74ac nanosecond pulse
In years past, I built a number of different high-speed pulse generators; including several avalanche-xsistor based units.
The #1 advantage of using avalanche pulsers is the ease of obtaining quite high output power.
24v is very marginal for avalanche circuits....you might try some of the recent 8ghz fT small-signal xsistors, with the -lowest- Vcc-max rating you can find; and see if you can get consistent operation under 24v.
I can pretty easily get 600ps transition times from a 2N2369A, but most of them require 50-60 volt supply for consistent operation.
But since you left out a whole lot of necessary specification information; it's real hard to give specific advice.
For example....
You didn't specify whether you need high output power, or not.
You say you need 5-10v pulse height, but you didn't specify the load resistance.
You didn't specify what repetition-rate (frequency) you need.
You didn't specify what waveform you needed; and whether or not waveshape quality was important or not.
These things are important because, for example, if you need kilowatt peak power, a typical small-signal SRD isn't going to work; whereas an avalanche will easily make that kind of high-power pulse.
But if you need 100ps edge-rates, an avalanche won't work; but an SRD will.
Also, avalanche circuits are generally hard to build for a rep-rate much above 100-200 khz. The heating in the xsistor becomes a real issue as the freq rises. Higher F ckts can be made, but they're much trickier. A 10khz avalanche, on the other hand, is very very simple to make.
Also, depending on what kind of shape you need for the low-frequency portion of the waveform, you can get your required 10v of signal -regardless- of the available power-supply voltage; simply by using a xformer at the output. This is true regardless of what device you use.
If you only need single pulses, or at most a few hundred Hz, then you can get very fast pulses with nothing but a relay and a piece of coax. For 400ps, you really don't even need the coax.....a short microstrip trace on the PCB will work for the xmission line.
To avoid contact-bounce and get a clean output pulse-train, you will have to use a mercury-wetted type of relay. Even at that, a 400ps overall-width will be very hard to achieve. Tek used to make a relay-type pulse-gen, which used their own custom relay, which was filled with high-pressure hydrogen to improve operation.
However, I've acheived decent results from regular 'commercial' type merc-wetted relays. You need to look at the physical construction and choose one which is set up closest to good RF practice...i.e. straight-thru signal-path, short connections, etc..
Bear in mind that a 400ps "pulse" implies edge-rates of 200ps, or even less; and that is a VERY high-frequency signal. No matter which "device" you design around, you will have to do a VERY VERY good job on the PHYSICAL design (layout, construction, choice of components and materials) to have any chance of achieving a usable result.
best of luck with it....high-speed pulsers are an enjoyable fun challenge to work with.
PS: the Tek 7904 scope with the right plug-ins is available pretty cheap (a couple hundred bucks) and is just barely capable of work in the 700ps and up arena.
A Tek 7104 with the right plugins is probably around 700-1000 bucks, and is just barely capable of useful work in the 400ps arena you're talking about.
Any Tek 7000 series storage scope, with the 7T11/7S11 sampling plugins, is fully capable of working down to the sub-200ps arena; and with the right front-end module for the 7S11, could probably be had for around 300-600 bucks.
If you plan on doing a lot of fast-edge/fast-pulse work, the sampling-scope setup is probably worth the investment.
You might be able to find an older sampling-type digital scope for similar money....like one of the old 54000 series HP units. Just make sure it's working (inputs not blown), because otherwise it's a useless piece of scrap....the repair parts are all custom and are big $$$$$.
ps; you may be able to get close to the pulse you need with nothing more than a logic-IC from some very fast modern family....i.e. something like 74ACxxx logic, but a bit newer and faster. As I recall, 74AC has edge rates right around 1ns or even a bit faster...so presumably the next-newer series might get into the range you need.
Differentiate the output-edge, and you might end up with a pretty short pulse....600-800ps or so...and to get the voltage you need, add a xformer at the output.