Triac EMI during commutation and keeping it on to solve the problem

Haven't tried that before, no, but it just might work OK.

One of the big reasons I like triacs is that with a somewhat inductive load, usually for me it is an inductive power resistor.... is that it turns off naturally at zero crossing at least up to a few hundred Hz.

So, maybe if you add a bit of indcuctance in series, it could slow that turn on at a few volts above/below zero crossing and still turn off OK ?

Nice app notes. thanks.

boB

Zero-voltage switching is used in most SSRs, I think. It’s been around for ages.

Cheers

Phil Hobbs

I tried a simulation, and looks very good. No cross over distortion at all. Now, to build it :-)

Yes, SSRs use zero crossing detectors, but they don't seem to turn on during the zero crossing. None that I have found anyway.

Reason is obvious IMHO, the SSR is a controlled device, they don't know the future, so if the Triac is meant to be turned on for the next period. In my system I know that.

Hmm, I don’t think that’s the case. The SSR basically just resynchronizes the control signal at the next zero crossing.

Been done for ages.

Cheers

Phil Hobbs

If there is a downstream converter, you can always use an extra winding to get triac or thyristor conduction.

The usual question of 'who's on first' apply.

This is also another possible route for EMI, at the converter's frequency.

If you capacitively couple the RCdiac circuit, you can advance it's firing angle by whatever degrees the RCdiac feed permits. You can also create a low power capacitive reservoir, with the RCdiac circuit downstream. Shorting the C optically will disable it. Triacs will fire with both polarity of gate pulse, though not as reliably in both quadrants.

To ensure triacs or thyristors don't turn off to high harmonic/reversing current flow, you need to use a pulse train, anyways.

RL

??? That makes no sense. It is relatively easy to create a pattern of on and off half cycles to approximate virtually any setting of the average duty cycle. You don't need to predict the control input. The duty cycle responds to the control input.

If the gate driver is powered by the switched voltage, it needs that voltage to be up some before it can fire the gate. 5, 10 volts maybe?

The gate driver won't blow up at the AC peak because the triac shorts its power supply! Dimmers similarly protect themselves by crowbarring their own power supply.

The alternate way would use a battery or supercap and some logic to fire exactly at the zero cross.

Transformer coupling into the gate would work, but the driver would need to know what the AC line waveform was doing.

If a certain percentage of cycles on are needed, then I certainly don't want the SSR to take the decision to delay turn on to the next cycle.

The power supply demand is probably quite small, so it will be up in about a millisecond.

Sorry, I don't understand your concern. What are you talking about? Are you not talking about a dimmer? If not, then it won't delay, it will just turn on. If you are talking about something like a dimmer, why do you care which cycles are on and which are off?

it's a SSR function, to turn on a motor, so ideally it is fully on all the time, but when commanded on from the off state it needs to turn on at the zero crossing, and during each period it also needs to turn on a the zero crossing.

The problem is that the triac turns off when the holding current is below a certain value, and since the diac/triac is supplied from the mains, it turns on again when some voltage has developed across the triac. Thus crossover distortion that leads to emission problems.

Rather than use an opto triac with internal zero cross logic to drive the bigger power triac could you use a photo-mos type ssr to drive the big power triac and do the zero cross timing yourself?

piglet

So, the zero-voltage turnon is using a diac? That's usually used for phase control (turnon at arbitrary times). If you want near-zero-volts switching, the diac has to have a boost capacitor (with enough charge to reliably trigger) and gate resistor (enough resistance to stretch the start current time out), and should trigger off a shifted sinewave, not the AC directly. You want a phase-shift network to fill the boost capacitor if you want a 30V diac to breakover when MT1 vs MT2 voltage is low (3V?).

Normal emi precautions ( >20KHz ) will employ RC snubbers, or just C across the switch.

These allow the switch to turn on more reliably, as well.

RL

Crossover distortion is not EMI.

It might affect harmonic distortion, if that's the actual issue.

RL

Yes, snubbers is a given. Also avoid the triac to turn on by itself during high dV/dt on the line.

It is. Quite clear from the measurements, that a discontinuity in the current leads to conducted emissions. The discontinuity some times turns the triac on several times, since it is right on the edge of the holding current.

I did a test with an opto to keep the Triac turned on even when the current is below the holding current, and now conducted emission is 30dB lower.

So it's conducted emussions you're tracking down - not crossover distortion.

In what frequency range?

Is your triac mounted on a chassis-grounded heatsink?

Do you have a filter, or differential/common mode caps?

What is the physical load?

RL