Progress in Solving Switching Problems

Posted by Rezwan on Dec 08, 2009 at 12:10 AM
Tools: Print | | (18) Comments | Discuss In Forums

See also accompanying photos courtesy of yours truly.  And note the headings are written by me - hopefully not perjorative.  Now, back to Eric:

Summary of progress

We spent much of the month in initial shake-down and tuning of Focus-Fusion-1’s performance.  We made good progress in solving problems with the switches that caused the capacitor bank to pre-fire before we could trigger it and for the current to rise too slowly.  Problems such as these are routine for new experimental facilities and are to be expected.  We expect to finish resolving the switch issues in December and begin moving on to our first major experimental campaign with the angular momentum coil. 

Problems with Switches

In early testing of the FF-1 device, we saw that there were two problems associated with the spark gap switches that we purchased from a supplier.

• Pre-firing.  Even at relatively modest charging voltages of 25kV, the switches pre-fired, meaning that one of the switches fired by itself before the bank had fully charged and before we initiated with the trigger. 
• Timing.  The onset of the current was much slower that we needed, and the switches were not firing at the same time.

How a switch works

The switch consists of two copper electrodes mounted on metal plates, with a slightly pressurized gas between them.  The plates are screwed into a plastic insulator which also serves to contain the pressurized gas.  In a hole in the upper electrode (which attached to the “hot” or upper plate of FF-1) is the tip of an automotive spark plug. 

When the trigger generator fires, it creates a pulse of high voltage (HV) which travels through cables to the spark plugs in each of the switches.  The negative HV pulse from the trigger adds to the positive HV sitting on the capacitor (which is in contact with the lower switch plate).  This increased voltage causes the insulating gas to break down, electrons are stripped off of atoms, and the current begins to flow between the two electrodes.

Possible pre-firing culprit

We think we have identified the reason for the pre-firing.  The plates of the switches that the copper electrodes protrude from are made of aluminum, a choice by the supplier that we were surprised at.  Aluminum has a low melting point of 660 C and is a poor choice for any material exposed to plasma.  We saw that most of the plates were heavily cratered and pitted, with the aluminum having melted and re-frozen into very rough surfaces.  This scarring was caused by pre-fire arcs far from the copper electrodes and in turn produced a surface that encouraged more pre-firing.

Action taken for pre-firing

To prevent pre-firing we are having copper inserts made that will cover the aluminum plates in the entire area that they were exposed to the switch plasma.  Copper’s melting point is 400 C higher.  We had two such insert sets machined, and tested these two switches by themselves last week.  As expected, this seems to have cured the pre-firing problem at least at low voltage.  We fired 8 times at 20kV with only one pre-fire.  We intend to continue testing next week at higher voltages, but are convinced that all switches need to have the copper plates made for them.

Action taken for slow firing

In the meantime, we heard from the supplier that he had made a mistake—again a very surprising one—in telling us to use nitrogen for the gas between the plates.  This leads to the slow firing. Instead, we could get quick firing with “switch gas”, a mixture of SF6 and argon.  So we expect this will solve the slow-firing problem when we receive the new gas this coming Thursday.