Thursday, April 23, 2009

oops, I may have screwed up

Last night I sat down with pencil and paper and ruler and compass (none of which have ever been known to crash, or refuse to allow me to change a dimension, or do anything else which a poorly designed piece of software might do) and pipe...

...and figured out that there is something wrong in the dimensions of my own design somewhere. Or rather, I believe that my design, as embodied in the 3D model, is jest fine thenkyewverymuch, because I already did interference and clearance checks on it using SolidWorks' built-in tools. BUT, I seem to have lost the shop drawings I originally brought home, and I ordered the aluminum based on how I THOUGHT the thing was built and dimensioned... and I think I screwed up / misremembered, because 4" aluminum appears to be about 1/2" too small.

I know right? Me? Not remember something accurately? UNPOSSIBLE!!


So I'll try to look into that today, if work and life allow me the time. I also need to get in touch with $PSHRINK today, dammit.

Oh, but to return to the spark gap project for a moment, I think I can solve the issue only two ways:
1. scrap the parts I've already cut from the 1/2" aluminum plate I just bought and get some slightly larger
2. use smaller diameter clamping rods to hold the thing together, so as to ensure they will clear the outside of the trigger plane electrode AND not sit outside the radiused edges of the end caps. Blah.

#1 may or may not work. I've been making a lot of design decisions on this thing by waving my hands in the air and making educated guesses based on past experience. But I didn't like the idea of making any of the parts weaker. You see, when a sealed spark gap fires, switching a whole bunch of power, some of that power gets converted to heat in the spark itself inside the switch, which expands the gas quickly, creating a powerful shock. (the same mechanism is the reason lightning causes thunder)

So the other thing I did last night was calculate how much force the clamping rods could withstand if I made them 3/8" diameter instead of 1/2". I am using acetal (Delrin) rods. The short version is that six rods of 3/8" acetal could withstand an instantaneous force of 5,964 pounds total, ignoring whether and how much the threaded holes in the ends and the fasteners threaded into them can withstand. I need to look up some information on machine screws pulling out of various materials and how one handles that.

Then I decided to look at how much PSI overpressure would have to exist on the exposed (inside the o-ring seals) end caps to cause that total force on the rods. The area of that circle is 1.767 in2, so the pressure required to create the above force is 3,375 PSI, which is pretty damned high. Not bad! And no, we don't have any idea what the peak pressure inside the switch will be for a given shot, because it will depend on a whole bunch of hard-to-calculate factors, including whether the switch is operating in multi-spark, single-spark, or pseudospark mode. (you probably don't want to know)

Now, there is another component which sees that pressure spike, and that is the cylindrical housing. We need to do the hoop stress calculation to see how much strain the walls of the housing can withstand, based on an internal pressure of 3,375 PSI. The housing material I have lying around is some variety of polyurethane, and I don't know which variety, or how hard it is. The harder it is, the stronger (not always true for many materials, but true for PU, I looked it up). I thought PU would be a good material for this, because it is more flexible and less brittle than other materials, and ought to withstand shock very well (literally, by stretching a little bit and recovering). So I did the hoop stress calculation for a 1/2" wall thickness tube of the lowest grade PU (4,500 PSI) given my dimensions. And it seems it can withstand 1,791 PSI.

Er, oops. So the limiting factor isn't the damned rods anyhow, it's the housing. Mind you, I wouldn't have used the 3,375 PSI internal pressure number anyway. You don't design something without a safety factor. For general engineering where humans aren't TOO likely to get hurt of the design fails, you use at least a 2X safety factor, which would reduce our maximum operating pressure to 1,687. For human-rated devices (airplanes, cars, bicycles, crutches, or things which will be used near people and which might explode, sending a few hundred pieces of shrapnel into the flesh of nearby persons, cough-cough) you use a 4X safety factor, which suggests 843 PSI, based on the rod's strength. That also gives us a 2X safety factor on the housing. Hmm. Maybe. Sure would be nice to know what kind of actual pressure pulses I can expect. Testing this thing is going to be interesting. I wish I had some kind of very fast (ie; capacitive or piezoelectric) pressure sensor I could connect to the switch through one of the gas fill ports. Then I could connect such a sensor to my storage scope during test firings and see what kind of pressure pulse is generated by what kinds of energies in which operating regimes. I know that the heat released must go as I2*R, and R will go up with the amount of gas in the switch that the arc has to pass through, so it seems likely that pseudospark operation - which is done at very low pressures - ought to generate the least amount of pressure. And it's where I want to be anyhow for other reasons like not eroding the fuck out of my electrodes.

I would explain more of what is going on with all this in terms of design features and why it has to be the way it has to be, but I very much doubt any of my readers care. Mostly I am just "thinking out loud" here, for my own benefit.

No comments: