Re: Tesla simulations vs. scope traces

From: 	Rscopper[SMTP:Rscopper-at-aol-dot-com]
Sent: 	Wednesday, December 24, 1997 11:59 AM
To: 	tesla-at-pupman-dot-com
Subject: 	Tesla simulations vs. scope traces

First to Jim:

The real coil is two secondaries; an 8-inch on bottom, and a 6-inch on top.
The two are separated by a small toroid (that is also connected to the two
coils).  The toroid  seems to cause all of the primary field to couple to just
the 8-inch.  I say this from the shape of the corona that can be seen when in

I took several measurements of the system Ls and Cs with a calibrated bridge
from work.  The total L of the 2-coils was 73.8mh.  The measured C was
0.349uf.  The measured RwindingsDC was 54.7ohms. I haven't decided how to
model the measured C yet, but I'm sure we'll figure it out. I used the
calculated self-C from WinTesla for the model.

To Antonio:

The reason the first C and the last L don't follow the exact formula, is that
I made sure the total was equal to my measured values without having to have
more sections.

I think the model is a good start, however a few things are missing.  If the
system is modeled by a mass on a spring that is thinner at one end and gets
thicker as the windings progress, It would react to higher frequency
vibrations at the thinner end, and then propagate to the other end as I
believe the current in a TC does.  If you then model the mass sitting in a
bucket of jello (toroid-C) it would take a few oscillations before it could
break-out of the liquid (sparks).  I not sure of the analogy for the R-
windings, and for the C made of the cylinder of the windings to ground (not
self-C), with humidity added in for kicks. But I don't think the system is as
simple as a RLC model.

To Scott:

The only spectrum analyzer we had at work that could measure the coil when
operating was a HP about 2ft X 3ft, and they wouldn't let me take it home (too
bad...I realy wanted to). So you live with what you have. I measured the
3.83Mhz with the scope on 500ns/div, I ran the scope trace points through the
FFT just to see if there were any other major components that weren't obvious.
The AC analysis of the model did give somewhat different results depending on
the number of samples I specified.

Simulations are only as good as the model, and this model is not yet complete,
but  it does raise some interesting questions.  

I also agree with JHC and others, that if you build enough prototypes and
check your results with the simulations, it can only verify the results, or
cause you to re-think your theories.

Scott C.