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

Re: measuring true self capacity of a coil.



At 08:43 PM 5/30/00 -0700, Jim Lux wrote:
>
snip....
>For that matter, how accurately are you measuring the dimensions of your
>test object?  A 1% error in the diameter of a sphere will give a 1% error
>in the capacitance. While it is fairly easy to measure a 1 meter sphere to
>an accuracy of 1 cm, measuring it to 0.1% (or 1 mm) is somewhat trickier,
>and even trickier is assuring that it is truly spherical to that accuracy,
>and that the surroundings are uniform to that accuracy. (unless you are
>measuring in a far field range.. like on top of a tall tower with a finite
>ground plane under the coil.)

That's a good point!  One of my test grids has a sphere I always assumed
was 12 inches in diameter.  E-Tesla5 always gave high numbers for it.  I
finally figured out it was really 13 wide by 11 tall with a metal edge
around it.  That small 1/2 inch error was very obvious when the computer
computations were done...  You have to be really careful to input good data...

>
>I would say that the uncertainty in the ACTUAL fRes introduced by HV
>discharge in the form of corona, leader, and spark formation is much
>greater than that from C modeling or measurement error.  What we need to do
>is think about ways to measure the actual characteristics of the spark
>channel as it forms.  I've been thinking that now that LOTS of computer
>power is available cheaply, we might be able to use a technique like that
>in Efield 5.x, except backwards... Put a bunch of e field probes and
>digitizers around the coil.  Let 'er spark. Now, calculate the charge
>distribution in the leader and spark that corresponds to the measured
>e-fields (lots o' computation, there..)

It is one of my dreams for E-Tesla to be a truly 3D grid instead of the 1/2
cylinder section it is now...  Then it could look at streamers.  Adding the
dimension of time would also be cool.  That takes the big number array
handling of something other than QBASIC...

>
>Or even something as simple as measuring the spark channel current from a
>deliberate breakout point on the toroid.  You could put one of those $50
>radio shack battery ProbeScopes inside the toroid measuring the current
>through a shunt or a toroidal transformer.

Fiber probes do that easily but this would be more available to those that
do not have big dollar toys...  Neat idea!!  

The sphere I talked of earlier has full 40 MHz input and output current
sensing with built in fiber-optic probes (just one of my many toys! :-)).
However, if anyone needs info on the values, shielding, calibration, and
tricks to make it work more than once :-O let me know...  The transduced
voltage levels are right in the range to go into an A/D recorder...
Perhaps a few high tech protection components would be a good idea too for
such a fancy electronic gizmo...

>
>So, given that sparks develop in a timescale of nanoseconds, has anyone got
>a clever idea on how to build a cheap high speed multichannel digitizer? 
>12 bits at 100 MSPS or 1 GSPS would be nice...  The video A/D's exist...

They seem to develop a bit slower.  Maybe 20MHz for the really fast stuff.
Most happens down at Fo...

Cheers,

	Terry