Switching a 3KW (240V) immersion heater: Solid state or mechanical relays?

[IanB]

... Is an SSR still the best option for this sort of usage?

I'm not sure an SSR was ever the best option. As Zero999 says, heat dissipation of an SSR will be significant. Go for something like the Finder 65 series mechanical relay, 30A rated with double break contacts. It is smaller, more efficient and way cheaper than the equivalent SSR.

Why not use a contactor (designed for switching high power loads) rather than a relay?

[bdunham7]

My pool pumps--1HP and 1.5HP 240VAC--are switched by relays that are roughly the size of two sugar cubes and are mounted on the circuit board.  That's a commercial product (Intermatic) and it has been working fine for a decade.  I don't think you need a contactor for a resistive 12A AC load unless you need exceptionally high reliability for safety reasons.  Your overall design should include a thermal fuse if overheating or sticking on would pose any hazard beyond burning out your heating element.  I personally would prefer a relay at those power levels to be mounted separately, like this type:

https://www.mouser.com/datasheet/2/418/8/TE_Connectivity_10112022_ENG_DS_T9A_2-3051264.pdf

A commercial design would probably just use a PCB-mount relay like this:

https://www.mouser.com/datasheet/2/307/Omron_(ENG)G5PZ_E-1843616.pdf

IMO the only justification for screwing around with an SSR is if you have a really high switching rate or a very difficult load. 

[IanB]

A commercial design would probably just use a PCB-mount relay like this:

https://www.mouser.com/datasheet/2/307/Omron_(ENG)G5PZ_E-1843616.pdf

An issue I have observed with such commercial designs is the very small contact area available on the PCB mount pins to carry up to 20 A at mains voltage (either 120 or 240 V AC) through a solder joint. Over time (perhaps due to stresses from thermal cycling) this can become a weak joint, and a weak, high resistance joint subjected to mains voltage will tend to become a hotspot, which will accelerate the weakening until the joint fails.

Good practice says that mains connections should be made with spade lugs and crimped connectors, or with screw terminals. The trend towards putting high power mains loads on miniature PCB mount components was never, in my view, a good trend, unless early failure and planned obsolescence is a consideration.

[wraper]

It depends, if you want to switch it occasionally, say once in 30 minutes, use mechanical relay. If it's supposed to frequently cycle, then use SSR.

[BeBuLamar]

Just because the outer case of the immersion heater contacts water does not mean the inside element does not run at red heat. The outer sheath generally will reach over 400C as the water boils and forms steam pockets,

If thats happening then time to change the thermostat in your immersion heater, 80c is about the maximum it should be reaching.As to the original question   properer relay/contractor for me,if the water heater is suppoised to be off i want it off not being kept tepid by the bleed current in an ssr.

A Crydom D2425 at 240VAC would have the off state leakage current of 10mA. It's not enough for you to feel the heat. Because it's 10mA at 0.2VAC (not the full 240VAC).

[Gyro]

... Is an SSR still the best option for this sort of usage?

I'm not sure an SSR was ever the best option. As Zero999 says, heat dissipation of an SSR will be significant. Go for something like the Finder 65 series mechanical relay, 30A rated with double break contacts. It is smaller, more efficient and way cheaper than the equivalent SSR.

Why not use a contactor (designed for switching high power loads) rather than a relay?

Sure, the relays I mentioned have the proper 3mm contact gap and 4kV isolation, so would do the job, but are chassis mount and spade terminal. A DIN rail mounted contactor might be be a better connection and form factor fit. I'm not sure exactly how 5U4GB intends to drive it, but both are available with 240VAC coils if needed.

[BeBuLamar]

... Is an SSR still the best option for this sort of usage?

I'm not sure an SSR was ever the best option. As Zero999 says, heat dissipation of an SSR will be significant. Go for something like the Finder 65 series mechanical relay, 30A rated with double break contacts. It is smaller, more efficient and way cheaper than the equivalent SSR.

If the switching isn't frequently then a contactor is fine. For the heater I often have the temperature controller to turn it on and off every second thus a contactor isn't suited for that.

Why not use a contactor (designed for switching high power loads) rather than a relay?

Sure, the relays I mentioned have the proper 3mm contact gap and 4kV isolation, so would do the job, but are chassis mount and spade terminal. A DIN rail mounted contactor might be be a better connection and form factor fit. I'm not sure exactly how 5U4GB intends to drive it, but both are available with 240VAC coils if needed.

[IanB]

Sure, the relays I mentioned have the proper 3mm contact gap and 4kV isolation, so would do the job, but are chassis mount and spade terminal. A DIN rail mounted contactor might be be a better connection and form factor fit. I'm not sure exactly how 5U4GB intends to drive it, but both are available with 240VAC coils if needed.

