Specifically, my circuit consists of a 5,500 watt hot-water heating element on a 220 volt circuit. I am controlling the heat output using a PWM signal - the heating element is turned on/off with a solid state relay. The PWM signal runs at a 0.5 Hz frequency with a duty cycle typically ranging between 80% and 100%

I have found a relay that is rated for 240V, 25 AMP on the control side. Is it safe to use this relay in my circuit? By 'safe' I mean do I risk a fire hazard or other life-threatening condition by using this relay in this manner?

Here is a link to the Data Sheet Link.

  • 2
    \$\begingroup\$ Define "safe" in the context of 240VAC and the PWM frequency and duty cycle you intend to use. \$\endgroup\$
    – jonk
    Aug 25, 2016 at 19:04
  • \$\begingroup\$ What is your PWM signal frequency going to be, that relay is rated at "25 - 65 Hz Operating Frequency"? \$\endgroup\$
    – Tyler
    Aug 25, 2016 at 19:06
  • 1
    \$\begingroup\$ What thermal resistance heat sink did you calculate you need? \$\endgroup\$ Aug 25, 2016 at 19:10
  • \$\begingroup\$ Whoever wrote the 'Product Features' for that unit seems to have forgotten that a SSR doesn't have any contacts to arc across... \$\endgroup\$
    – brhans
    Aug 25, 2016 at 19:10
  • \$\begingroup\$ @jonk, Tyler I've added that info to my main post. Tony I have a heat sink from the previous SSR I was using, I was just going to use that same one \$\endgroup\$
    – wesanyer
    Aug 25, 2016 at 19:14

5 Answers 5


Use a couple sizes up. 40A SSRs are easily available. You must consider the 100% duty cycle case when the heater is operating for a long period of high demand.

Here is a typical derating curve for a 40A SSR based on the ambient temperature maximum and your heatsink performance.

enter image description here

You should stay well below the curve. A 1.5°C/W (natural convection) heatsink is quite large (maybe 5" square x 1.5" tall).

Do note that SSRs will typically fail 'on' and you must ensure that no safety hazards arise when (not if) this eventually happens.


Don't size to the load. Up size your SSR and leave headroom. I'd go with the 30 or 45 amp model. And those SSR's also dissipate power as heat. At higher loads you will need to mount it to a heat sink, consult the datasheet, page 5.

  • \$\begingroup\$ If I should not size to the load, then what should I size to? The load times a safety factor? \$\endgroup\$
    – wesanyer
    Aug 25, 2016 at 19:41
  • \$\begingroup\$ Go for >50% load. It might hold up at its maximum but you have to account for ambient temperature and humidity plus the fact you have to mount it to a heat sink. It might be happy to handle 25A all day at room temperature on a heat sink in a temperature controlled environment but a basement or boiler room can get warmer in the winter months. \$\endgroup\$
    – Mister Tea
    Aug 25, 2016 at 20:17
  • 3
    \$\begingroup\$ @MisterTea - I think you mean go for >150% of load. \$\endgroup\$ Aug 25, 2016 at 22:10
  • \$\begingroup\$ @WhatRoughBeast Doh! You are correct. I meant to say >=150% of the load. \$\endgroup\$
    – Mister Tea
    Aug 29, 2016 at 16:22

I am controlling the heat output using a PWM signal - the heating element is turned on/off with a solid state relay.

There are two types of SSR: "random" type which allow variation of on-time in each cycle (Figure 1) and zero-cross types (Figure 2). You have selected zero-cross type.

enter image description here

Figure 1. A "random" triggerable SSR can vary the voltage of the output.

For on-off control the triac will be switched to give a load waveform as shown in Figure 2.

enter image description here

Figure 2. On-off AC time control.

These devices use zero-cross circuits to switch the load on at zero-cross to minimise electromagnetic interference. The triac itself switches off at the end of the next half-cycle. This approach works well for loads such as heaters which respond slowly to power.

PWM usually refers to high frequency switching on and off. Power switching using an SSR would be done at a much slower speeds with on and off times of 0.1 s or longer (even up to minutes) to suit the thermal response of the heater.

Is it safe to use this relay in my circuit?

If it's rated for 25 A it should be fine on a 25 A heater load.

