keeping thyristors off

[BobM3]

I'm trying to control a universal motor with triacs (actually "Alternistors" from Teccor). The armature is connected to 4 of them in a bridge configuration. I'm having trouble keeping the two that are suppose to be off during dynamic braking from turning on. The dynamic brake event is "dynamic". There is a sudden increase in circuit current through the conducting triacs (with corresponding sudden voltage increases across the non-conducting triacs) as well as voltage spikes from the commutation. I've tried various snubber circuits. I've tried heavily filtering across the armature (reducing the commutation voltage spikes). The problem gets worse as the triacs get hot. A scope shows that the voltage does not get near the rated blocking voltage for the triacs and the dv/dt does not seem to be any where near the limits for the triacs (or the optos). The offending triacs do manage to stay off for anywhere from 5-15 ms after the start of dynamic braking. Maybe a combination of volts and dv/dt and temperature is turning them on? I read in another thread that thyristors need time in the off state to build up to their rated blocking values. I hadn't heard of this. Are there any technical articals out there on this?

 

[mc5w]

What you need are SCRs. Does this bridge turn the alternating current into direct current? If so, triacs are not the right device. You can buy a $75 drive from Baldor that will turn 240 volts AC into the 180 volts DC for a DC motor. You can add a form C relay between the drive and the motor that cuts in a dynamic braking resistor and cuts out the drive.

My experience is that triacs are never fully off after removing the gate signal. In the case of incandescent light dimmers a small amount of leakage does not matter too much. In the case of a blinking light sign the leakage current does slightly illuminate the bulbs in the "off" state but that actually reduces thermal stress on the bulb filaments. From a reasonable distance the moving light illusion works just a well if the bulbs glow a little when "off".

 

[Skogsgurra]

I do not agree there. The glowing (yes, to make the filaments live longer) has been designed into the circuit (firing pulses never removed but more delayed). A thyristor or triac has a very low leakage current when turned off.

I think that your problem is either dv/dt firing (transients) or a rating problem. Also, keeping the gate-cathode (MT2) resistance low helps.

The turn-off time is normally in the 50 - 100 microseconds range (for power devices) but there are faster animals than that. And slower as well. But 50 - 100 is typical.

http://www.eng.abdn.ac.uk/~eng524/eg3555/stud1.html

gives a crude overview and some generic data.

Gunnar Englund

http://www.gke.org

 

[ScottyUK]

I agree with skogsgurra that dv/dt could well be the problem. Part of the problem may also be that the triac conducts twice per cycle, compared with the thyristor which conducts only once. This allows less time for the triac to recover blocking capability. Inductive loads such as the motor will worsen the situation as the available recovery time is shorter than for a resistive load. I think a thyristor pair could be a better choice than the triac for the application, unless you are really looking to save the last gram of mass and the last penny of cost. Having said that, triacs are certainly used in this application with success so it may be a case of either poor layout or some tweaks to the design being needed.





----------------------------------

If we learn from our mistakes,
I'm getting a great education!

 

[Warpspeed]

Yes I can well understand your nightmare !!

Triacs can be very difficult beasts to control, particularly in nasty inductive circuits. The problem being that they conduct in both directions, and sometimes will just refuse to turn off at the zero crossing.

As others have already said, beware of susceptibility to dv/dt switching, and make sure the gate driver has a very low shunt impedance when it is supposed to be holding the triac gate at zero volts.

I am in agreement with mc5w that for difficult loads, SCRs are far better behaved.

I have had all sorts of dramas with using Triacs to control laboratory centrifuges fitted with brush motors, and Triacs driving neon sign transformers. In both cases a redesign of the circuit using SCRs completely eliminated the smoke and flames.

 

[mc5w]

I have also built a moving light illusion controller that used opto isolated triacs that drove power triacs. The leakage of the opto triacs in the allegedly off state was enough to make the power triacs to conduct enough to make the light filaments glow a little when "off".

SCRs also have an advantage which is that you can feed a small amount of reverse bias into the gate during the off period which will help the SCR resist any voltage spikes that could cause the SCR to accidentally fire. If a spike is applied across the cathode and anode of an SCR the capacitive displacement current can act as a false gate signal. This is similar to excessive dV/dT.

 

[Mike2468]

I have used alternistors that are actually SCR pairs that share one leg so they look like a triac. So, is your Teccor a triac or is it an SCR?

The other writers have already pointed out what I think your problem is. The voltage surge is causing false activation. At warmer temperatures, the problem is accellerated.

 

[BobM3]

Teccor's Alternistors function like triacs but may very well be back to back SCRs. Their published dv/dt ratings are actually higher than their SCRs' ratings. I'm going to try SCRs. I'm thinking I'll probably need snubbers for them also. Any good suggestions for snubbers? (I've seen some suggested circuits with caps across all 3 pairs of the 3 terminals - seems excessive.)

 

[Warpspeed]

It depends on what you want to snub. In some situations the odd incoming mains spike can be allowed to cause momentary false turn on, and that in itself can save the SCRs from seeing a destructive voltage (they turns on to save themselves). As spikes can easily reach several Kv, at low source impedance, snubbing can be a mixed blessing.

Repetitive fast load generated transients can be far more troublesome and usually need to be snubbed. That can be done either across each SCR or perhaps across the load itself.

For EMI reasons placing some heavy snubbing directly across the motor may not be a bad start, and it may be all you need to do.

 

[mc5w]

A soft start that I took apart because somebody ruined a terminal by overtorquing it has SCR modules that have 2 inverse connected SCRs inside. These modules seem to have internal resistance-capacitance snubbers inside. I did not break open a module to find this out but rather it seems reasonable that manufacturers of these modules nowdays would include a snubber so that they can have reasonable reliability during the warranty period. The module manufacturer is more able to determine what resistance and capacitance values will produce an optimal snubber.