Triac / Relay combo to reduce heat dissipation

[Joe Voytovich]

I've been trying to resolve an issue with a pump controller project where the customer needs a long run time (5 minutes) and this time length generates too much heat on the control triac. Currently the triac is PCB mounted; the board doesn't provide enough heat sink space, and a separate heat sink cannot be too big as space is limited.

I had this idea of using a triac / relay combination to reduce the heat in which they would be connected in parallel. The triac would turn on first, then 1 second later the relay would close. This would drop the voltage across the triac to almost nothing and the heat would go away. At turn off, the relay would open first and with the triac still on and pulling current, there should be no arcing since the triac drop is only 1 volt.

So... no relay contacts arcing and no triac heat build up.

Any real drawbacks on this? It seems like a really ingenious solution here, so I'm wondering what else I need to consider.

One thing though is the contact ratings. Since the relay will not ever switch the full line voltage (120 VAC), can I get away with using a relay with 12V contacts? I'm looking at some small ones for about $3 with 10A contacts.

Thanks for any input.

 

[FvM]

The usual solution in industrial control and home appliances is to use a contactor rated for the full motor power and tolerate contact arcing or possibly try to reduce it with passive means. A solid-state switch/contactor combination might be used in special cases with high switching frequency to save contactor lifetime. I have already heard about it but I don't remember the application.

Using a relay that isn't at least able to reliably disconnect the load in case of a controller failure doesn't sound like a good idea.

 

[andre_luis]

...I had this idea of using a triac / relay combination to reduce the heat in which they would be connected in parallel. The triac would turn on first, then 1 second later the relay would close...

Although using relay you would face to voltaic arc issues, I´m not totally confident to ensure that employing a silicon based device for starting a inductive load should be the better approach.

I had used relay for control coarse devices such as compressors, and most protective actions on design concerned to layout and circuit scope, assuring proper decoupling for induced spikes became from the load.


+++

 

[chuckey]

Just use the triac to fire up a decently rated relay, which uses its contacts to run the pump. All work external to PCB, in a little box?
Frank

 

[Joe Voytovich]

The triac would also run the pump for shorter periods alone. There are some run conditions where the pump would only run for less than 10 seconds, so the relay would not be used. The relay would only be on for long runs to alleviate the heat. I don't want the relay to be the main control device, since would have a more limited life cycle.

The board is a redesign of another board that has run for years without any problems using a triac alone. The problem now is that the customer - since we're now using a PIC as the brains - wants to be able to run for much longer times. The previous board was all discrete logic and didn't have much flexibility.

 

[chuckey]

OK then use your triac to fire up another in a properly heatsinked enclosure.
Frank

 

[Joe Voytovich]

This pump is for a toilet system, with a large shoe-box sized plastic well for the pump, and the controller sits in a partially submerged unit which contains water level sensors. We cannot mount outside the controller box, and it is sealed against water, so air flow is nil. Fun, huh?

These are consumer products that get mounted many times behind walls, etc, so maintenance is difficult. It needs to be ultra-reliable, that's why I'm hesitant to use a relay only.

Think of this as a hybrid car. It would use the triac for short runs, then kick in the relay for longer runs to save heat, then disconnect to allow the triac to bring it all to a smooth, peaceful stop. Or at least that's how I'm going to explain it to my boss when he gets back next week. :grin:

 

[andre_luis]

...the board doesn't provide enough heat sink space, and a separate heat sink cannot be too big as space is limited...

Now it is clear that you are not dealing with hard devices, as at least I presumed at begin, but compact apparatus so that in fact a TRIAC based solution seems the best approach.

Taking a review at the original stated problem, sound as the issue concerns mostly to layout and mechanical scope.

If possible, post more details about its geometry.


+++

 

[chuckey]

As your PCB is surrounded by water, use that to cool the triac. Bolt a piece of angle to the side of your box, bolt the triac to the available surface.
Frank

 

[Joe Voytovich]

Can't do that either! The board has to be inside the cover and electrically isolated. The pump has the same issue with heating - it's totally sealed in plastic and has a thermal cutoff if it runs too long. The pump is then off for 1/2 hour. We're adjusting the controller software to try to prevent that as much as possible. We use a simulated heating curve in the program that measures the AC current and increments based on that. When the heat counter gets to a certain level, the pump is shut off for a while. When the counter "decays" to a certain level, the pump can turn on again.

- - - Updated - - -

The board size is 4.2" x 3.0" (106mm x 76mm). It's got 2 PICs on board, power transformer, triac, 2 pushbuttons, 3 LEDs, a buzzer, 4 connectors and AC power cord connections. It also handles user inputs via the buttons and LEDs.

 

[andre_luis]

...When the heat counter gets to a certain level, the pump is shut off for a while. When the counter "decays" to a certain level, the pump can turn on again...

If I understood correctly, the circuit does not properly measure temperature inside case, but solely perform estimation based on previous known temperature decaying behavior, and it means that a closed-loop control is not being performed

I presume that including a thermostat control you could achieve a better performance, not heating too much or not wasting time due to some eventual oversized time predicted for cool.



+++

 

[Joe Voytovich]

After looking at some relay specs and thinking about the circuit some more, I might be leaning now to just using a relay. By adding a snubber across the contacts, I can minimize the arcing as much as possible. Also, what I was wondering is if the triac is on and the relay closes, will the triac then effectively be off until the relay opens? I'm trying to picture the circuit operation at that point: the relay begins to open, and the voltage rises and at a certain point the triac will see enough voltage to turn on. This will effectively clamp the voltage across the relay contacts and the arcing will be gone.

Also some relays have an electrical life of 100,000 cycles - for a resistive load. Drop that down to 70,000 cycles then for a motor. At 10 cycles a day, that's almost 20 years. That should work. And the relays I've seen to handle 16A are less than $2. Sounds like a good alternative to putting an aluminum battleship inside the box to try and cool a triac.

- - - Updated - - -

The heat counter is for the heat in the pump itself, which is sealed and cannot have a thermocouple added to it. We've asked numerous times. The triac is on the main circuit board. We could add a sensor on this, but it's still a band aid in that the triac will overheat way before the 5 minute run time is reached. Also, we found that if the triac gets too hot, it latches up and refuses to turn off even when the gate is shorted to MT2. It's sort of like having the gas pedal stick to the floor in your car. The pump will then run until the thermal device trips and shuts it off for 1/2 hour. Then the customer calls and wonders whats wrong with the pump. He has to be told to come back in 1/2 hour and try it again. That doesn't leave the customer with a good impression, especially since it will always happen under the same circumstances.

 

[FvM]

I agree that the overloaded triac is obviously a bad solution.

Also, what I was wondering is if the triac is on and the relay closes, will the triac then effectively be off until the relay opens? I'm trying to picture the circuit operation at that point: the relay begins to open, and the voltage rises and at a certain point the triac will see enough voltage to turn on. This will effectively clamp the voltage across the relay contacts and the arcing will be gone.

If a sufficient gate current is flowing during relay disconnection, the triac voltage won't rise above on-state voltage.

 

[betwixt]

I use a relay/triac solution here for 2HP water pumps. I actually use two relays, one in series with the incoming power (to isolate it) and one across the triac. The power on sequence is 1. close the isolating relay, 2. turn the triac on, 3. short the triac out with the other relay. The reverse happens to switch it off again. Both relays are rated at 250V/30A and each step is timed at 0.5 seconds. So far there have been no problems and by using a zero voltage crossing circuit to fire the triac it doesn't even produce the tiniest click of radio interference.

Brian.