Analog output overvoltage protection

[ArdWar]

Commercial calibration references are often specced to survive 1000V backfeed into its outputs, but with a caveat of several tens of mA max current. Sounds nice, that's exactly how TVS and steering diode protection work. However I don't think such strategy will survive a low impedance backfeed. At least not if the output also need to provide actual low impedance during normal operation (in addition to the low effective impedance afforded by 4-wire connection).

So, how should I protect from someone hellbent on connecting a power supply to my devices? No need for 1kV cat ratings. I **think** 50V should be enough, lab PSUs over 30V aren't that common.
ADG5401F sounds nice and it also gives me building block for 4-wire connection, but I'd like to hear other's opinions on this.

[Smokey]

What is the output impedance of your analog line?  What is it supposed to do?  Is it high bandwidth?  Etc..

[T3sl4co1l]

Back-to-back depletion MOS might be of interest.  Maybe with a sense divider afterwards, so the op-amp can correct for noise/error of the limiter device.

Bourns TBU things might also be worth a look: https://www.bourns.com/products/circuit-protection/tbu-high-speed-protectors-hsps

Tim

[Geoff-AU]

Place a sidactor hefty enough to blow up the other equipment first

TBUs can be a good choice too if the RDSon is acceptable.  The reference application in the datasheet shows what you need to do if over-voltage protection is also required. SIDAC or gas discharge tube (GDT) are good for OVP.

[ArdWar]

What is the output impedance of your analog line?  What is it supposed to do?  Is it high bandwidth?  Etc..

It's a voltage source/reference, no need for high bandwidth. Impedance might be wrong terminology choice here, since it relates more to drive headroom (e.g I can't put a 100R sense resistor if my supply is only 3V above my output and I want to source 30mA).

Back-to-back depletion MOS might be of interest.  Maybe with a sense divider afterwards, so the op-amp can correct for noise/error of the limiter device.

Bourns TBU things might also be worth a look: https://www.bourns.com/products/circuit-protection/tbu-high-speed-protectors-hsps

Tim

Duh, for some reason my mind completely blanked on the usual depletion mos/jfet limiter. Got too focused on the usual two BJT current limiting buffer often posted here
Nonlinearities should be compensated by the 4-wire, so there should be no problem there. Although gotta need to test if it does rob me some precious ppm.

That Bourns device looks nice to try. It does foldback protection!
Didn't expect such thing available from Bourns out of all manufacturers lol.

[Kleinstein]

A simple solution can be a PTC type fuse and sufficent clamps / TVS at the actual amplfier. To get a reasonable low output resistance one would likely want a separate path for the feedback, unless there are separate drive and sense terminals anyway. The feedback path can have protection with simple resistors (still with suficient voltage rating).

Back to back depleten FETs are an option, if needed for a lower maximum current with a resistor in between.
Instead of depeten FETs one could also use normal MOSFETs and a PV OK to provide a gate voltage and 2 BJTs and a resistor between the FETs to limit the current to a slight better defined value and less overall resistance.
This version can also work as a switch to turn the output off.

Most solutions would add some resistance and thus a bit extra headroom needed.