  • \$\begingroup\$ So are you saying that a) PWM is not suited to this use-case, or b) I am using the term 'PWM' incorrectly? \$\endgroup\$
    – wesanyer
    Aug 25, 2016 at 19:22
  • \$\begingroup\$ If by PWM you mean the control strategy of Figure 2 you are OK. If you mean high frequency switching (anything greater than mains frequency) then it won't work. You need to study what a triac is: one they turn on they don't turn off until the current falls to zero. \$\endgroup\$
    – Transistor
    Aug 25, 2016 at 19:26
  • \$\begingroup\$ @wesanyer: At your low on/off frequency (once every few seconds or so), it should be fine. I haven't bothered reading the datasheet, but in general SSRs will drop perhaps \$2V\$ or so and I'd therefore expect about \$50W\$ dissipation for continuous use (80% to 100% duty cycle qualifies as continuous, to me.) You need to be sure that your airflow and heatsink combined with the SSR can radiate that much (actually, more would be much better) on a continuous basis. Also check for any derating required! \$\endgroup\$
    – jonk
    Aug 25, 2016 at 19:30
  • \$\begingroup\$ Well, by PWM I mean that every 2 seconds, my control algorithm is either going to keep the SSR turned on for a certain fraction of that 2 seconds commemorate to the duty cycle. For 80% duty-cycle, the SSR will be on for 1.6 seconds and off for 0.4 seconds. For 50% duty cycle it will be on for 1 second off for 1 second. \$\endgroup\$
    – wesanyer
    Aug 25, 2016 at 19:31
  • \$\begingroup\$ @jonk yea that's where I went wrong with my last setup. It worked fine for a while but when I opened up my enclosure and saw scorch marks on the hot lines I got scared- I never should have left the SSR/heatsink in the enclosure. I plan to mount the heatsink outside the enclosure this time and run a fan over it when in use. \$\endgroup\$
    – wesanyer
    Aug 25, 2016 at 19:33

This is a good question - in part because the subject is more complex than it may appear. and the "obvious" answers may be less good than they seem.

Short version:

  • If all is as you say and as it seems then this device is notionally OK to use if employed correctly.

  • But there are several factors that can make this a bad choice.

  • Your selected Optocoupler is a "zero crossing triggered type so your PWM may work differently than you expect. Or may not

Depending on factors discussed below, a relaay based solution may be superior for your purpose.

Opto22, the indicated brand, are a very reputable and experienced manufacturer. Their data sheet can be relied on to be correct, with all the normal qualifications. Understanding the data sheet and designing appropriately is essential.
How hard can it be to "design" an SSR hot water control cicruit.
Actually not overly hard BUT get under the minimum limit and bad things happen.

Be CERTAIN that the part is genuine - its well known enough to be counterfeited.

The data sheet says Vout_on_drp is 1.6V worst case.
At 25A dissipation = V x I = 1.6 x 25 = 40 Watts.
Operating temperature = 40 TO 100 C Thermal resistance Rth is 1.2 C/Watt

Temperature rise = Trisemax Watts x Rth = 40 x 1.2 = 48C Say 50C.

Tmax = 100 C so heatsink max = Tabsmax - Trisemax = 100 - 50 = 50C.
Assume a max ambient temperature Tamb of 30C. That may be low in some locations.
Heat sink max rise = (Thsmax - Tambmax)/ Rpax = (50-30)/40 = 0.5 C/W.
That's either a very large unblown Al heatsink, or something cold and large creatively used as a HS ot a smaller heatsink with fan.
With fan failure = heap of slag.

The Opto22 claims <= 1.6V max drop at full load.
It probably achieves this.
Clones and fakes may not.

... I have a heat sink from the previous SSR I was using, I was just going to use that same one

Is it rated at 0.5 C/W? :-) :-( . .... !

More horsepower!:

That 'design' assumed Imax is as stated. But that's at nominal 220 VAC (presumably). At Imains = 230 V then current may be 25A x 230/220 = 26A and power in SSR up by about 10% to about 45 W.

It would not be uncommon in many cases for a 220 VAC line to be run at 230V. Or more on occasion.

As per datasheet - ALL Opto22 SSRs have internal zero crossing circuitry.



TRIAC triggering:

@Transistor gives some details on TRIAC firing.

Once triggered,
. all TRIACs
. when switching AC loads
... (except for special cases that should not apply here)
.... remain on until the end of the load half cycle in progress when the trigger voltage is removed.

