Monday, February 21, 2011

playing with arc-flash for fun and profit

I worked on the big pulser (for the railgun) over the weekend.

 Lately, job #1 has been to figure out a way to connect the capacitors to the "hot" collector plate.  The means had to be inexpensive, reasonably low effort, and capable of handling the full 60,000 amps from each capacitor.

 The arrangement I am trying first is to clamp four soft copper straps to the capacitor's bolt terminal with a nut, and clamp the opposite ends under the bolt rings of the old ignitron housings.  The rest of the housing won't carry current.  Some nice low-rise clamping rings would be better, but I don't feel a pressing need to make six of them (with eight holes each, ugh) right this minute.

 I made the straps and installed them yesterday.  This involved cutting out forty-eight 1" x 4" x .030" straps from copper sheet (with sheet metal sheers), removing all the corners and sharp edges on the belt sander, and punching forty-eight half-inch holes near one end of each strap.

 After that, I felt an urge to know the leakage rate of the capacitors.  I decided to perform a "megger" (a high voltage insulation tester) test on all six caps.  Now meggers, by their nature, typically put a fair mount of voltage across the unknown.  My instrument, an old Genrad unit, puts out anywhere from 100V to 500V DC.

 Then, before I could start with the measurements, I had to figure out how I would safely discharge the cap bank after it had been charged by the megger.

Step the first was to wind half a dozen turns in my test leads onto some big ferrites.  That should, in theory, keep the hash from the arc out of the detector circuitry of the megger.  I would disconnect it by hand before shorting the bank if I trusted my gloves, but I don't.  Anybody out there have a set of lineman's (Class 2 or 3) gloves in test-passing condition that they don't want?

Step two was to don rubber gloves, leather gloves, face shield, and ear muffs.

 I got my measurement on the low range (100 V) and decided to take measurements at all the other ranges too.  I discharged the caps after each measurement.  I used a "chicken stick" (insulating shorting stick with copper bar at one end) and an extra little bit of brass lying on the top of the collector plate to strike the arcs to, so the collector plate wouldn't get too badly torn up.

 At 100 volts, the 'pop' was impressive, but one does not need hearing protection.  A modern firecracker is louder.  At 200 volts, I decided the hearing protection was a good idea after all.  At 400 volts, I worried what the neighbors might be thinking.

 Note that these are 10,000 volt caps.  So I was playing with 4% of the total energy this cap bank can store.  I won't lie: when this machine lets go with a full charge, I don't want to be in the same room with it. Fortunately I won't have to.  I have a nice concrete block wall separating my workshop from the rest of the garage.  With that and the remote control panel I'm building, we ought to be golden.

 In reality, we don't want this (a short circuit discharge) to happen at any voltage.  Without any significant inductance or resistance for a load, the very high currents and high voltage reversal caused by high frequency ringing (since the "load" inductance is so small) stress the dielectric in the caps rather badly.  It is to be avoided.  These caps don't grow on trees.  Ordinarily, when the caps need to be discharged (either to remove residual charge at the end of a shot, or due to an aborted shot) the caps are drained more slowly (and safely) by a large bank of power resistors connected to the cap bank through a pair of high voltage relays.

 This is also why I'm testing insulation and clearances to the cabinet and the like.  I do not want an uncontrolled arc happening out in the open.  Aside from the obvious undesirability and damage, it will also be EXTREMELY LOUD.  We're talking broken-windows-loud.  Under normal operating conditions, I expect most of the energy to be absorbed or muffled in various ways.  In fact, I am going to rather a lot of effort to ensure that happens.

 Figuratively speaking, I have in the past had the local villagers light torches, take up pitch-forks, and storm my castle because I frightened them. News media, emergency services, and attorneys were involved.  So, let's just say I am not comfortable allowing extremely loud noises to escape from my property these days.

 Oh, and I hope to have the shop's old rolling door finally ripped out and replaced with a wall before the first gun firings take place.  That should cut down on the blast noise a LOT.

After doing the above, I decided I really need to do a proof test on each individual capacitor, and that means disassembling everything I assembled over the weekend.  Well, to be honest, I had other reasons for doing that too.  I need to do a proof test on the collector stack, and THAT has to be done with the capacitors disconnected (but with the stack still mounted on the caps) for what I hope is an obvious reason.

Photos to come later.

PS: I often forget what I've described and what I have not.  It occurs to me that I have not described the capacitor bank high current connections, which I've been referring to above as "the collector stack" so I'll do that now.  The capacitors I have are very early pulse caps using 1960s technology.  They are set up for relatively low inductance connections, although improvements have been made in pulse cap design since these were made.  The capacitor connections are made between the case and a single 1/2-13 brass bolt located in the center of a round, convoluted, hard rubber insulator.  The return connections to the case are made to four brass blocks soldered to the corners of the case.  Threaded holes are provided in said blocks to accept 3/8 brass studs.

To minimize stray inductance, the six capacitors are bussed together with aluminum plates separated by a thin layer of strong insulating material.  The ground/return plate sits directly on the capacitors' ground return blocks.  It has six holes about 4" across to provide clearance for the high voltage connections.  A 1/2" thick plate (could be different material, and thinner) of phenolic-linen laminate goes on top of that. It also has big holes to clear the high voltage connections.  The "hot" or output plate has big holes too, but they are surrounded by rings of eight 1/4-20 threaded holes so that "something" can be connected there. Originally that "something" was an ignitron on top of each capacitor, which sat inside of a coaxial housing which connected to those rings of holes.

I am still using those housings and the original connections for my new connections, only because they are handy and they eliminate - for now - the need to make a whole bunch of new parts.

No comments: