Monday, November 2, 2009

Spiral generators

 I haven't posted anything in a while for a good reason: the micro-Marx generator has been on hold while I waited to see whether I won a spiral generator which recently popped up on eBay.  "What the hell is a spiral generator?" I hear some of you say.  Please visualize me putting on my pedant's hat while I throw together a description cribbed from a patent.

 A spiral generator is an interesting fast pulse generator which works in a way similar to a Marx generator: a number of opposing electric vectors (put simply, directions of electric charge) are suddenly rearranged to be in series addition.

 As near as I can tell, spiral generators were first reported by Fitch and Howell (PROC. IEEE, Volume 111, No. 4, pages 849-855, 1964).  The earliest patent on a spiral generator that I've found (1967) was assigned to a fellow who was working for Tobe Deutschmann Labs at the time.

 I'll try to describe how these things are constructed.  First, imagine a rolled-up strip transmission line (stripline) with an additional layer of insulation between turns. Each of the N turns of the spiral consists  of both an "active" and a "passive" stripline. The "active" stripline is the one you have before you rolled the whole thing up into a coil (a spiral seen from the end, thus the name).  The "passive" stripline is formed by the inner conductor of one turn and the outer conductor of the next turn, separated by that additional insulation layer.  This image should help make this clear:



 The shaded gap between the inner (A) and outer (B) conductor strips represents the active stripline formed by the two strips, and the unshaded gap represents the passive stripline.

 The outer end of the active line is connected to a switch (the rectangle with two semicircles at the upper right). Also shown at upper right are the biasing resistors and a trigger pulse isolating capacitor associated with switch operation.  That's a distortion-triggered spark gap switch, for which there is no standard electronic symbol.

 The remaining ends of both lines are open. The output terminals are normally connected with the reference or ground terminal being connected to the outer active stripline and the high voltage output terminal (C) being connected to the inner passive stripline.  Note that in this rather misleading drawing (which I stole from an old patent) the connection of the outer stripline to the reference terminal is not shown.  In fact, this drawing makes it look as if the active stripline is shorted to itself at the center, which is NOT the case in a real device.  No, I don't know why this drawing looks that way.  It does not agree with the patent description.  I guess the patent examiner was asleep at the wheel when this patent was awarded.  I couldn't find a better image, and haven't the time to draw one.  If I do, I'll replace this one and maybe edit the post for clarity.

 Initially, a DC charging voltage Vc is applied to the active stripline causing both striplines to be charged to the voltage Vc with the electric field vectors of the active lines oriented in one direction and the electric field vectors of the passive lines oriented in the opposite direction.  In other words, before the switch closes, the charge across the insulating layers A-->B points in one direction (toward the outside, let's say, just for example) and the charge across the insulating layer B-->A points in the other direction (toward the inside).

 The generation of the high voltage output pulse is caused by the alignment of the active line vectors with the passive line vectors.  When the switch closes, a wave travels down the active line, reflects off the open circuit at the inner end, and travels back to the switch, reversing the polarity of the active line relative to its initial charge.  Now all the charges point in the same direction, causing (for a very, very brief period of time) the voltage between the active and passive lines to add, or double.

The voltage across each active and passive line pair is then ideally (assuming perfect materials and construction and thus no losses) -2Vc. If there are N pairs of active/passive lines then the total output voltage ideally is given by the equation:

-2NVc.

The time for the spiral generator to erect is given by the equation:

t = 2πND/v

where N is the number of turns, D is the mean diameter of the spiral and v is the propagation velocity of electricity through the stripline (aka local speed of light).  Note that the speed of electrical propagation through materials is always slower than the speed of light in a vacuum, although said slower speed can be maximized by the use of transmission lines.  A stripline is a form of transmission line.  That "wave" I mentioned is moving at the propagation velocity




-=o=-

Now then, I just bought a very fast high voltage spiral generator from eBay.  The manufacturer doesn't say that's what it is, but a white paper I've got which mentions this specific product refers to it as a "commercial spiral generator".  The discussion in the white paper, the connections to the device, and its physical shape assure me that's what it is inside.  Not that I care all that much, I'm more interested in what it does.