I see, so it may be a question of physical mounting arrangements? Given that the overall apparatus needs to be in a box or an enclosure with cables coming in and out, and the hardware mounted inside it.

[Gyro]

Pretty much, yes. Once you get to the point of it being a decent sized relay with reasonable (non-PCB) terminations and the same double break 3mm clearance contacts as a contactor, then it does really come down to preferred mounting and termination (screw for a contactor). The main difference with a contactor is flexibility, ability to DIN rail mount in a cabinet with wiring looms, and typically having rather more 'ways', aux contacts etc. [Edit: Coil power is correspondingly higher for a contactor that is closing more contacts of course.]

[BeBuLamar]

I use many DIN rail mounted SSR with integrated heatsink and the form factor is about the same as a contactor. I prefer SSR for heaters.

[rteodor]

Resurrecting an ancient thread that Google directed me to, since reputable-brand 25 and 40A SSRs from the likes of Digikey are now a few tens of dollars I've been looking at using one to turn power to a hot water cylinder off during peak-rate times, so it'll heat off cheap off-peak power rather than whenever the thermostat tells it to including at maximum rates, in effect emulating the old ripple control that hasn't been used for years.  This means it'll only switch on and off once a day, and be carrying 2-3kW of load for brief periods while heating.  Is an SSR still the best option for this sort of usage?

SSR's work very well for boiler control. I have a boiler with 2 heaters 800W and 1200W in which I changed the mechanical thermostat with SSR's. And they get a bit warm, maybe 40...50 degC but that is mostly because they do not make proper thermal contact with the boiler plate they are mounted on. For safety I kept the mechanical protection thermostat in place and I recommend the same to anyone regardless if they use a SSR or contactor.

Recently I used the same SSR's and relays to control a 3kW load. That load has two IEC connectors (computer mains power) in parallel probably because those connectors have a 10A rating. At full load they barely get warm. The Recom module is actually warmer that the SSR's.



Now I think maybe I jumped over the horse with two SSR's and two relays in series.

[rteodor]

Found the thermal pics too:

[Zero999]

I don't think anyone has said a solid state relay won't work, just that it will result in poorer efficiency, compared to a mechanical relay.

Example:
The current is 13A, for a 3kW load, at a mains voltage of 230V. Calculate the power loss in a solid state vs mechanical relay:

Solid state relay:
https://www.farnell.com/datasheets/3307636.pdf
On state voltage 1.25V, which doesn't change much with current.

P = VI = 1.25*13 = 16.25W
Coil power  = negligible

Mechanical relay:
https://docs.rs-online.com/8672/0900766b81587c7b.pdf
Maximum contact resistance = 50mΩ. Note this is the maximum, it's likely to be much lower.

P = I²R = 13²*0.05 = 169*0.05 = 8.45W
Coil power = 720mW

Total power  = 8.45 + 0.72 = 9.17W

In reality, the mechanical relay will waste much less power than the above calculation shows. Probably under half, as the contact resistance will be much lower than 50mΩ given on the data sheet.

[Gyro]

The Finder  65.31.xxx  https://docs.rs-online.com/ac53/0900766b816d5220.pdf

Power lost to the environment: Coil + contacts at full rated current (30A): 3.1W

[JustMeHere]

In general.   Resistive loads are easily switched by SSRs and Inductive loads are best switched by mechanical relays. 

[5U4GB]

As Zero999 says, heat dissipation of an SSR will be significant. Go for something like the Finder 65 series mechanical relay, 30A rated with double break contacts. It is smaller, more efficient and way cheaper than the equivalent SSR.

Good point, thanks!  One thing about the Finder 65's specifically (65.31, 65.61), Digikey is listing them as obsolete and unavailable while element14 gives no indication of problems, is that just a glitch with Digikey?

And a general question about SSR heat dissipation, just for my own education, the data sheets for the Panasonic SSRs I looked at as examples had fairly modest heatsinks even for 30-40A loads and had curves at 10A for no-heatsink use, do they really dissipate that much heat?

[5U4GB]

It depends, if you want to switch it occasionally, say once in 30 minutes, use mechanical relay. If it's supposed to frequently cycle, then use SSR.

For the specific use I have in mind, i.e. controlling when the HWC element runs, the switching will be very occasional, however at least for the 'on' cycle it'll probably have to carry the full load.  I haven't pulled the thermostat apart but given its age I'm assuming it's the bimetallic-disc style, which means the cylinder will mostly likely have cooled to the point where the thermostat contact is engaged so the relay contacts will carry the full load once it's energised.  For the 'off' cycle it's less likely, the heating element would have to be energised when the relay switches which is only the case a small percentage of the time.

[5U4GB]

SSR's work very well for boiler control. I have a boiler with 2 heaters 800W and 1200W in which I changed the mechanical thermostat with SSR's. And they get a bit warm, maybe 40...50 degC but that is mostly because they do not make proper thermal contact with the boiler plate they are mounted on.

