5kV 500MHz scope probe

[apchar]

I'm trying to check the fidelity of a 5kV 1nsec pulse from a Blumlien pulse generator circuit. I have a scope that can sample fast enough but I don't have a 1000:1 HV probe to go with it. The closest I've seen is 6kV and 400MHz but the 6kV spec is only up to about 100MHz. It degrades from there. It'd be handy to have a divider permanently in circuit anyway.
I understand that a simple resistive divider won't cut it. A capacitor has to be added to make the dividing ratio independent of frequency & that complication puts a dip in the frequency response as the divider goes from resistive to reactive. But what if I don't need any DC response. The waveform will always be a single very narrow pulse, though the width may vary from 1-20nsec. Couldn't I just use capacitors & skip the resistors? If so, what gotchas should I look out for?
Any advice is appreciated.
Apchar

 

[canadaelk]

I needed to do something like that and built a 10:1 resistive divider. I adjusted the 10:1 probe compensation cap to get a good square wave

 

[crutschow]

Couldn't I just use capacitors & skip the resistors? If so, what gotchas should I look out for?

You can just use just capacitors if you don't need DC response. Keep the cap values small and make sure the generator can drive the capacitive load you select.

Of course the capacitor connected to the supply needs to have greater than a 5KV voltage rating.