I've designed a circuit which controls 3 relays that switch inductive loads. Specifically, the loads are these solenoid valves: http://www.orbitonline.com/products.../34fpt-auto-inline-valve-without-flow-control. I couldn't find much in the way of electrical specs for these valves, so I plugged the AC/DC transformer supplying them into a Kill A Watt, and measured about 0.05 A of current draw (at 120 Vrms) for each solenoid. At the 28 VAC on which these operate, that should mean a RMS current of about 0.2 A each.

With that in mind, I selected the Spartan 9007-05-01 relay to switch the solenoids: http://media.digikey.com/pdf/Data Sheets/Coto Technology PDFS/9000_Spartan_Series.pdf

It has a rated voltage of 200V, and a rated current of 0.5 A.

After acquiring all the components for my circuit, I soldered them all on to my PCB, and tested for correct operation with about 1000 cycles of all of the relays. They worked flawlessly. However, today I connected the loads to the relays for the first time, and all three failed within the first few switches. (The switched contacts are now permanently connected.)

I suspect this is due to the contact arcing and voltage spike which accompany the magnetic field breakdown in the solenoids. I was not ignorant to this principle when I was designing the circuit, so I included a varistor (specifically the MOV-14D151K http://www.bourns.com/data/global/pdfs/MOV14D.pdf) across the switched contacts of the relay, with the intent of protecting the switched contacts of the relays from voltage spikes and arcing. The varistor is rated for a maximum continuous RMS voltage of 95 V, a maximum voltage of 165 V at 1mA, and a maximum clamping voltage of 250 V.


Can anyone explain why my relays have failed under load, and how I could have better protected the relays in my circuit?

 

I don't know if this one that you want as a solution.... since you are talking about the inductive loads then your circuit surely will suffer for the inductive kickback, Its no good if the switching is so frequent.....
There are two solutions to protect your relay from this,
first is the easy one, a diode......
second, need more component is the Snubber Circuit.

I hope this will help you..

 

You need a much larger relay. Even with snubber.

I use a 5 amp relay with snubber on 120vac sol. load and often see failures.

There was a place to ask about "other" loads in your link.

An ssr may fit in your footprint.

 

This is only my guess....
But maybe you put the inductive loads too close to the relay or the inductive load magnetic field is too strong....

At the first few switches, it's still working because your load is not fully inducted.
But after that few switches, the inductive load reach the top of its magnetic field stronger than the magnetic field in the relay, which is causing the relays in permanent state.

Just an opinion. CMIIW.

 

Swap the relays for 120v AC rated SSRs with zero cross firing.

 

Reed relays have low contact pressure and are designed for low current resistive loads. They thus have little tolerance for inductive loads that cause contact arcing so it's not surprising that the contacts immediately welded with the solenoid load. You should use a solid-state relay or a standard relay with a much higher contact rating (at least 5A), as noted in the other posts, to prevent contact failure.

And why did you use a varistor with such a high voltage rating? It should be much closer to the 40V peak of the AC supply. Two back-to-back 50V zener diodes or a 50V transient-voltage suppression diode (TVS) would probably provide a sharper clamp than a varistor.

 

You need a much larger relay. Even with snubber.

I use a 5 amp relay with snubber on 120vac sol. load and often see failures.

There was a place to ask about "other" loads in your link.

An ssr may fit in your footprint.

Something like examples on attachment below.

 

I have several of those Orbit valves in an irrigation system. They are all driven with 24Vac from various irrigation timers. Those use TriAcs to switch the valves on and off. I dont know if they use zero-crossing switching or not, but I doubt it...

 

Another thing I've done to save relay contacts. Solder a triac to the pads, using the relay to trigger it. Still have the isolation from the relay.

Might have to cut a trace or two, but saves redesign of existing.

There is a special one for inductive loads that I get from digikey.

 

Thank you for all the advice and comments so far. I will be unable to (re)work this little project for about a month, but I'll be sure to update the thread with my attempts/solution.

I think I picked that varistor to make the circuit more generally useful for the relay, rather than the particular load I had in mind. Obviously that was a mistake, because the one I chose was inappropriate.

 

I believe I used these from digikey.

http://www.littelfuse.com/data/en/data_sheets/littelfuse_thyristor_qxx15xx_qxx16xhx.pdf

http://www.digikey.com/product-detail/en/Q6016LH3/Q6016LH3-ND/409685

 

Thanks to your suggestions, I've got it working reliably under load now!

I switched to this zero-cross solid state relay (composed of triacs): http://www.digikey.com/product-detail/en/AQH2213/255-1530-5-ND/570668
and this varistor: http://www.digikey.com/product-detail/en/MOV-07D470K/MOV-07D470K-ND/2407528

I also tried the circuit without the varistor, and it did not destroy the SSR. The zero cross feature seems to do the job by itself.