I’m interested in using phase angle control to modulate the power output of AC resistive heaters. I have looked into various circuits for switching AC (triacs and SCRs), but am mostly curious about differences between leading edge and trailing edge phase angle control, specifically EMI implications. I have read that leading edge phase angle control results in significant EMI when switching on part of the way through an AC cycle. My questions:
Does trailing edge phase angle control also result in significant EMI? Why or why not?
What are the pros and cons of leading vs trailing edge control for resistive loads?
Is there a go to circuit for controlling AC resistive loads (while not producing excessive EMI)?
I would appreciate any advice or resource recommendations. Thanks!
Any time you introduce a sharp discontinuity you’re going to generate harmonics and EMI. An alternative is to deliver an integer number of whole cycles followed by an off period. Another alternative is to do phase angle control with a MOSFET instead of a triac, because you can slow down the edges a bit.
@JuliaTruchsess has some very good suggestions. Most AC resistive heaters have enough mass that their temperature can be controlled very easily with integer multiples of full-cycle ON periods separated by integer multiples of full-cycle OFF periods. It’s really just like PWM, but slower than usual. The key is to switch the TRIAC both ON and OFF at the zero crossings.
Julia’s other suggestion to use MOSFETs instead of TRIACs to allow slower ON-OFF transitions is quite doable, but requires additional circuitry. Zero-crossing TRIAC controllers are commonplace and simple to use, so that’s the route I’d take myself.
I’ll take a firmer position: Switch on zero crossings and use whole cycles on of off. Whole cycles because parts of cycles contribute to undesirable power factors. This another reason (in addition to EMI) for using whole cycles. If you look at the regs that apply to your design, you’ll almost certainly have a power factor limit. As Jverive observes, the heaters have so much thermal mass that zero crossing control is every bit as effective and controllers are readily available.
I would not consider phase angle modulation, TRIAC or MOSFET for any but the lowest power applications, and then I’d be running them off DC.
Leading edge has higher switching loss, but since it turns off at almost zero current the ringing with mains inductance is lower and produces less HF EMI
I should have been more specific. I’d like to drive quartz tube IR heaters which are essentially incandescent lamps, as I understand it. It sounds like the best way to avoid EMI is to only switch at zero crossings. A few issues introduced by only switching at zero crossings are low output resolution and visual flicker.
Assuming I have to make a tradeoff between EMI, control resolution, and visual flicker, what are the best ways to address power factor and current spikes? Can I just add a smoothing capacitor/inductor or other filtering?
Thanks for all the help!
Can you elaborate on how MOSFETs would slow down the edges? Do they just ramp slower? I’m considering something like this MOSFET driver.
Thanks, Julia!
That particular circuit is just a couple of FETs that turn on/off with opposite polarity inputs. This type of circuit is very commonly paired up with a PWM controller, but the FETs aren’t really there to control rise/fall times of the load current. The R2-C3 snubber may reduce EMI for non-resistive loads, but they aren’t likely to be all that helpful.
Designing switching circuits to pass EMI testing is not easy, and if you’re looking for a quick and easy fix it will be best to hire someone who understands EMI handling in power control circuits. You might be able to get away with a simple “soft-start” circuit based on MOSFETs or other switching devices, but even these require a fair amount of knowledge of the EMI limits you’re trying to meet - that is, if you don’t want to have to just try different circuits and/or component values to find what works.
Actually no, that circuit has its MOSFETs configured as an “analog switch”, and they both turn on with the same polarity. The switching can be easily slowed with some extra gate capacitance, but very close attention would need to be paid to power dissipation, which would increase with slower rise/fall time.
That’s correct. Which reminds me that there’s another kind of flicker you need to pay attention to, which is causing voltage dips, or “flicker” on the line when you switch a heavy load. This was the very first type of EMC regulation ever legislated, back in ancient days.
You’re right; they fooled me by being drawn in different orientations on the page. They are not both ON at the same time, and the circuit is very similar to circuits that use back-to-back SCRs on alternate half-cycles to act much like a DIY TRIAC.
Any capacitance added to gates will slow the devices down a bit, but too much delay will increase EMI by causing turn-off to occur past the zero-crossing.
While the discussion so far has been concerned with MOSFETs vs TRIACS etc. I would propose a different approach entirely. Instead of chopping the mains voltage I would simply rectify the mains voltage and then use PWM to chop the DC at a much higher and more easy filtered frequency. This approach, being only slightly more complicated, would give you all the resolution you wanted. You could even use a synchronized rectifier to reduce any harmonic distortions on the mains supply. For the output stage you could use MOSFETs or even IGBTs.
I had thought about suggesting rectification but wouldn’t it be as noisy on the PWM as sharply switched AC? At least there’s frequency control over your rectified PWM, I suppose. Any other pros / cons to switching rectified mains in EMI terms?
Switching the AC mains is done at the line frequency, especially when using TRIACs. Using PWM to switch DC can be done at 10-20KHz easily, much easier to filter. Depending on the current, the PWM is unlikely to get back on the mains with a sufficiently sized capacitor on the DC bus. There will be some harmonic distortion on the mains when using just a plain diode bridge rectifier, again depending on the current. This is how most servo drives and inverters work, although this application should be much simpler as it is driving a resistive load with no back EMF to deal with. There would certainly be no “flicker” and the system could be designed with as much resolution as required.
Using PWM to switch DC can indeed be a good solution if, by changing the switching frequency, the electrical noise it generates is easier to squelch than the phase-modulated AC switching with a TRIAC. That said, selecting capacitors, inductors, and other components to reduce EMI (both radiated and conducted) is very often quite difficult. In short, even seemingly simple designs can be rife with complications for the designer.
But PWMing rectified DC doesn’t necessarily mean blunter edges than PWMing AC? The benefits are basically the relative isolation of the noise from the AC side in the rectified approach?
Yes, the topic and the initial question are focused on phase angle control of AC and a resistive load, specifically with TRIACs. However, after the op’s second post this seemed more like a description of the load and his current approach and less of a constraint. So I chose to go “outside the box” and address the main concerns he raised in that second post.
No, it does not, in fact it will generate EMI. But that EMI is going to be much easier to mitigate than polluting the AC mains. As mentioned, depending on the power requirements, the rectifier will introduce harmonic distortion on the AC mains which is significant problem to be reckoned with but still manageable.
Obviously switching a TRIAC on mid-cycle is not a good approach from an EMI perspective. A “cleaner” approach using TRIACs would be to switch at zero-crossings and control the number of half cycles during which the load is on during the control window. For example, using a window of 1 second would give 100/120 half cycles giving a resolution of approx. 1%. However, the load would be switched on/off at a rate of 1 Hz and, depending on the power required, could result in significant flickering of other lighting etc. Reducing the window size to 10/12 half cycles reduces the control resolution and increases the rate of flicker to 10 Hz.
The proposed solution eliminates all of these issues and results in much more manageable EMI. The EMI is constrained to the local system, lines to the load could be kept short and shielded if necessary. The switching circuitry could be contained within a faraday cage. The higher frequency of the PWM is more easily filtered.