Peanut gallery:
- ESD is mainly RF, and pretty big at that.
- Fight back in kind.
- Instead of using a, say, 1V TDR pulse and sampler, do it at 100V, or
10kV. Now your hardware is beefy enough to withstand such treatment, and doesn't care.
- Bonus: you can get a pretty good step generator just by sparking through air. (You'd probably use something more controlled, like a reed or vacuum relay.) ESD itself is standard around 1-5ns rise (IEC 61000-4-2).
- Plus, you can use the generator head for ESD testing and physics experiments!
There's also a lot of wideband energy, so you can cut down on the energy seen by your critical devices by reflecting or absorbing the rest. TDR could be done with bandpassed wavelets, for example. In which case you're just doing old fashioned RADAR, with everything but the rotating antenna.
Of course, your time resolution always follows the total bandwidth, whether it's baseband or not, so this isn't really desirable. Perhaps instead of DC-1GHz bandwidth, you'd go with bursts in the 4-6GHz range instead. Which avoids a lot of ESD energy thanks to filtering, but now you're dependent on quality cable too, which might hardly succeed at all -- like with a modest length of RG-58, or give or take whatever kind of nonsense the attached antenna might reflect at those frequencies.
Continuing this aside... how would you build a 10kV sampler, anyway? Tempting to use the input signal (i.e., assuming the return event is still ~kV), plus a triggered spark gap (e.g., UV radiation from another, carefully timed spark) to measure it. But sparks are pretty bad at timing and breakdown to begin with. Perhaps that's the kind of domain you'd harness electron physics for your advantage? Perhaps use the input signal as a kicker for bunches in a cyclotron, so you get the synchronous detection needed, and for the anode, a series of small plates, or perhaps an interleaved binary sequence or delay line, giving the electron equivalent of a smear camera for the readout.
If you've got a precision timed pulse generator, you can certainly still use the diode gate method. Tube diodes of ~1pF and 10kV are plentiful, unfortunately I wouldn't expect their bandwidth to extend much beyond
500MHz or so.
Tim