I am going to be switching power to three 2.7kW heating elements (240VAC, ~11.1A each) using a microcontroller.

I'm trying to decide between mechanical relays and SSRs. The mechanical relays are much cheaper, and are tempting to use, rather than the expensive SSRs and associated heat sinks, etc.

However I'm aware that a zero-crossing SSR would lower conducted and radiated EMI. I've seen other question/answers that mention this, but they don't quantify it, hence this question.

My toaster and my countertop hotplates, with their simple bimetal bistable thermostats, pay no attention to zero crossing as they cycle. With the relays, my cycling times would be no faster than 5 seconds, I estimate, possibly much slower. This is in a residential setting, with a couple wireless devices nearby, but nothing critical in terms of EMI sensitivity.

Are there any unforeseen consequences to the mechanical relays?

Would EMI be a significant concern?

Assuming generic mechanical relays, would the SSRs offer significantly improved lifetime over the mechanical option?

  • \$\begingroup\$ What are these elements heating? You could probably get the relay cycle time down to minutes, with a corresponding improvement in lifetime, how long the cycle life lasts. \$\endgroup\$
    – Neil_UK
    Commented May 29, 2023 at 18:20
  • \$\begingroup\$ Design is always seeking for optimal ballance between price, efficiency, reliability etc. \$\endgroup\$
    – user263983
    Commented May 29, 2023 at 18:41
  • \$\begingroup\$ Relays will require a deadband, therefore the regulation will not be as good. The exact regulation will depend on your physical system, but expect a few degrees or worse. An SSR with a PID controller can regulate to less than a degree without disturbances. \$\endgroup\$
    – Mattman944
    Commented May 29, 2023 at 23:00
  • \$\begingroup\$ @Neil_UK - yeah I expect a realistic cycle time of a minute or two at the shortest, but I wanted to cover my informational bases. The heating elements are part of a sauna, so there is a large thermal mass. \$\endgroup\$
    – user272901
    Commented May 30, 2023 at 19:32

2 Answers 2


Zero-crossing SSRs typically do not have zero EMI. In fact, it's worse than a relay in some ways because there is EMI at every zero crossing when the SSR is carrying current. This is because the gate current for the thyristor (typically) must come from current passing through the load, and the thyristor does not snap on until the voltage across the thyristor allows that current to build up to a usable level. For example, an MOC3042 switching a thyristor with 50mA gate trigger current and the recommended part values (fig. 10) would not switch until the mains voltage reaches almost +/-20V.

Further, the SSR will typically dissipate about 1W/A of power (see the SSR datasheet for real numbers, but that would be 11W from this rough estimation). That requires you provide a big heatsink, all the more so if your ambient temperature could be high. Cooler is better for longer life. SSRs generally fail in the shorted condition. In case of a heater failure by short, the SSR will generally fail before an inexpensive fuse. Lifetime is less predictable- a surge that would be ignored by a relay could kill it at the next electrical storm if it's not thoroughly protected. Cycling faster (perhaps 2-3 seconds) may yield better SSR life because of thermal fatigue failure modes, however fast cycling is more likely to yield annoying flicker in lighting etc.

A mechanical relay only creates significant EMI when it switches (but that EMI is typically much more and much more variable). It is only slightly affected by heat. The lifetime in terms of switching operations is limited but fairly predictable. Mechanical relays can fail shorted but usually they fail open.

Now if your relay is good for 100,000 operations at your required current, then at 5s/cycle, that's 500,000 seconds or less than a week operating 24/7. Not very practical for an industrial situation. If you have a relay or contactor rated for 1,000,000 operations and have 30s/cycle and operate for an average of 3h/day that's more like 7 or 8 years. Typical electrical life ratings for relays and contactors switching resistive loads are from 50,000 to 1,000,000 operations.

  • \$\begingroup\$ Thanks -- so in terms of the level of concern re: EMI from mechanical relays; given that my household appliances that switch large currents with no regard to zero crossing don't cause me any issues, is it safe enough to assume that switching 33A/240V every couple of minutes is also unlikely to be a problem? \$\endgroup\$
    – user272901
    Commented May 30, 2023 at 19:46
  • 1
    \$\begingroup\$ @user272901 Yes, I would say it is unlikely. \$\endgroup\$ Commented May 30, 2023 at 19:48
  1. If you want to have low noise, then always switch at the zero point.
  2. The contact surface of the switching contact wears away when it is switched off while current is flowing across it. But even when switched on, the contact is worn out by the bouncing and the plasma formation.

You can simply use a triac and a MOC 3083M.

  • The problem here is the voltage drop over the triac.

With 1.4V voltage drop and 11.25A of current, you get ~16 Watt of heat. (2.7kW heating elements @240VAC => 11.25A)

  • If you do not want the heat, but you want the endurance of the contacts, then you can combine triac and relay. First switch the triac and then the relays to bridge the triac to switch the heating plate on. To switch it of you have to switch off the relay and then the triac. So you will never get burned contacts on the relay and you do not have to worry about the heat.

I use as relays the HF3F-L. It is a 1 channel latching relay for switching 10A. But there are other types as well available which can switch 16A.

  • \$\begingroup\$ Ah so you are wiring the relay and triac in parallel, switching the power on and off with the triac but using a relay to maintain the connection -- do I have that right? E.g. to turn on: activate the triac, engage the relay (which doesn't arc because it's in parallel), then de-activate the triac; that way the triac isn't running continually and the relay isn't arcing? And that way you get the longer life of the relay and no heat build up from the triac. \$\endgroup\$
    – user272901
    Commented May 30, 2023 at 19:41
  • 1
    \$\begingroup\$ @user272901 - Yes. I had enough problems with this contact-burning, so that I use simply a better solution. To use high voltage MOSFETs for this is to expensive. With the triac/relay combination I can make a motor softstart-function, then let it running and this with a very low possibility of damaged contacts. \$\endgroup\$
    – MikroPower
    Commented May 30, 2023 at 19:52
  • \$\begingroup\$ I've been designing a form of this idea with a zero-crossing SSR and a relay combo... do you think it's smart to have a snubber across the switched output? I struggle to see where it would be useful in this arrangement, but maybe I'm missing something. Perhaps the change in current in the SSR when the relay engages makes the snubber useful? (I'm obviously fuzzy about this, forgive me.) I guess it's smart in the event that the SSR fails. \$\endgroup\$
    – user272901
    Commented Jun 5, 2023 at 20:00
  • 1
    \$\begingroup\$ @user272901 - Your load is only a little bit inductive, so this should not be a problem. There are triacs available which are "snubberless" like the BTA20-700. If you would add a VDR, then you have to add a thermal fuse in case if it breaks. I use a BTB16-600BW since several years for a 2kW heater without a snubber or any over-voltage protection. \$\endgroup\$
    – MikroPower
    Commented Jun 5, 2023 at 23:43
  • \$\begingroup\$ Re: a zero-crossing SSR and the parallel mechanical relay: it seems like turning it off won't work reliably, because when you "activate" the SSR it won't really activate because the load side is shorted by the relay. Do you think it is smarter to use an immediate SSR instead of a zero-crossing SSR in this case? (I asked this question here: electronics.stackexchange.com/questions/670970/… ) \$\endgroup\$
    – user272901
    Commented Jun 19, 2023 at 4:20

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.