I am designing my own oscilloscope, hopefully able to do about 10 MHz in ranges of -5 to 5 volts, -50 to 50 volts and -500 to 500 volts, as well as current and ohm measurement. If someone plugged in an input of +/-50 volts while the machine was selected for the +/-5 volt range, it would (probably) destroy the op amps and other components. I could probably prevent that with a zener diode & regulator of some kind. However, how can I handle negative voltages? I have tried reversing zeners and adding other diodes, to no avail. Plus, the zener diodes would get incinerated if they try to short out too much watts, which would defeat the protection mechanism, applying full voltage to the op amps and might even destroy the ADC and whatever the scope is attached to.

A circuit which shut off the inputs until the machine was power cycled would be nice, but I can't figure out how to make one. Ideally, it should be capable of <5 ms reaction time (a fuse would not be suitable here because it wouldb be too slow, plus I'd like to avoid replacing them.)

Thanks,
Tom

 

How about back to back zeners? They won't conduct until the breakover voltage is reached, in either direction.

 

Ok, thanks, I'll have a look... Thanks for the idea.

[Edited; I reread your post]

 

Realize that he does really means "back to back", meaning in series.

<Soapbox>
Lots of times people say "back to back" to mean in parallel, like you would use regular diodes to do the same sort of thing. But parallel is actually more like "side to side".
</Soapbox>

Another thing to consider is using parallel/side-to-side signal diodes at a point where a normal signal would not cause the signal to go above a few tenths of a volt, like at the virtual ground input of an inverting op-amp. In normal operation, the voltage at the "-" input to the amp would be held at zero volts by the feedback path. If the op-amp was not able to balance this signal because the front-panel input is too high, the "-" input would be clamped within the +/- voltage drop of the diodes, with the current being limited by the input resistor. Not enough voltage to damage the op-amp.

 

Diodes ( signal ) to the supply rails will limit the voltage to the rails. Put in a high impedance or current limited part of the circuit. Look at the cap. of the diode and the leakage to make sure they do not mess up the circuit when not conducting.

 

A problem with using zeners or diodes clamping to the voltage rails, is that they are placed across an active signal. The junction capacitance, in conjunction with the current limiting resistor, will create a low pass filter. This circuit is looking for a 10MHz signal. This low pass could easily have an effect on the signal depending on the value of the resistance and the diode used. What is the corner frequency? It could be calculated once the components are known, but I think it's easier to just design it where this isn't a consideration.

Placing diodes at a virtual ground (the minus input of an inverting op-amp) is at a place where the signal is not varying (or at least not very much). I'd have to check some references, but I'm pretty sure the effect of the capacitance at this point is the actual capacitance divided by the open-loop gain of the op-amp (usually up around 10^6 or higher). This would greatly reduce the effect of any junction capacitance.

Also, any contribution by the forward conduction or reverse leakage is insignificant at zero volts. Even well below 0.7V, a diode can have significant conduction. Low current, signal-diode clamping at a virtual ground point is just a good way to go to get rid of a lot of this type of problem.

A similar way, that might even work better, is to include zeners in the feedback path of the op amp. This would provide a gain break-point at the voltage of the zeners.