stand-off voltage of zener TVS device for a relay coil

You are almost always "safe" with just a diode-drop up to about 1 volt over the rated coil voltage, for doing no harm to the rest of the driver circuitry.

I have a current (as in 'now') design using 28VDC relay coils. They are rated for a maximum voltage of 32 VDC, and the TVS snubber device is rated for 42 VDC. My relay drivers will withstand 50V at their outputs.

Find out what the Vds-max. is for your FET. 80% of that max. is a good place to start in sizing the TVS.

Good luck, Tom

Why not just an simple diode across the coil. Most applications don't care about the slowing of the turnoff due to the freewheeling currents.

It is 100V.

I guess you're saying it's good to have as high a clamping voltage as possible, as long as the driver circuitry is not endangered. You mention a place "to start"---what would make somebody change from that

You have a rather narrow range (28V to 50V) to work with, so you might benefit from a lower clamping factor. I found the following at the link provided by martin.

I read this. Sounds like crap to me. They do not quantify "optimal" life.

So if a relay is good for 100,000 closures, and spending all this money on suppression gets you up to 100,041 closures, I say: "Forget it."

As for the 9.8mS claim using just a diode, this seems excessively long to me. And in any event, is an event less than one AC cycle in duration assuming 60-Hz. Next thing you know, they'll want us to use a zero-crossing detector.....

I remain unconvinced that anything more than a simple reversed-bias diode across the relay coil is needed here.

And remember, this "study" is coming from a company that makes relays (and therefore has a vested interest in lowering warranty returns, etc...), and also makes zeners, diodes and resistors. Coincidence?? Hummm.

-mpm

I searched their website for the zener diodes and TVS devices recommended in those appnotes. I did not find any. I did find gas-discharge tubes, MOVs, capacitors, and resistors. They do not recommend capacitors and resistors for this application. They give honorable mention to MOVs, but they clearly prefer TVSs.

Probably their P&B people want the customer to use the relay products in the best way.

It sounds quite reasonable to me.

Assuming a relay doesn't fall to bits mechanically first what it is 'good' for depends on what the contacts are switching. The faster you open the contacts the faster you extinguish the arc drawn between them and the less contact material get vaporised.

It is a shame they didn't go as far as running some trials, not hard to rattle a few loaded relays for a couple of months with different coil 'suppression' circuits and then inspect the contact wear.

The other app note shows V and I versus time for a 1) 12V relay alone,

Is the extra cost that much of an issue? Digikey is selling TVS devices for $0.11 each in quantity 750. If that is $0.08 more than a diode, the difference is under 5% of the cost of the relay, probably closer to 2%.

The article smacks of a "white paper". They should have run the trials.

I am really tempted to peek at their relay offering to see how many might come with internal reversed bias-ed diodes...

Hey Matt. I thought your TVS's cost more than that. In small quantities, it's probably a wash.

Still, designing for cost efficiency is a REQUIRED skill in today's job market! When you're making a million widgets, every penny saved is worth $10,000.

-mpm

Actually, I suspect for AC loads it depends on WHEN the contacts are switching.

Which brings me to a story about a transfer switch (a big relay). I'll keep is short...

Mnf = Zenith. (I mention this because it is a real POS! - Don't ever buy one.)

This baby had the most elaborate zero crossing detector you ever saw. It must have cost a couple hundred dollars - easy.

This xfer switch was used to switch a FM transmitter from main to emergency generator power automatically. If power was lost, the transfer switch would start the generator and then switch to it. When main (utility) power returned, it would switch back.

The phase monitor would try to throw the switch at the exact instant in time when the two AC waveforms (obviously not synched to each other), crossed the zero volt reference at the same time. (This would guarantee minimum voltage on the contacts, etc... and is a great idea!)

The problem was, the switch timing was so bad (awful, actually!), that I don't know, anywhere from 10mS to maybe as high as 500mS might elapse before the contact would actually engage. (Poor mechanical switch actuation, not a command problem!). Like I said, real piece of sh^t.

And of course, you can guess what happens next: Generator output on top of cycle, utility on bottom, switch closes... main breaker pops. Damn!! Off the air. Again! (And isn't this exactly what an "emergency" generator is supposed to prevent??

Problem was we had 8 of these babies in service at the facility. (Hey, they were cheap and I didn't spec 'em!!). Our "solution" was a forced timed shutdown after utility power restored, and we ended up throwing the switches in sequence, a couple at a time. Totally not the ideal fix.

-mpm

When you're making a hundred widgets, every hour spent on engineering is worth around a dollar apiece.

Best regards, Spehro Pefhany