New type of TC RFI filter
Tonight while trying to get my spark gap model to fit real measurements
(not quite there yet :-0), I think I found what causes those giant current
spikes. Also, I think I have a way to eliminate them now.
It appears that the primary coil has self capacitance just like the
secondary coil. The space around the primary becomes charged and stores
energy in the form of capacitance. For my coil, I measured this
capacitance as 38.9pF. At 21kV that works out to only 0.0086 joules of
However, that energy is basically shorted by the spark gap and is
dissipated in about 500nS. So we have a pulse with a burst power of about
17 kilowatts. This pulse appears to be responsible for ringing every stray
L an C in the primary circuit and generates all kinds of RF noise up into
the GHz region.
Modeling shows that a small inductor (100nH) in the primary path would
easily stop this pulse. Unfortunately, a very small capacitance (5pF)
across that inductor would simply restore the high frequency path again...
Thus, it would almost be impossible to make the inductor with low enough
parallel capacitance to work. At thousands of amps, GHz frequencies, and
20+ kV a noise filter is not basic.
However, one may be able to make the high frequency filter with just two
pieces of heavy (like RG-8) coax. The circuit would look like:
Two x 2-foot long sections of RG-8 coax would be placed in-line with each
lead from the spark gap to the primary coil. The coax shield would be
grounded on the spark gap end and left open on the primary end. The ground
connection between the two coax cables would need to be very good! They
would also need to be grounded to the local system ground but with less vigor.
This may really help those that are having problems with RF noise...
It will take me a day or so to test this out, but I think it will work.
Any comments or suggestions are welcome...