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

Re: ScanTesla program

Original poster: Terry Fritz <teslalist@xxxxxxxxxxxxxxxxxxxxxxx>

Hi Steve,

I've already done the 3-d plots by hand :P I used the parametric simulation feature to do blocks of 10 simulations varying one parameter, and did 10 of those blocks, then typed the 100 results into Excel for 3-d plotting. The parameters I investigated were streamer resistance and capacitance and the dependent variable was peak primary current.
http://forum.4hv.org/attachments/loadingmap.gif

Yes!! That is the type of stuff the program should do! Like you, I just don't want to sit there working for hours running models and typing in numbers. The new program should make it all automatic. Of course, it can do many parameters are run a very large number of models that we could never do by hand.

Antonio Wrote:

Could be something similar to the programs Optesla and Optmag, that I
wrote to search the design space for capacitor-discharge Tesla coils
and magnifiers (http://www.coe.ufrj.br/~acmq/programs). I didn't add
a load in your style, but this would not be difficult.
I have a DRSSTC simulator in the program Sstcd. Almost your structure.
In that case I didn't use an exact solution, but this would not be
very difficult to implement.


Jim Wrote:

1) Do this as a matrix equation as in classical network analysis (like
Spice, MicroSim, etc.) use.  You build up an matrix representing all the
interconnections and admittances, then solve it.

Very close to the same thing as Antonio's program but more designed to scan a large number of parameters. Right now I really don't want to find closed form solutions since the program may evolve in ways that make that very difficult. If the loads start to get time dependent like in Steve's model or other changes are made (output feedback), fixed solutions equations my be difficult to change or intractable. The P-spice style number crunching is orders of magnitude slower but it can handle any situation with relatively minor changes. Eventually, it may be obvious that things don't change much and it would be best to solve the equations to take advantage of the higher speed. At that point, the "answers" might be directly calculable anyway. Also, the matrix math is a little more than "I" want to chew on 0:-)

I don't want the program to be tied to Excel or Matlab since it should be a total freebie that anyone can run. It may not be the optimal or best way to do things, but it should be a good place to start out. Those program are nice in that one can fiddle with things easily. But LCC is almost interpreted language like these days so fiddling is pretty easy even in C. I "think" you can even import C type programs into Excel and Matlab as sort of a "function engine", but all that is far beyond me and my use.

Of course, if anyone does not like it the way it is, it should be very easy for them to change to their tastes ;-)) Or they could write their own! 99% of it is just typing and debugging... There is no real "hard theory" behind it.

It is really not programmed toward a specific goal but rather a way to do a vast number of test automatically and just see "what the computer thinks" the best numbers are. I remember with the LTR coil "hand scanning" leading to far better running parameters... Hopefully, eventually, it will help find optimal coil parameters to match up to longest streamer length. With the DRSSTC, that data is now readily available and a tool is needed to crunch a lot of numbers in a very versatile way. Microsim just is not able to do large numbers of parameters and report them well. Since the "engine" is pretty easy to reproduce, we might as well write our own program that we have total control over.

I think Steve well knows like me how much time one can spend running models in MicroSim and writing down the results one by one to make graphs, find trends, or find the best spot. Basically just using time (far less) to write a program that does all that while we sleep ;-))

Cheers,

        Terry