk/tmp/TriacSolution.PNG
>> >
>> >Apparently, sometimes the circuit spontaneously turns on the triac.

>> >It's probable due to a transient, high dV/dt, turning on via "rate of ri=

>se of offstate voltage" limits.
>> >
>> >The triac used is BT137S-600:
>> >
>> >https://www.mouser.dk/datasheet/2/848/bt137s-600g-1520710.pdf
>> >

>> >I am using a snubber to divert energy, and also have a pulldown of 1kohm=

> to shunt energy transients that capacitively couple into the gate.
>> >

>> >The unit is at the client, so have not measured on it yet, so trying to =

>guess what I should try to remove the problem.
>> >
>> >I could:
>> >
>> >Do a more hard snubber
>> >Reduce the shunt resistor
>> >Get a better triac
>> >Add an inductor in series to limit the transient
>> >

>> >One thing I though of, since I turn it on all the time, and it is not ve=
>ry critical that the timing is perfect in terms of turning it on in the zer=
>o crossing, was to add a big capacitor on the gate in parallel with shunt r=

>esistor R543. That will act as low impedance for high speed transients.
>> >
>> >Good idea, or better ideas?
>> >
>> >Cheers
>> >
>> >Klaus
>> It's a sensitive-gate triac. R542 and 543 look big to me. They could
>> be smaller and bypassed.
>

>If there are motors in the vicinity, you want to at least use twisted leads=

> in all feeds of the gate circuit.

One could consider using a VDR (voltage dependent resistor) parallel to the triac
to prevent spikes due to inductive switching reaching very high voltages.
Possibly add a spark gap in series:
https://panteltje.nl/pub/mains_protector_IXIMG_0501.JPG
I have such a device in the mains, it has spar gaps with VDRs in series parallel to the mains.
https://panteltje.nl/pub/mains_protector_IXIMG_0503.JPG
Works great so far...
Bought at the local market for a few Euro years ago.
VDRs are nice

[John Larkin]

On Sat, 2 Sep 2023 11:20:49 -0700 (PDT), Fred Bloggs
<bloggs.fred...@gmail.com> wrote:

I doubt that would make any difference.

[Fred Bloggs]

Twisted pairs make a HUGE difference. It's more for reducing susceptibility than emissions.

Could be electric field susceptibility, but magnetic dominates in industry with industry grade current surges and similar, and electric tends to introduce flicker type triggering, not 'spurious.'

[John Larkin]

On Sun, 3 Sep 2023 05:38:52 -0700 (PDT), Fred Bloggs
<bloggs.fred...@gmail.com> wrote:

Sometimes. Probably not here.

I wonder how far from the triac the opto is.

[Fred Bloggs]

It drops 1.3V minimum at 10A. It has an R theta-JC of about 2. If the application is high current, it needs a heat sink, so it may be off board.

I^2T is only 21, which is kind of weak.

The max rate of rise of turn off commutating is min. 10 V/us, again on the low side.

But dVd/dt is a minimum of 200 V/us with the gate open, that's to trigger a commutation from the off state, which is pretty good but not outstanding. It could be that, and if so a standard L shunt C off the line is all that's needed.

Don't know how you get sensitive gate with 30mA trigger current.

The kicker is VGT, gate trigger voltage. At 400V across main terminals, it could be as low as 0.25V @125oC, making for 0.4V at 25oC. Table 6. That kind of number indicates a vulnerability. He definitely should guard the gate drive.

https://www.mouser.dk/datasheet/2/848/bt137s-600g-1520710.pdf

[Fred Bloggs]

Snubbers designed around 'nominal' values are almost always wrong. He needs to determine worst case phase lag of load current relative to voltage. That fixes the snubber specification. Although snubbers don't have much to do with 'spurious' turn on.

[whit3rd]

It's not just the gate you have to worry about, there's dV/dt sensitivity on the main
terminals of a triac as well. Inductive pickup from the wires is a good candidate when
there is spurious triggering happening... dressing the internal wiring to the gate isn't all that should
be a concern.

