Kiloampere pulse with sub nanosecond rise time

[belax]

Hi.

I am stuck.

For magnetic force experiments I need to develop a circuit with parameters listed below:

pulse width: few nanoseconds (1-5 ns).
pulse rise time (very important here): < 1ns
pulse fall time: not important, can be more than 10ns
pulse current: >1 kiloamps.
repetitions rate (important): > 1 mhz.
load: strait copper wire 2.5 sq.mm with length 300-400 mm.
jitter: not important.

I've attached a scratch of what i want.


In other words I need to put a high current pulse with picosecond rise time into a straight wire.


I have looked through lots of articles and forums, but still have no idea.
1) Stack of Avalanche bipolar transistors - rise time is good, but current 10's of amps.
2) SRD. Looks like it will work, but can not find simple schematic and what is more important SRD diode itself.
3) PCSS. Semiconductor photoconductive switch. GaAs. Not available on the market and lasts for few billion shots.

Need help.

Thank you.

 

[FvM]

I fear you didn't yet calculate load impedance (e.g. analyzed as transmission line). kA refers to 100 kV, unfortunately. That's the point where the design idea the latest becomes infeasible.

 

[betwixt]

Its a tough one!

The only thing I can think of is a gas discharge tube, somethig like a giant neon tube breaking down or triggered at 100KV but don't ask me to design one! With 1,000 seconds between pulses it might be possible to charge a huge capacitor then dump the charge into the load. Your written spec doesn't match the diagram which states >1 MHz rather than >1mHz which would make it even harder, if not imposssible to do.

Brian.

 

[FvM]

I also read MHz (mHz won't need to be mentioned at all), effectively excluding all kinds of gas discharge devices.

 

[belax]

MHz - megahertz
mHz - milliherts.
That was my mistake. Of course, I need a repetition rate of >1 mega hertz.

Ok. But I am not talking about 50 ohm impedance.
Of course there will be a back route for current, so the load is U type wire.
I think I need to calc an impedance of a free copper wire with ground far away (1-2 inch away from each other). Something tells me it will be less than 1 ohm, but give me some time.

- - - Updated - - -

I have made calculations. The results looks like a nightmare.

The inductance L of a 600mm length wire (and 2mm in diameter) will be around 700 nH.
The imedance |Z|~=|X_L|=2*pi*f*L = 2*3.14*1GHz*600nH=4.2k ohm
So, we have the voltage amplitude of around 4.2 megavolt for 1 kiloamp current.

If we calculate using formula U=L*(di/dt), di=1000 amp, dt = 250 ps, we get 2.8 mega volt.

Are there a way to achieve a kilo-ampere sub-nanosecond pulses on a straight wire with a reasonable length (at least 30 mm)?

Thank you.

 

[FvM]

A twin lead with wire diameter d=0.1" and separation D=1" has about 350 ohms transmission line impedance. https://en.wikipedia.org/wiki/Twin-lead#Characteristic_impedance That's what the signal generator "sees" as load for the first few nanoseconds. A finite element simulation of the exact geometry can you give more information, but it sems impossible to drive kA with < 1ns rise time into the load with less than hundreds of kV generator voltage.

I'm not sure what's the purpose of the setup and if you have any options to implement a lower load impedance?

 

[Warpspeed]

Even the most minute amount of inductance will stop you.
I cannot see that this is a feasible project.

 

[Easy peasy]

the repetition rate of 1000,000 Hz makes this project a very high power proposition, as in order to overcome the wiring inductance, say 1kA in 1nS and 50nH total o/p loop, (V/L = di/dt) gives you 50kV required as the driving voltage, as a minimum, 50kV @ 1kA is 50MW, for 3nS say at a 1 MHz rate is 150kW unless you can recover the energy in the o/p "inductor" which adds complexity... so have you done anything similar before...?

- - - Updated - - -

on the plus side you only need 200pF @ 50kV to support the 1kA for 1nS, with 5kV droop on the volt source...

 

[Dan Mills]

Hell of a project right enough, I hope you have military project scale budgets.

1 GHz? How wideband can you get a really butch TWT or Gyrotron or similar?

I have heard tell of TWT amplifiers working at these voltages and current levels, tend to be in concrete bunkers for the XRays and were more of a thing back when looking for the bear coming over the pole was a major concern.

I would be thinking wide copper traces over a ground plane rather then wires in air, probably with a teflon dielectric, you might be able to get the characteristic impedance down to a few ohms, at which point some of the modern gallium nitride power devices and a matching transformer becomes at least worthy of consideration.

I would start by going back and considering what problem you are really trying to solve, odds are there is a more reasonable way to do it.

A microsecond rise time and a PRF of a few hundred Hz is standard sort of radar technology, a PFN with some pulse compression using a saturatable reactor will get you that.

Regards, Dan.

 

[belax]

Ok. I understand now.
My main initial fault was in misunderstanding of straight wire impedance.
I thought it was less than 1 ohm, and I need less than 1 kv of voltage amplitude to achieve a kiloampere pulse.
I will dig further in solving my task.

Will be back.

Thank you all.

 

[dick_freebird]

1kA/1nS is 1GA/s dI/dt. Pushing that into a 1nH (~1mm of wire)
inductance would need a kilovolt behind an infintely fast switch.
And you want a foot of wire, not a millimeter. Not to mention
that any force must involve more inductance than a bare wire
(no second body, no force).

I would recommend one of the faster HV eGaN devices, run from
a negative supply via pulse transformer (common-source switch).

 

[Easy peasy]

Sadly, without significant resources and technical engineering know how, this project is well beyond the means of a junior engineer.