high side switching + tri-state

[phenol]

Here's the deal-im about to build a circuit that delivers a series of 200-1000ns 50-150v pulses into a mostly resistive (20-200ohm) load. After each pulse, the load will be switched into a low current source and the voltage across it measured. When i say switched, i don't mean relays and such. Because things have to happen fairly quickly and frequently, relays are a no go. This means that the switch has to be in high-z mode essentially during the measurement phase such that it doesn't divert current from the test current source/load.
I'm planning to use IR2125 as a high side mosfet driver and disable the bootstrap supply in favor of an isolated dc-dc brick delivering 12V directly to Vb-Vs. This will hopefully eliminate the bias currents that would normally flow thru the load in a standard bootstrap configuration. The fast dv/dt transients appearing on the Vs node may couple thru the isolation transformer and upset things on the "low" side...
Any ideas how to skin the cat in a different way?

[Paul Price]

Post a schematic.

Sounds like an excellent circuit for controlling current for enhancing interrogation of prisoners to achieve results while minimizing  creating tell-tale lesions and burns from the electrodes.

(Sometimes I wonder what the actual purpose of circuit designs here are for!)

[phenol]

It is supposed to be a trimming device--with each high voltage pulse the resistance of a special layer goes up or down (depending on the ingredients used to make it) a little. as soon as resistance hits a certain set point, the process stops.
the current source is 10mA, compliance range 5V. the resistor's cold end and current source are both referenced to ground, and so is the measurement circuit (a mcu). The 'hot' end of the trimmed resistor goes to the source of the high side switch AND to the current source via a reverse blocking diode.
I don't have a diagram yet, it's still only an idea, but i'm going to sketch something later on.
edit- added a rough sketch.

[phenol]

It probably will, but what are the chances of it surviving a dead short across the load (and such is not unlikely during handling)? I opted for a beefy NMOS switch and some sort of SC protection. IR2125 does have such a feature, albeit as slow as a wet wick due to the 500ns blanking time. I am also considering potential firmware foul-ups whereby the pulsing output is stuck in on state.

[phenol]

i prototyped my version with ir2125 and a small murata 5v-12v isolated converter as a high side supply. It seems to work fine, also with a shorted output. This is when diode reverse recovery charge and layout come into play...
Anyway, I've been looking around for something similar to IR2125 with an integrated short circuit protection, but to no avail. I'd like something without 'hard-coded' blanking time.