Thursday, November 12, 2009

VIG / spiral generator update

PT-55N front.jpg

PT-55N side:rear.jpg

Since my last post, I have learned that there are two likely failure modes for this spiral generator in particular:

1) the dielectric of the striplines could have failed (ie; puncture).

2) the krytron used to switch the striplines could have worn out.

I have just tested the dielectric by the relatively simple expedient of applying the specified charge voltage and measuring the current.

The unit is specified to operate between 5kV and 7kV. I have on hand a small HVDC supply nominally 5.5 kV @ 0.5 mA, however it develops 6.6kV across an open circuit.

The unit charged up to 6.6kV in a few seconds, and the current at 6.6kV was 0.1 mA, which is exactly what it is supposed to be.
This is good news.

The only way I can think to test the krytron is to actually try firing this thing. According to the manufacturer, that requires a +250V pulse with a 10nS rise time. That's a darned fast pulse, but it should be do-able with a small, fast pulse capacitor, a FET, and some tight construction geometry.

10nS equates to 100 MHz, so presumably I can use ordinary RG58 cable for this, but I wonder whether there is something better. I was once given a rule of thumb by an RF engineer regarding preservation of pulse shapes and signals: if the shape of the signal is important (ie; single- shot rise times, square waves, etc) use cable and connectors rated for at least 2X the maximum frequency of interest and preferably 10X!

The trigger input connector is a BNC. 50Ω BNCs are good to 2GHz or so, and RG-58 (foam dielectric) is good to about half that, depending on how much attenuation one is willing to endure. Looks to me like I just need to figure out how to generate that pulse.

Then there's the minor issue of how to observe the output. The unit comes with a 1000:1 output monitor with 50Ω output impedance, which is tres convenient, but how well my storage oscilloscope will capture this remains to be seen. My scope's bandwidth is only 500 MHz, and for now, I don't even have a sufficiently fast vertical plugin for said scope, but I think I can borrow one from a friend.

Another interesting fact I learned is that the generator has only 1,500 ohms of output shunt resistance (measured across the output). This is the manufacturer's spec. The resistance looking into the charging connector - at 500V, the maximum voltage my megohmeter will deliver, is 135 MΩ. I suspect the "keep-alive" supply for the internal krytron is derived from the charge voltage through a resistive divider.

And finally, I'm going to start referring to these things as Vector Inversion Generators, because it seems that more pulsed power researchers use that term than "spiral generator". In any case, "VIG" is faster to type.

1 comment:

Railgap said...

It has been suggested by a friend of mine (who is 10 times the engineer I will ever be) that a better way of generating pulses with ridiculously short rise times would be to use an avalanche transistor.

Zetex makes several parts good to over 250V (C-E breakdown) which can generate pulses with rise times of less than 2 nS, if the physical geometry of the circuitry is done right.

At such short pulse times, we start to care about all the stray inductances in the circuit, so the physical construction of the circuit begins to matter.

A nice little fast pulser circuit is presented here:
http://joule.bu.edu/~hazen/avalanche/