[Prev][Next][Index][Thread]

How do I make my HV chokes?




----------
From:  Malcolm Watts [SMTP:MALCOLM-at-directorate.wnp.ac.nz]
Sent:  Sunday, February 22, 1998 3:41 PM
To:  Tesla List
Subject:  Re: How do I make my HV chokes?

Gary, all,
             I have scanned my memory banks after reading your post:

> From:  Gary Lau  20-Feb-1998 1052 [SMTP:lau-at-hdecad.ENET.dec-dot-com]
> Sent:  Friday, February 20, 1998 10:40 AM
> To:  tesla-at-pupman-dot-com
> Subject:  Re: How do I make my HV chokes?
> 
> >>Gary Lau wrote:
> >> I have seen advice advocating L-and/or-R-only protection networks posted
> >> to this list many times but have not seen any rationale for using these
> >> configurations beyond anecdotal "I've been using this and had no failures
> >> yet".  Can anyone defend series-only protection networks in a more
> >> analytical manor?
> 
> >From:  Greg Leyh [SMTP:lod-at-pacbell-dot-net]
> >Although the NST HV winding does have a large L,
> >it also has a significant parasitic shunt capacitance,
> >on the order of hundreds of picofarads.
> >Perhaps these series LR schemes work against this
> >shunt capacitance to form an LC low-pass.  
> 
> Sorry, I don't buy it.  The parasitic shunt capacitance is a distributed
> capacitance.  At the outermost xfmr secondary turn, there is effectively
> no capacitance to shunt HF energy to ground, and the outermost turn is as
> likely a location to carbon track to the core as any other.  A lumped shunt
> capacitance, outside the xfmr case, is required to keep the HF nasties
> out of the NST.

I seem to recall Ed Harris measuring a shunt capacitance at the 
terminals on the order of a couple of nF.

The R may be needed to keep the filter from self-resonating
> >if the NST's self-C is not lossy enough.
> 
> The R is needed to keep the choke from self-resonating (actually just
> reducing the Q and resonant-rise) against either it's self-C or external
> shunt-C.  I believe the concept of the shunt-C wanting to be "lossy" is
> wrong.  Lossy implies a non-reactive parasitic component, i.e.
> resistance.  In a low pass filter, we want a simple, reactive shunt
> capacitor.  Any series resistance in the shunt-C diminishes it's effectiveness.

Interesting argument. Yet, you agree that there must be some Q-
killing loss in the circuit. You need resistance to achieve a 
Butterworth response in a filter. Can't series R be modelled as an 
equivalent shunt element and vice-versa?

> Having ranted on this topic now, I must confess that despite following my
> professed beliefs and meticulously constructing an L-R-C-safety gap
> network, and measuring and confirming that my chokes aren't saturating, I
> still killed a 15KV/60mA NST, later reviving it by depotting and
> repotting in Vasoline.  Could it be that NST's in tar are doomed
> regardless, but detarred, rebuilt ones have better survival rates?
> Sorry for this anecdotal theory ;-)

My theory (which I have yet to test properly) after examining safety 
gap operation: The chokes themselves and any other inductance between
the gap and the transformer terminals is the real killer. As soon as 
the gap fires you have energized chokes with energy stored in them 
and nowhere else to go except back to the transformer. My money says 
that positioning the transformer as close to the main gap as you can 
with only some series R and some "lossy" shunt C should do the trick.
I will condemn one of my own neons to checking this out soon.

     I mused on the placement of the main gap vs cap across the 
transformer terminals too. This whole thing seems to be a bit each 
way. If you position the gap across the transformer, the mechanism 
noted above is at work. If you position the cap across the 
transformer, the transformer appears as a shunting element in the 
resonant circuit formed when the gap fires. I'm opting for the gap in 
this position so at least the tank operation won't be affected.

Malcolm