Now as far as I know, all spiral generators require a very fast switch to initiate the inversion-inducing pulse.  Yet this particular device requires a pulse of only 250 V to fire, which doesn't sound like the trigger pulse for a spark gap switch.  I suppose it could contain a solid state switch.  Spark gap switches have finite lifetimes and usually require periodic maintenance to reach their rated life, but there doesn't appear to be such a device attached to this unit.  Since the case is sealed, I'm somewhat skeptical that it contains a spark gap switch inside.

This makes me curious about what the guts of this device look like, but I'm unlikely to ever know since PAE seems to have built it from phenolic laminate epoxied together.  By the way, this is one of the few exceptions to my usual hatred of products which cannot be opened or serviced.  These things cannot be serviced in any case, and in all other respects, PAE's design is entirely reasonable given the voltages involved and the requirements of functionality.


And finally, we get back to why the micro-Marx generator has been on hold.  Its primary purpose in life was to trigger the distortion-triggered spark gap switch I built.  This item, assuming it is in working order (and I have convinced myself that it most likely is) will trigger said switch better than anything I am capable of constructing.  It is faster than any Marx generator I'm likely to build, and has adequate output voltage.  The Marx would have had 2X the output voltage of the spiral generator (probably desirable) but would have a much slower rise time (undesirable).

I find it very amusing that the patent I took the above image from uses a distortion-triggered spark gap switch to fire the spiral generator, when spiral generators are often uses to trigger large spark gap switches.

I will probably go ahead and build the Micro-Marx eventually, because I intend to build a larger one some day and I need to learn the practical construction issues of these things at a smaller scale before I attempt the big one.

But for now, the need for the small Marx is much less urgent than it was, unless this new toy turns out to be an $80 paperweight.  Of course, if it's in working condition, it'll be the steal of the decade for me.

(note to self: call up PAE and ask 'em how much these things sell for and also what the stored energy is)

EDIT: they cost about $2,600, and have 40nF of capacitance, which equates to about half a joule of stored energy.  As pulsed high voltage power supplies go, that's wimpy, but as trigger generators go, that's plenty.

I expect my next post will concern my testing of the PT-55N after it arrives.  How the hell am I going to trigger the thing?  It wants a trigger input of not less than 250V with a maximum rise time of 10nS (!) and a minimum pulse length of 150nS.  The voltage is no big deal, the pulse length is no big deal, but that rise time might be tricky.  Perhaps a fast FET, a low-inductance capacitor and some careful construction geometry will do the trick.

-=0=-

I am also still working (albeit very slowly) on the "mad scientist light switch project".

3 comments:

Todd said...

Hey, I found one of these trigger generators in the trash at work many years ago! I still have it stashed in a cabinet.

It might be broken however, since someone has pried one end of the case open. As I read your description of spiral generators, the image of the guts of this thing popped into my head. And of course the photo clinched it!

You wondered about the switch. This thing contains a _Krytron_!
That's primarily why I rescued it from the recycling bin.

-Todd

Bill Lemieux said...

No kidding? The exact same make? I remember you mentioning finding a krytron in some piece of gear in the trash, but I didn't remember what it was. Shame it was disassembled, it's almost certainly unsalvageable now. But its construction might be interesting to examine.

A krytron makes good sense. In PAE's letter to me, they did mention this thing requires a license to export, but that's true of most very fast high voltage pulsers or switching devices, whether they are bulky or not.

I wonder how they managed the keep-alive supply for the krytron. I suppose it could all be done with dividers. It does require 0.1mA from the charge supply, which seems a bit much for simple dielectric leakage if the thing is well made and back-filled. It has a fairly narrow operating range - 5kV to 7kV.

I also discovered that the lifetime of these things isn't so long - they rate the PT-55 for 10,000 shots... IF you run it at 5kV for long life.

They also suggest that for best jitter, you run it at the max charge voltage of 7kV and the max trigger voltage of 600V, but they don't say what the shot life is under those conditions.

I've got a KN-22 lying around somewhere. I suppose that's one way I could fire this silly thing.
Well, maybe. The KN-22's minimum anode voltage is 400V, typical is 4,000. I suppose one could divide the output. Two other krytrons will operate down to 300 volts.

Krytrons also have fairly low shot lives...

Or I could just save up $270 for PAE's trigger module. *sigh*

Bill Lemieux said...

Oh, and krytrons happily trigger at few hundred volts.
Why didn't I think of that?