Malicious high voltage protection

If you only need to detect a closed contact, you may be able to put a high impedance between the user terminal and the sensitive circuitry. Perhaps a high value, high voltage resistor (3.5kV are fairly cheap), or a Y-capacitor if you sense the continuity with AC. Obviously after the resistor (near the sensitive circuitry) you want a clamp like a TVS etc. and another small resistor before the micro pins.

As well as putting a high impedance in series, you could put the thing in a metal box, and make sure the metal terminals are close enough to the box metalwork that the air gap will break down before the series resistors do. That way, if they do damage it with high voltage, it would have to be in the manner of melting the box with an arc welder. I expect there is limited pay-off in protecting it any more than that.

When I was young, the naughty kids would get piezo gas lighters (the big wand-shaped ones you'd use for a stove or BBQ) and expose the inner electrode to make a makeshift ESD gun. They would use these to zap the coin slots on arcade machines, to try to get free games.

I had thought of transformer coupling the input (i.e., two "floating" conductors) and driving from the "inside" of the coil (sensing the reflected impedance of an open vs short). This would protect the return path, as well (adversary couldn't "bump ground"). I suppose that would still work with some large resistance in series -- depending on how finely he could resolve the excitation current.

This is "field" wiring; hard to imagine he'd be practically able to armor the entire runs. I.e., any protection would have to be at the device end of the run (assume the device itself is secured).

Yes, a friend used to call from abroad (europe?) and his phone call was limited (in duration) to how long he could keep flicking his lighter (I guess it would crash the MCU in the phone? leaving the line intact).

Such attacks can be thwarted by simply denying the attacker ANYTHING. I.e., cause the arcade game to do a hard reset. Or, the phone to drop the connection (unless the MCU tells it to keep the connection alive -- which it can't do if it is crashing!)

Bottom line, it's not an unreasonable thing for me to suggest he add to his RFQ (as it represents a potential vulnerability). Just let their proposals suggest ways of meeting that goal...

Thanks!

--don

If you are protecting with high resistance in series, no need to protect the ground with a transformer. That can also have a high resistance. But you need a separate ground for each switched input.

The other end of the resistor simply needs a zener or other voltage limiting device and you are done.

But a spark gap would be useful to protect the protection resistors. I think these can be bought as specified devices. I've seen them in monitors and small TVs.

No, it might be very useful.

Can the usual startup self test routine simply be repeated whenever the thing is idle - would that do it? Adding trs if necessary to toggle the inputs.

NT

Late-model Simpson VOMs were equipped with an overload button; if one over-stressed the box, a (presumably magnetic) protection relay tripped and a yellow reset button popped up. Nothing worked until you manually reset (pushed the yellow button down). It's hard to generalize this to a fancy interface, but for two or four wires, it'd be an option. One still needs a few milliseconds of other protection components, but this would keep those from overheating.

Sounds like it should be in the RFQ as a *requirement*.

It should not be too difficult to protect against that kind of an attack- it's likely no worse than some conditions that happen naturally in process control (lightning strikes, RFI etc.) but I would expect it to cost somewhat more in parts and engineering, and it might take a bit more expertise than average.

--sp

Yes, I have a couple of 260's with the reset button, lovely meters.

Jamie

build a test mode in and run a periodic test forcing the input open and then closed.

eg use photo transistor opto isolators so you've got three phototransitors one normally closed to pass the input signal inwards, one normally open to provide the test signal, and one feeding the signal through to the processor.

disconnect the input, check that it reads open, short the input check that it reads closed, read the input, repeat at 10Hz or faster.

how often is the customer wrong?

feels like it would add a buck or two for each input plus a couple of hours coding time.

if the adversary cuts the conductors and connects a counterfeit input device how can that be detected?

if he can't be absolutely certain the input is correctly wired there's no need to be certain that it's working correctly.

one solution is tamper resistant sensors and encrypted signalling,