There's a thought, mount the heatsink plates of the SSRs against the metal sides of the boiler so even if they do get hot they're helping the boiler along :-).

[Gyro]

As Zero999 says, heat dissipation of an SSR will be significant. Go for something like the Finder 65 series mechanical relay, 30A rated with double break contacts. It is smaller, more efficient and way cheaper than the equivalent SSR.

Good point, thanks!  One thing about the Finder 65's specifically (65.31, 65.61), Digikey is listing them as obsolete and unavailable while element14 gives no indication of problems, is that just a glitch with Digikey?

And a general question about SSR heat dissipation, just for my own education, the data sheets for the Panasonic SSRs I looked at as examples had fairly modest heatsinks even for 30-40A loads and had curves at 10A for no-heatsink use, do they really dissipate that much heat?

I don't know. RS in the UK have stock of some coil voltages, others are back order (June). Farnell (element 14), UK of course, have most voltages in stock. Presumably you only need one, maybe a spare. I don't know what coil voltage you're looking for.

Regarding SSR dissipation. Zero999 did the sums (reply #37), they look correct to me. Heatsinking requirements depend entirely on thermal resistance to ambient, ambient temperature, and maximum permissible / desirable operating temperature. In the UK, hot water tanks live in the airing cupboard (for airing -not drying- laundry). Inside temperature is significantly above the rest of the house. I don't know your situation there.

The AC1 rating of that particular relay is 7500VA. De-rating by ~50% for a heater load is my sweet spot for a long and happy life. I probably wouldn't be tempted to go for a lower contact current. The cost saving doesn't really make it worthwhile and you typically lose the desirable 3mm contact separation.

Edit: As I indicated in Reply#38, the total power loss of the finder relay is 3.1W at full 30A contact load. Given that the coil contributes 1W, running the contacts at <15A probably takes the total dissipation to <2W. Omron used to do a similar spec relay in the same for factor, I don't know if they still do.

[rteodor]

SSR's work very well for boiler control. I have a boiler with 2 heaters 800W and 1200W in which I changed the mechanical thermostat with SSR's. And they get a bit warm, maybe 40...50 degC but that is mostly because they do not make proper thermal contact with the boiler plate they are mounted on.

There's a thought, mount the heatsink plates of the SSRs against the metal sides of the boiler so even if they do get hot they're helping the boiler along :-).

Oh, but they are already mounted on the boiler .
The first thermal image is the underside of the boiler after about 30 minutes with both heaters on. The SSR's are mounted on the inside so they are not visible.

Anyway its the valve coils that switch the water circuit that, thermal wise are much more problematic than SSR's. Those suckers use about 25W for half of the day and get to about 85 degC.

[5U4GB]

Regarding SSR dissipation. Zero999 did the sums (reply #37), they look correct to me. Heatsinking requirements depend entirely on thermal resistance to ambient, ambient temperature, and maximum permissible / desirable operating temperature. In the UK, hot water tanks live in the airing cupboard (for airing -not drying- laundry). Inside temperature is significantly above the rest of the house. I don't know your situation there.

Yep, saw those, just trying to reconcile them with what the data sheets were implying based on the smallish heatsinks.

In terms of what's called a hot water cupboard here and which I realised as I typed it probably sounds odd to people from other countries, it depends on the cylinder, older/cheaper cylinders would bleed heat like crazy while newer/better ones don't, so the temperature in there will only be marginally higher than ambient.  In fact in some parts of the country, which are located about 50km from the surface of the sun with people audibly crackling and in some cases bursting into flames as they walk down the street, the hot water cupboard may be quite a bit cooler than the outside air (seriously!).

Thanks for the thoughts on the relays, will go for a DIN-rail-mount one unless the electrician says otherwise.

[jonpaul]

For safety and life, use Properly rated industrial contactor.

Most relays are not robust for tat app

The SSR are not as reliable.

j

[Zero999]

For safety and life, use Properly rated industrial contactor.

An immersion heater is not  a safety critical application. If the switch fails short then it'll get too hot and the internal thermal fuse will trip.

[Gyro]

... and a thermostat, normally set to a water temperature of around 60'C. It is normal to leave them permanently powered in some households (just the mandated manual switched isolator on the wall).

[agehall]

Related question - I want to switch a power strip in my virtual pinball cabinet using 5V so that it powers on when the computer inside it powers up. I’m expecting less than 10A constant load on the power strip.

I’ve never felt confident estimating the heat output if I was to put a decent SSR to do this switching for me. Given the discussions here, I get the sense it would be perfectly OK to use a 16-25A SSR for this without a big heat sink. I would probably just screw it to the cabinet body (made out of wood) and cover it on the top to prevent anything from falling on it by accident.

Thoughts?