FITS box

We'll have over 400 relays in this box. It could get hot.

We can play the trick of pulling them in at full coil voltage and then dropping to holding voltage. That will cut power dissipation by at least 4:1, likely more.

We were discussing boogering various switching regs to do that, but we don't really need a regulator... just the switching part.

This should work:

The Ixys driver claims to be very resistant to latchup, so we don't really need the catch diode.

On a sunny day (Sat, 16 Oct 2021 10:05:29 -0700) it happened snipped-for-privacy@highlandsniptechnology.com wrote in snipped-for-privacy@4ax.com:

You can do the same with 2 NPNs, 1 diode and 2 supply rails, no PWM needed. Maybe RC as delay to switch off the first NPN on the high supply rail.

That would work, maybe use a pfet. Or the Ixys and 2 diodes. It does need another main supply. Might be a big deal compared to one SO8 per board. The inductor shouldn't be bad. The 2nd supply rail switcher would need an inductor.

The uP can round-robin the four relay boards. Select one, pop it up to

24 volts, bang the relay driver shift register, wait maybe 25 milliseconds, drop the voltage, move on to the next one.

Relays have a big pull:drop voltage ratio, but it's not always well specified. 50% is probably conservative, 1/4 power.

Often the reduced operation current has limited specs, so no guarantee with vibration and temperature changes (you need to take high ambient temperature into account)

The Omron G5Q relays are unusual in that they have a specified hold voltage, 30 or 40% for parts in the family. Under 40 milliwatts hold.

Many relays specify a guaranteed dropout voltage, typically 5% of full coil voltage. That doesn't help here.

We'll have to test our selected mult-sourced relays. The PWM hold set factor could be in a cal table.

High ambient temp is avoided by cutting the coil voltages! I'd rather not have a fan in the box, but it might be necessary.

We can impose restrictions on how many relays can be energized at once. Some combinations make no sense. 150 relays on at 40 mW each is only 6 watts, compared to 400 relays at 400 mW = 160 watts. A similar product, made by someone else, gives the users full relay control with no setback, and the manual is full of big-font warnings about frying the box.

You don't see that method being used much because the relay loses all of its vibration/g resistance. Just a little bump to the chassis can cause it to drop out. You might want to test this yourself.

Surely not all, and not a little bump at, say, 1/2 coil voltage.

The reluctance of the magnetic loop goes way, way up once everything's seated. It would take a big whack to open up the gap much.