[Klaus Vestergaard Kragelund]

The opto is just next to the Triac, and with a good ground plane, so no
twisting of the gate traces needed,

> It drops 1.3V minimum at 10A. It has an R theta-JC of about 2. If the application is high current, it needs a heat sink, so it may be off board.
>
> I^2T is only 21, which is kind of weak.
>
> The max rate of rise of turn off commutating is min. 10 V/us, again on the low side.
>
> But dVd/dt is a minimum of 200 V/us with the gate open, that's to trigger a commutation from the off state, which is pretty good but not outstanding. It could be that, and if so a standard L shunt C off the line is all that's needed.
>
> Don't know how you get sensitive gate with 30mA trigger current.
>
> The kicker is VGT, gate trigger voltage. At 400V across main terminals, it could be as low as 0.25V @125oC, making for 0.4V at 25oC. Table 6. That kind of number indicates a vulnerability. He definitely should guard the gate drive.
>
> https://www.mouser.dk/datasheet/2/848/bt137s-600g-1520710.pdf

The combination of the voltage rate of rise and the capacitance from
M1/M2 to the gate is what triggers it, right?

So just adding a capacitor on the gate would be a good way to protect
against noise, right?

[Klaus Vestergaard Kragelund]

Yeah, you mean the leakage from the snubber when the triac is turned off?

[John Larkin]

On Mon, 4 Sep 2023 23:59:08 +0200, Klaus Vestergaard Kragelund
<klau...@hotmail.com> wrote:

>On 03-09-2023 18:05, Fred Bloggs wrote:

Some triacs need 150 mA and 1.5 volts to trigger. Some have low ohmic
paths from gate to MT1, which helps reduce sprious triggering.

>>
>> The kicker is VGT, gate trigger voltage. At 400V across main terminals, it could be as low as 0.25V @125oC, making for 0.4V at 25oC. Table 6. That kind of number indicates a vulnerability. He definitely should guard the gate drive.
>>
>> https://www.mouser.dk/datasheet/2/848/bt137s-600g-1520710.pdf
>
>The combination of the voltage rate of rise and the capacitance from
>M1/M2 to the gate is what triggers it, right?
>
>So just adding a capacitor on the gate would be a good way to protect
>against noise, right?

I'd bypass the gate and the optocoupler receiver. Either could be
triggered by a bit of capacitively-coupled noise.

As suggested, R542 and R543 could be smaller, both bypassed by as much
c as is compatible with your speed requirements.

[Klaus Vestergaard Kragelund]

Agreed, and thanks for the suggestions. Will try it out :-)

[piglet]

Yes, I'd put a capacitor across R543. At a guess 0.1uF a good first try,
maybe even upto 0.47uF. R543 at 1k is far too high and maybe not even
necessary since a lot of triacs that size have on-die resistors gate-MT1
- you could measure one.

Is there a special reason the triac side supply is negative ground? That
triac is capable of positive gate triggering but is much less sensitive.
I almost always have the logic supply positive ground triac side so the
gate is driven from the -3.3 or -5V.

piglet

[Fred Bloggs]

The triac has its greatest susceptibility at turn off. Spec says it could be as low as 8 V/us., in an 8A circuit, Tj=95oC. If the rate of rise of voltage across part exceeds that, it may re-fire. The snubber is used to reduce that rate of raise.

A lot of SS switches have fairly large OFF state leakage, on the order of 100mA. But most of that is because of an offline ancillary power supply and not the snubber.

[Fred Bloggs]

What does that U1 on the schematic represent? A SPST with 5 ms bounce? V3 is line 225VAC 50Hz, I see that. That could be a problem. Is the spurious triggering occurring when you throw that switch?

>
> Cheers
>
> Klaus

[Klaus Vestergaard Kragelund]

The switch U1 is a simulation device to turn the external power source
on at 5ms (so peak of a 50Hz mains cycle).

That is done to see how a large dV/dt can affect the system (if that is,
the simulation model of the Triac supports M2-Gate capacitive coupling
modelling.

[Fred Bloggs]

Unless the model is specifically advertised to model that kind of thing, it probably doesn't. You can put a current probe in the gate lead to confirm.

For some representative transient waveforms see:
https://en.wikipedia.org/wiki/IEC_61000-4-5

[Klaus Kragelund]

The changes was implemented, and now the spurious triggering is gone :-)

I still need to be sure that I have sufficient robustness, so it doesn't creep up again...