many relay drivers

TVS across the coil. That sometimes helps a bit with keeping the rail clean, and (unlike the shunt rectifier approach) lets the contacts open at full speed.

Cheers

Phil Hobbs

I'd like to use a sot-23 fet and nothing else. I'll brickwall 120

7x20mm relays and have to put the drivers between the pins on the bottom side of the board and I need room for hundreds of fat traces too.

It's tough to get a flyback boost of 10:1, and I'm hoping that a relay coil can't do that. I could use a 100 or 150 volt fet and the spike voltage might not get that high.

Dropout would be fast!

So use a couple of SC-70s instead. (It needs to be a bidirectional TVS.)

Yes, the high voltage transistor can do the work as long a it can take the energy. I do that too when I need turn-off to be fast. Otherwise, I just use a cache diode across the coil.

The circulating current with a regular diode will slow the switch off but quite a lot. Sometimes it's ok but sometimes, like with line connected inverter/charger, we need it to turn off in a couple of ms rather than 10 ms or so.

boB

This is the guts of a TE PCJ, 7x20 mm SPST 5 amps.

I was a little concerned about mag field interaction so now I know where the coil is. I'll test them too.

A bunch of people make the same drop-in part, some under $1 each.

I'd prefer to use the 5x20mm parts but they are not as common.

In some cases we can measure the contact currents and open the relay if the customer pumps in too many amps. In that case, we'd prefer to open the contacts as fast as possible.

If I do add a part, it may as well be a zener to ground.

tirsdag den 25. juli 2023 kl. 03.16.00 UTC+2 skrev John Larkin:

literally what the SSM3K357R is made for as shown in the datasheet

We can probably do a footprint that allows the SSM or some other part. We're paranoid about availability and sole-source lately.

The SSM is a slightly unusual package but it looks like it will work in a standard SOT23 footprint. I'll get some and try them.

tirsdag den 25. juli 2023 kl. 04.24.47 UTC+2 skrev John Larkin:

afaict it uses the same footprint, the legs are just flat with the bottom instead of gullwing

I'd be concerned that whatever I thought the max flyback voltage could be, there will be a combination of circumstances that conspire to produce something higher on the day when that's most inconvenient.

E lets you choose the turn-off time as a trade against FET drain voltage.

Sylvia.

What I usually do if there are multiple coils is share one zener between all of the coil drivers, and have an individual diode from each drain to the shared zener.

Having the zener to the supply rail rather than to ground is slightly more energy-efficient, if the relays switch often, but requires some amount of bulk capacitance to be present on the supply rail, which there usually is anyway.

(on further thought the capacitance is only more necessary (than with the zener to ground) if there is long wiring to the power supply, with significant inductance)

I have used slew rate limited ...

a less minimal version of your "G". I also like "B" it is very cheap very reliable and can tailor release time and peak voltage just the way you like.

piglet

Yes, there could be a clamp supply that is some multiple of my +12 relay coil supply. But I'd have to distribute it to all 120 relays, which would probably add another layer to the PC board. A zener can dump to ground.

I don't think we'd switch often enough to fry a zener. Relays don't specify coil inductance, so I'll have to measure a few to estimate stored energy.

I guess some of the energy is stored mechanically, so static inductance isn't the whole story.

This is what I have so far:

Little SOT23 mosfets under each relay and an FPGA driving them. Lots of traces.

Sure, but I need to route fat traces to 120 relays on a 6-layer board. Parts get in the way.

B loses power when the relay is on, but the numbers aren't bad. The relays use 200 mW and if the resistor is 4x the coil resistance, the resistor loses 50 mW and the flyback is about 60 volts.

..

they come in difference version with different coil voltages, 5V coil is 125R, 12V coil is 720R

Here it is:

Nice little part. It zeners at 72 volts.

Thanks for the link.