Conflat Leak

September 8, 2018 at 1:29 pm

In my previous post I wrote that I was aiming to do fusion last weekend.

Once the high voltage line was fully wired, I was ready to evacuate the chamber and test the high voltage supply – 30 kv, 20 ma.  Pumping down the chamber takes some time, usually upwards of an hour.  First, the roughing pump pumps from atmospheric pressure to <10 microns (~10-5 atm), then I start the turbopump.  This is a high vacuum pump that takes the pressure down another few orders of magnitude, between 0.1 and 0.01 microns.  Fusors actually operate at about 10 microns, but it’s desirable to have them filled with pure deuterium.  The first you get to 10 microns it’s all air in the chamber, but if you pump down to 0.1 microns and then fill it up with deuterium to 10 microns, you’ve got a 99% pure deuterium atmosphere (0.1 microns of air, 9.9 microns deuterium).

Anyways, once I started the roughing pump, something was obviously wrong.  There was an audible hissing sound, indicative of a gross leak.  Installing the high vacuum line involved changing only one vacuum connection, so I immediately suspected that.  The high voltage feedthrough I had just installed was sucking in air through the conflat seal – this is not supposed to happen.

Generally, a conflat leak is solved by tightening the bolts a bit more, so I tried that first.  No joy.  Then I replaced the copper gasket with a new one, still no joy.  I couldn’t see any defects in the knife edges on the ports, these bite into the gasket.  If there’s a nick in them it can prevent a proper seal forming.  Again, no joy – I was stumped, and decided to ask one of the physicists I was to meet at my first week at Trinity College Dublin.

At TCD, Prof.  Cormac McGuinness advised me to do another examination of the knife edges.  That’s what I’ll be doing this weekend.  If I can find a hair, dust particle, or nick in the edges I’ll have identified the source of the problem, and should be able to fix it.  If all goes as expected (unlikely!!), I’ll be doing fusion shortly.