So if a TRIAC is conducting and triggered or retriggered at the 0 degrees zero crossing point then if the trigger voltage is removed anywhere in the ~~= 0 to 175 degree range, will turn off at the 180 degree zero crossing.
This applies to a pure resistive load. Inductive or capacitive loads may add complications which should not matter here.

"Zero Crossing" TRIAC designs only fire on or near the load signal zero crossing and then stay on for at least the next half cycle.

Random triggerable / non-zero-crossing types turn on when triggered (more or less) and then remain on for as long as Vtrigger is maintained plus until the period to the next zero crossing point.

SO applying PWM at a rate >> s x line frequency (100 or 120 Hz for 50 & 1\60 Hz mains), will trigger on each PWM pulse but will not turn off until the next mains zero crossing. So a PWM with eg 1 kHz frame rate ans say 1% - 99% duty cycle will permanently turn a TRIAC on. ie PWM frame rate > line frequency does not work.

SUB line frequency PWM frame rate: However, if PWM frame rate is << 2 x line frequency the PWM will work 'somewhat correctly' - ie waveform may be slightly 'distorted' as turnoffs will occur at the next zero crossing and a ZC version will also turn on at zero crossings.

This means that a ZC TRIAC's on times will be quantised in 1/2f time periods and a non ZC TRIAC will have its on periods after the 1st ZC after triggering quantised in 1/2F steps but the first on period MAY be 0 to 1/2f long.

In this case:

The PWM signal runs at a 0.5 Hz frequency with a duty cycle typically ranging between 80% and 100%

I'll assume 50 Hz mains as the arithmetic is (v slightly) tidier. Half cycle period = 1/(2 x 50) = 10 mS.
So PWM frame period at 0.5 Hz is 2000 ms = 200 x mains half cycles 80% - 90% PWM will be on for 1600 to 1800 ms typically = a variation of 20 half cycles. So ~=5% per step across 80%-90% are available.
In some applications this would be unnacceptably coarse but if this is a brewing application (as a browse through profile suggests it may be :-) )(but, it may not) then the resolution is probably good enough due to thermal time constants.

Affect of PWM frame rate on heating control:

Water has a 'thermal mass' of ~= 4.2 J/cc
ie it kaes ~= 4.2 W of heating to raise 1cc of water by 1 degree C.
Beer or soup of ... will be similar enough for general sanity checking.

Rate of change of temperature of water with energy input is ~=
T change = Watts / cc / 4.2 degrees C/second
or T change ~= kW / litres / 4.2 degress C per second.

In this case at 4.2/5.5 ~= 76% PWM you'd get approx

Tchange = 4.2 kW / litres /4.2

or about 1/litres degree-C / second with no thermal losses.
ie 1 C/second with 1 litre
10 seconds per degree C with 10 litres
100 seconds per degree C with 100 litres

A PWM cycle time of 2 seconds and a 10 litre load will rise ~=0.2C/second at 4.2 kW. 100 litres = 0.02C. Given that the PWM period can be fine tuned in 5% steps, fluctuations of around 0.05C per 2 second PWM frame can probably be achieved with as little as 10 litres of "load".

Switching cycles:

At 0.5 Hz rate that's 1800/hour. A switch rated at 100,000 cycles would last ~- 55 hours. At a 500,000 cycle rating that's 277 hours.

If this is for a water heater in continuous daily use neither is acceptable.
If it's eg a beer brewer then 277 hours before replacement may be acceptable.

A relay will do this job with no heatsinking and minimal heat generation. If a relay is used it MUST be a quality part whose specs are guaranteed correct. There are many possibilities - but OMRON is a good starting point.

OPTO22 also sell SSRs with lower thermal resistances - usually due to having higher rated currents.


  • \$\begingroup\$ This could be a great answer, but it rambles on a bit and is hard to follow as well. I noticed you write 'editing' at the bottom, but that was six hours ago. Are you done editing? \$\endgroup\$
    – wesanyer
    Aug 26, 2016 at 12:37

No! Electrical codes say never design any electrical component for more than 80% of its rating. So, for a 25 amp heater, you want something > 32 amp rating for the SSR, the fuse, the copper wire, etc. AND you want a heat sink that is sized to remove the heat. For a 25 amp Opto or Crydom, that is 25 x 1.6 = 40 watts of heat to dissipate. For a Power-io SSR, that is 25 x 1.2 = 30 watts of heat to dissipate. For a Fotek that is 25 x 1.6 = 40 watts, but there are many counterfeit Foteks that do not meet any specification.


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.