I'm looking to use an optocoupler+MOSFET to switch "undefined" DC power sources; the opto I'm looking at is FOD817D and the MOSFET is FDPF55N06. The power sources may range from 5V (USB power) to 24V, but I could see this being used to switch as high as 36V at some point in the future, and Digikey lists the MOSFET's Vgs(max) as ±25V... because the power source may be as low as 4.8V I didn't want to use a voltage regulator to limit voltage, so I was thinking I could clamp it with a zener; something hefty like a 1N5355B so it wouldn't clamp until 18V, and can handle up to 5W.

Is this a reasonable idea, and if so, what size (resistance and wattage) resistor should I have for current limiting that would allow for fast enough switching, but still permit enough current to run everything? If not what alternatives are there? Most regulators that can take input voltages ≥36V seem to have a dropout of at least a volt, which would mean that when running on a voltage of 4.8V (USB with some power loss) I'd end up below Vgs(th) and the MOSFET wouldn't be fully on.

 

How fast do you need to switch?
For low frequency switching a 1kΩ resistor in series with the gate, with a 10V Zener to ground should work.
The maximum power dissipated in the Zener for a 36V signal input would be 260mW.

 

Yeah this is all low frequency -- the idea is basically to interrupt power for ~2-3 seconds, so it'll be

ON...(indefinite)...OFF...(2-3 seconds)...ON...(indefinite)

Would something like this be sufficient? (or do I have something way screwy here)

 

Would something like this be sufficient?

You need a resistor from the gate to ground, to turn off the MOSFET when the opto is off, since a floating MOSFET gate has no defined voltage..

And that circuit has continuous current going through the Zener.
Better to connect the Zener to the gate so it only conducts when the opto is on (unless, of course, the opto can't tolerate 37V).

Do you really need galvanic isolation between the µC signal and the MOSFET, i.e. Is the µC common isolated from the MOSFET common?

 

You might not get over one amp though your MOSFET if you only have 4.8 volts of gate drive. If your undefined load need an amp or more and the 4.8 volts is a real expectation, you might want to change to a MOSFET that needs less gate to source voltage or even to a bipolar transistor.

 

Since this is a relatively "slow" switching arrangement,
You can use a "Photovoltaic" Isolator that actually generates its own Voltage.
There are no other components needed.
See attached PDFs.
.
.
.

 

Ultimately this device is intended to let me interrupt power (remotely) for a series of consumer-grade devices I'm testing, some of which are USB powered, some of which have their own power supplies.

The galvanic isolation is mostly a "just in case" feature, since this could theoretically have 8 different voltages from 8 different power supplies running around inside it. Theoretically all their grounds should be "0V", but since these are likely cheap Chinese-sourced wall warts I figured it was better to not find out what interesting effects I'd get tying all their grounds together.

Moving the zeners to the gate is a good plan I think, Vce on the optos is 70V so they shouldn't even break a sweat, and thanks for the reminder about biasing the gate!

One alternative idea I had is that, since the μC is 5V, I could just use an obscenely hefty 5V supply for this unit and have it be the power supply for any USB-powered devices I'm controlling. That would eliminate the need for optocouplers on that side, and I could use lower-voltage MOSFETs for that too. Then only the DC plug circuits would need the opto+zener setup.

The trick is, I'm building for expandability; right now I've only got one or two devices. Realistically I don't expect any of them to be >24V at most, but I don't want to be two years down the road and pick up a new device only to go "well crap this thing has a 30V power supply, oops".

The photovoltaic isolators sound interesting, but they're substantially more expensive than a few optocouplers and zeners ($18.08 vs $5)