16A / 240VAC Solid State Relay - Low Power Loss?? What options

[mrpackethead]

Hi, i have to switch a load that is 16A / 240VAC,   The switch needs to be controlled by a Microprocessor,  and it is possible that we could want to send it an on/off command quite regularly ( up to once per minute ).     The whole load switch needs to be enclosed in a din rail mount..

I've considered.
(1) Mechanical relay:    Pros - nearly no thermal issues.. Cons;  Limited life time, contacts will spark..
(2) triac:  Pros - very long life, Con - 16A triac will get hot, and thats going to be a big thermal issue.
(3) Solid State relay - thats really just (2) in a single box.

I've just seen some notes about using Mosfets to switch AC.. Does anyone have any idea about this? In theory, FETs could be set up to switch with quite low losses..

[tom66]

One thing I've wanted to play with before is the idea of using a bridge rectifier to switch AC loads.

Essentially the bridge rectifier rectifies the AC into DC... then you control the DC by shorting + and - using a cheap MOSFET, creating an effective short circuit (minus two diode drops) between AC ~.

Losses then are primarily in the bridge diodes, but you can get 15A ones cheap enough.

Simulation example... the ground is present to make the simulation work.

[tautech]

They used to use Alternistors in some Mig welders of the transformer type @ double the current you require.
This was a stud triac and was used with a zero crossing controller IC for variable power control.
But the control was set by the operator for a secondary output current and the Alternistor was switched on & off by the operater when welding.
The switch was LV DC & low current so should be easy to interface with a micro.

Yes it may give off some heat, it was on a good size heatsink and was fan cooled.

Now all this is IIRC, had a hunt through my stuff and all I could find was the IC: U2008B
Sorry no datasheets even, all on a long dead PC.

But it may sow a seed in another direction that may be worth looking at. 

[Psi]

What about some sort of hybrid system using a relay and a triac.

Have the triac only for making and breaking the connection. The relay then takes over and holds that state.
That way the relay contacts never have to deal with arcing from switching the load current so they'll have a good lifespan. Conversely the triac only needs to handle the load for a few hundred milliseconds so it wont get hot.

[richard.cs]

I'd be going for either two fets back to back or the solid state + relay suggestion so long as the coil power is acceptable. Bear in mind that if you allow no arcing at all you might have contact oxidation problems. What's your budget, how much dissipation do you think you can allow and how wide will your module be?

[KM4FER]

This sounds very much like my pottery kiln where a microcontroller frequently switches 240V 20A relays on/off to control the temperature rise and final temp.  I've never had a relay fail but from pottery discussion boards I understand it does happen.  For long relay life many pottery people advocate using mercury relays which apparently almost never fail but cost about 3 times as much.

A more interesting question is what happens if your switching mechanism fails?  It could go either way: open or closed.  I've seen photos of pottery kilns where the relay contacts welded themselves together and all the pots in the kiln melted and ruined the whole kiln.

earl...

[SeanB]

Get some ABB industrial contactors, with appropriate coil for your supply voltage ( you will probably be limited though, they come in 12VDC/24VAC, 24VDC/48VAC, 110VDC/220VAC and 240VDC/400VA) and a 16A contact rating, which will typically be a 4 pole configuration of 3 16A double break contacts and a 10A auxillary contact. Lifetime is around 5 million cycles at part load, 1 million at full load. DIN mountable directly, and they are standard items. Expensive, use around 10W in operation and make a clunk on switching, but they do go on forever.

I have some that are on 30 million cycles or so, and they are still working well.

[jlmoon]

This sounds very much like my pottery kiln where a microcontroller frequently switches 240V 20A relays on/off to control the temperature rise and final temp.  I've never had a relay fail but from pottery discussion boards I understand it does happen.  For long relay life many pottery people advocate using mercury relays which apparently almost never fail but cost about 3 times as much.

A more interesting question is what happens if your switching mechanism fails?  It could go either way: open or closed.  I've seen photos of pottery kilns where the relay contacts welded themselves together and all the pots in the kiln melted and ruined the whole kiln.

earl...

I have a CRESS heat treat oven the is 240V single phase and it uses a pair of Crydom d2440 SSR's

http://www.crydom.com/en/Products/Catalog/AdvancedWebPage.aspx?operation=1&intl=&SearchText=D2440&mfgname=Cryd&mfgID=1303&E=S&RunSearch=yes&SearchList=Part+Number

If your switch a light weight inductive load, these should do the trick

[mrpackethead]

Even if this idea does'tn work, i'll give you a lot of points for thinking about it.

What about some sort of hybrid system using a relay and a triac.

Have the triac only for making and breaking the connection. The relay then takes over and holds that state.
That way the relay contacts never have to deal with arcing from switching the load current so they'll have a good lifespan. Conversely the triac only needs to handle the load for a few hundred milliseconds so it wont get hot.

[tautech]

Resistive load or inductive?

[mrpackethead]

In this particular instance the load is highly resistive, you can assume PF > 0.99.

[mrpackethead]

I'd be going for either two fets back to back or the solid state + relay suggestion so long as the coil power is acceptable. Bear in mind that if you allow no arcing at all you might have contact oxidation problems.

 What's your budget  - This is a price senstivie application, so it needs to be low cost.

how much dissipation do you think you can allow

Ideally i'd want to be sub 1W, or its going to get too warm.

 and how wide will your module be?   Currently planning on a 100mm wide module, sorry not exactly sure, its not in front of me.

[tautech]

In this particular instance the load is highly resistive, you can assume PF > 0.99.

IMO the simplest way would be to use a multi-pole contactor, poles paralelled for durability and choose a coil voltage that is easy to interface to.
Possibly cascade relays an option too. Coil voltages maybe can be supplied from a Din rail LV supply?

[Alex Eisenhut]

They're not all triacs. There's IGBT and IGCT-based stuff out there.

[gxti]

What about some sort of hybrid system using a relay and a triac.

Have the triac only for making and breaking the connection. The relay then takes over and holds that state.
That way the relay contacts never have to deal with arcing from switching the load current so they'll have a good lifespan. Conversely the triac only needs to handle the load for a few hundred milliseconds so it wont get hot.

Good idea. I wonder if the oxidation problem could be mitigated by only using the TRIAC to open the circuit, and use the relay to close it. I imagine most of the wear on relay contacts is during opening while full current flowing, and not on closing when the current is just ramping up and limited by inductance.

Still, I'd guess that oxidation isn't much of an issue when switching 16 amps. I know it can be a problem when switching signals with effectively no load on them.

[mrpackethead]

turns out that the spec that i was given was wrong. the number of events that the device is expected to cope with was reduced by a factor of 100.    This makes a Relay a realistic option.

thanks for you input though, some great idea.s