Optocoupler safety

The spiral track will open like a fuse.

..... Phil

Consider the 180R resistor to be the fuse. The optocoupler failure mode to overcurrent in this range of power levels is a low power dissipation short circuit.

RL

Use a fusible resistor.

--sp

The MOC gets blown right off of its pins leaving a 150kV isolation barrier on the board.

Internal failure of the triac will almost certainly be to a short-circuit condition. Is there a hazard if it won't turn off?

Jon

She's worried about a triac failure causing a breach of the isolation barrier causing an unsafe condition. It's not an unreasonable concern.

I don't happen to share your optimism that there will never ever be a bad solder joint or a broken vibrated off lead on a triac.

--sp

Well, I guess one could check some samples of the opto to see if they actually would fail in that manner.

Yes, I got that part. If you really want to figure out how to protect against every possible fault, you ought to check with the Jet Propulsion Lab. They seem to be pretty good at it. I tend to agree with a few other posters that most likely one of the resistors will pop first. Perhaps choosing a fusible resistor or one that you have proven yourself will fail in a safe manner would be a good plan.

Jon

The external components are supposed to be chosen to limit dissipation in the opto-coupler package, _and_ be of sufficient wattage to prevent gross failure... be an engineer already !! ...Jim Thompson

If you are concerned with a failure in the opto breaching the gap, why does that require a problem elsewhere in the circuit? Can't the opto fail on its own?

I don't know the details of every opto, but I can't imagine a failure mode that would breech the isolation barrier. Isn't the isolation barrier solid material in these things? Or is it just an air gap inside the plastic? I suppose an air gap could be compromised by stray material if the chip actually explodes.

I did find this paper

which suggests that my concern is not entirely unfounded, although the paper indicates that TI's own products are pretty robust (of course, it's TI saying that).

My idea of using a non-resettable fuse won't help - they don't blow anything like fast enough. Ditto a fusible resistor.

In the event, on further study, I've learnt that using a triac to control a load whose character is not know is problematic - one needs to go to a circuit that monitors the current and tries to trigger the triac repeatedly while the triac is not conducting. Either that, or a circuit that just applies the gate current continuously while the triac is meant to be conducting. That then requires some way of supplying that much current continuously.

So I've decided to use a relay, and be done with it.

Sylvia.

Also, the LED will have a fine wire going to the anode of the chip from one leg of the package and similarly for the opto-triac. The wire will have a low fusing current - circa 1 amp.

In practice, the possibility of an isolation barrier breach is FAR more likely on the PCB due to contamination with carbon particles or a liquid - like electrolyte from a crook electro.

.... Phil

The diagram came from

The resistive load version, figure 5, doesn't seem to have been engineered that way. Disconnect the external triac from the load and 180 ohm resistor, and I'd expect something to melt rather quickly. If that thing is the internal triac, it wouldn't surprise me if the isolation barrier were compromised.

Maybe one isn't supposed to assume that TI knows what they're doing.

Sylvia.

When the LED biased on, the op-triac conducts and triggers the main triac into conduction soon as there is enough voltage across it to provide the required gate current.

If the main triac commutates off for any reason, the voltage across it rises instantly and so will result in re-triggering of the triac, long as the LED is still on.

Unlike 100/120Hz pulse firing, this scheme works with any load.

.... Phil

.... Phil

This seems a nasty failure mode in another respect, when driving an inductive load, since the result can easily be a voltage spike well in excess of the optocoupler's isolation level.

Sylvia.

Then you already have an inductive load and switch in series, so have already taken the necessary precautions to limit voltage spikes.

Your original concerns seems questionable too, as LEDs survive large overcurrent IRL. What overcurrent does is shorten their life. Other bits of circuitry will be dead long before the LED.

NT

The triac circuit limits voltage spikes by turning off at the zero current crossing. The snubber only needs to deal with the residual spike cause by the turn off not being exactly at zero. It will be inadequate to handle the much larger spike that occurs if the failure happens at the point of highest current.

The scenario I described has the high voltage circuitry on the detector side, which is not an LED.

Sylvia.

I've designed many a chip and then have the applications department write up a totally crock-of-you-know-what datasheet.

Note the fine print disclaimers >:-} ...Jim Thompson

When I was a stripling, I learned a lot of electronics by assuming that the app note circuits were garbage and figuring out just what flavour of garbage they were. I usually found stuff, and it's a fun intellectual exercise.

You learn the good app note authors that way, for sure. Pease, Yamatake, Dobkin, Williams, Widlar, maybe Jung. Any others?

Cheers

Phil Hobbs