Re: Static Gap Break Rates

["Tesla list" <tesla@xxxxxxxxxx>]

Original poster: "Gerry  Reynolds" <gerryreynolds@xxxxxxxxxxxxx>


Hi Ed,

>    Try setting the peak voltage near the breakdown voltage of the 
> gap and playing with the capacitance.  You can get conditions where 
> a spark only occurs after several cycles elapse, a condition I've 
> seen with a "real coil".  The real coil has variable leakage 
> reactance which I don't know how to simulate.

I've done that both in simulations and for real.  The waveforms looks 
the same.  One problem is how we define the peak waveform.  We can 
define it as Vs_oc_peak (the peak no load voltage of the secondary), 
the peak of the steadystate response (with no SG firing), or as the 
maximum peak voltage including both the steadystate and transient 
responses.  If using a 15KV NST, the first definition will result in 
~21.2KV peak (assuming 120Vac input).    For the second definition, 
it is very possible to have a Cp/Cres value where the steadystate 
response will be higher than Vs_oc (not even counting on transients 
caused by the SG) and could take several cycles to "ring up" to this 
voltage.  Some may call this resonant rise.

You can calculate the steadystate voltage on the cap by figuring the 
series RLC circuit values.   Measure Rp and Rs and calculate the 
turns ratio N and R will be Rs + Rp*N^2.   L can be calculated from 
the commonly used xformer impedance equation (Z = Vs_oc/Is_sc), given 
the line frequency and R, by figuring the XL first from vector math 
and then calculate L.  You can simplify things by neglecting R and 
assuming that all of Z is due to XL. For the NST's Ive measured, this 
is 99% accurate.   C will, of course, be the Cp in the TC 
primary.  Once you do this, the steadystate response will be:

Vss = Vs_oc *  (1/LC) / sqrt [(1/LC -w^2)^2 + 
(wR/L)^2]          where w =2*pi*line_frequency

It could take many cycles to get to maximum voltage when close to 
resonance and, if your SG is set this wide, it will take many cycles 
for it to fire. How many cycles, I think, depends on the Q of the 
circuit and where Cp is in relation to Cres.  I dont know how the 
variable leakace reactance affects this as I havent explored this 
realm.  The simulation models do have saturable transformers 
available and I suspose one could experiment.

> Something else I've seen in a simulation which I think is impossible 
> is the sparks occurring on only on one half of the cycles, a 
> condition which would result in DC flowing in the transformer.

I have even seen a whole burst of firings where it only fired on the 
one half (both for real and in simulations).  Actually, I've seen it 
for real even before I ever did any simulations and was wondering how 
that was possible.  In simulations, the one sided firings lasted for 
a while followed by opposite side firings then something else 
happened.  I think it all depends on how the natural and forced 
frequencies beat with each other.  It can look like DC for the window 
that the behavior is occuring in but I believe like you do, if one 
waits long enough, the opposite firing will occur and the net DC will 
be zero.  Without a digital scope, my observations were limited and 
excluded the bigger picture.

Take care,
Gerry R