Friday, June 8, 2012

a new coaxial high current pulse transformer for pulsed power

(I say "new" in that I have yet to find anything in the scientific literature describing this exact geometry. I recognize someone may have done it and I just haven't seen the evidence / paper / photograph yet.)

This is a (rather poor, hastily made) 3D model of a high current, coaxial pulse transformer design I am fiddling with.  Its purpose is to match the relatively high source impedance of a high voltage pulse capacitor bank to the relatively low load impedance of an electromagnetic launcher - AKA railgun.  Coaxial geometries -- which is to say, with the current path of the one winding located inside the current path of the other winding -- can achieve coupling efficiencies over 90%.

This device is loosely based on a design by Pappas, et. al.1 but has been reconfigured from a solenoidal coil to a toroidal coil so as to contain the magnetic field within the device.

If I limited myself to using materials I already have on hand (so as to reduce costs) as a starting point, I derive maximum current ratings of about 65,000 amps on the input side and about 250,000 amps on the output side, for pulses of 4mS or shorter (Onderdonk). I am still working on the mutual inductances and what-not, but I seriously doubt I will be able to realize currents that high from my complete system. When I get done slogging through the math, or at least trying to understand the math since I am still a relative math cripple, it may prove impossible to have a transformer with sufficient current loop area to retain the magnetic field adequately and the correct transformation ratio, in which case I will probably scrap the idea of using a transformer entirely. A solenoidal transformer with fewer loops will create too much EMP. That doesn't negate the value of this idea for others of course.

Some leakage is inevitable, especially given the asymmetry necessary to make this design work and the small number of current loops I can manage given the amount of cable I have, but a toroidal configuration will nevertheless contain a large percentage of the generated field whereas a solenoidal design has considerable external field effects, which can be problematic with instrumentation, as noted by Beach2.

The secondary consists of a single electrical turn, divided into seven loops disposed radially around the device's axis of revolution.  The seven loops - shown as silver in the model - are fabricated from metal tubing, each end of which is inserted into one of the two circular current collector plates.  One plate is connected to the center conductor of the coaxial output and the other plate is connected to the shield of the coaxial output.  The two plates and the coaxial outputs are separated by plastic insulation.  The collector plate + insulator stack is provided with eight holes and 16 connectors to fasten the tubing to the plates.

The primary consists of a single piece of 15kV 2AWG "jumper" cable, pulled through the loops of the secondary and connected at to the shield output collector plate at one end.  The other end is provided to the driving pulser through a coaxial shield connected to the inner conductor output plate.  Thus, the secondary and primary form an auto-transformer electrically.

An additional annular brace may be added to the equator of the loop array.  This provides additional mechanical bracing to prevent discourage physical stretching of the loops.  Note that the brace may be made of metal welded to the loops at the same point on each, since all loops are shorted together.  this is probably stronger and lighter -- and certainly cheaper and easier to implement - than any equivalent arrangement of insulating materials.

All conducting paths have been sized with sufficient cross-sectional area to remain far below (50%) of the fusing current (Onderdonk Fusing Current formula) for copper.  This will result in significant heating of the device.  Increasing material mass would be impractical for the given project because large portions of the material are already on hand, which define many design constraints.  I currently have half copper parts for the output current collector, and I have the 27 feet of #2 HV cable required.

Assembly of most of the joints is non-trivial. If copper is used, most of the joints will have to be heat-shrink permanent connections.
The heat of silver brazing would be detrimental to the conductivity of the copper. Alternatively, the same machine could be built using machined and welded aluminum parts. This would be cheaper, however all of the dimensions would have to be increased. Nevertheless, a unit made from aluminum would weigh (and cost) much less than the equivalent ampacity and fusing-current-limit in copper.

For a high inductance transformer such as this to be used between a railgun and a capacitor-discharge current source, large "free-wheeling" diodes would need to be placed across each capacitor, or the capacitor bank, to absorb the residual back EMF from the transformer.  Depending on where during the discharge pulse the armature departs the rails, opening the secondary circuit, the back-EMF from the primary could be quite large.  Pulse caps do not like voltage reversal.  These diodes would add significantly to the cost of using such a pulse transformer.

This work was supported by a grant from The Joss Research Institute, Laurel MD.

1."High Current Coaxial Pulse Transformer for Railgun Applications", J.A. Pappas, M. D. Driga, W. F. Weldon, Proceedings of the Fifth IEEE Pulsed Power Conference, 1985 [link]

2. "Design and Construction of a One Meter Electromagnetic Railgun", F.C. Beach, Naval Postgraduate School Monterey CA, 1996

No comments: