For Triac you are switching at 100Hz, 1500 times below the conducted emission start at 150kHz,so the harmonics are way down, easy to filter (no filter)
For switching at a DC supply, you have the ESR of the electrolytic and the remaining noise needs a filter. Not a simple filter if the load is high. (probably over 500W)
.
Also,now you are above 75W, so then you need to do PFC.
The switching edges of the two solutions are not the same either. For 100Hz a slow edge is not a problem since switching losses relate to 100Hz. For a HF DC switch, you need fast edges or efficiency will be bad
Heaters have a high inertia so the speed of switching is less relevant. I do not know the wattage of these but I assume we are talking kilowatts. At such wattage, dissipation in your semiconductor elements can become an issue as well as EMI problems.
An approach i did not see come along is switching with a normal relay with a long PWM period to control the power to the heater.
In case you are concerned about EMI or lifespan of such a relay I got the following suggestion:
Place a triac in parallel with the relay contacts.
With a micro controller you fire these a couple of cycles before you switch ON and during you switch OFF the relay.
This way you achieve the following advantages:
The ON time of the triac is very small so no heatsink required.
You switch during the zero crossing so no EMI.
Re relay contacts take over which results in close to zero losses in the switching element (relay contacts).
The relay contact never see the stress of contacts closing and opening under load which means they achieve a long life span.
Such solutions are used in controllers for injection molding machines which cycle day and night heavy resistive loads.
Both FCC and CISPR 22 compliance testing are only concerned with radio interference so it makes sense that their testing methods start at 150KHz. For the purposes of power quality in an AC system we’re far more concerned with THD at frequencies much closer to the line frequency.
While the op hasn’t posted the specs of his heater elements I would not be surprised if they are a couple orders of magnitude greater that 75W. And I did mention that the use of a simple bridge rectifier might result in THD that might need to be managed, i.e… PFC.
Slow edges might be good for EMI but bad for power dissipation when switching several kW. 10KHz is not exactly HF.
The proposed solution, while not necessarily perfect, is common in the power electronics industry.
Thank you for all of the suggestions! I hadn’t considered rectifying the AC. I’ll have to do some more research into efficiency and power factor correction with that approach.
Corrected: you switch the triac during the switching off of the relay.
It is not only the heatsink cost savings. You avoid EMI suppression components which can become quiet bulky and costly at high power levels. So in terms of component count/ product size and heat generation a relay combination with a relative light triac might be a good solution.
This very low frequency type of PWM would be a very good solution if the heater had a large thermal mass, but the heater is just a halogen lamp with very low mass.