[Prev][Next][Index][Thread]
Re: More arc simulations
Hi Terry, Richard
I mentioned once before how valuable empirical data is
in model validation. However, we now have mathematical
schemas that allow us to build the empirical data into the
actual model. I suspect, with the use of fuzzy sets, that we
can intersect the lab work with the existing models into a
new type of coil model that will behave like the real world
and have a firm mathematical foundation. I don't think this
is an easy project, but when I finish this academic term
in November, if you folks are willing to share existing models
and empirical data, I will attempt to create the "Fuzzy Coil
Model" and share the results with all on the list.
John
John W. Gudenas, Ph.D. Aurora University
Department Chair of Computer Science and Mathematics
347 S. Gladstone Ave. Aurora, IL 60506
630-844-5539
-----Original Message-----
From: Tesla List <tesla-at-pupman-dot-com>
To: tesla-at-pupman-dot-com <tesla-at-pupman-dot-com>
Date: Thursday, October 15, 1998 8:55 PM
Subject: Re: More arc simulations
>Original Poster: Terry Fritz <terryf-at-verinet-dot-com>
>
>Hi Richard,
>
> Models can do wonderful things, They can test new ideas and designs
>quickly, simply, and cheaply. They can search out and find design problems
>and suggest optimal component values. My earlier posts had modeling data
>that took about three hours to find. If I had to make real coils to do
>those tests, it would have taken years. There is no doubt that models are
an
>extremely powerful tool. However, the problem with models is getting the
>darn things to agree with the real world!
> It is easy to point and click up all kinds of models while sitting in
>the stuffed office chair with a mouse in one hand and a beer in the other
>(my preferred method :-)). However, unless the modeler drags himself into
>the lab (ok, clear spot in basement :-)) and actually holds the computer
>printouts up to the scope photos, the models will never really work. In
>the past, that comparison has been quite depressing! Hooking a scope
>to the output arcs to measure real-time currents is not easy to do. As
>far as I know, Greg and I are the only ones. Greg's data was collected
>early in the year and my probes got good enough to do it last week (my
>top terminal is too small to climb into with a scope in hand :-))
>Although, Richard, you may have done this too and haven't mentioned it
>in the time I have been on the list. I know you have most of the tools to
>do it ;-)
> In the past, there was not good measuring equipment to feed the models
>with the accurate real world data they need. A big hot arc burning the
>paint off the ceiling is real nice but the models need to know currents,
>voltages, phase, etc. Without some way to tie the models to actual
results,
>they will stray wildly off course. However, with good data, they can lock
>right onto a system with remarkable accuracy. We now have the tools to
>supply very good data to the models and in turn they are producing results
>that are finally agreeing with real results.
> It is very gratifying to know that the results the models now give,
>agree with what the true experts like yourself have noted for years.
>Indeed, the models show that all the parameters have to work together to
>produce a nice output. They also show that the little tweaks and rules we
>do by experience are in fact very valid once all the complex interactions
>are considered. Better yet, they suggest new little tweaks we may have
>never suspected. Indeed, when a model shows something and one can say
"Sure
>enough, that's just what Richard has been telling us for years!" it's
>probably working! :-)
> Unfortunately, the present modeling systems are too complex for the
>"average guy" to use easily. Hopefully, once we can get these big complex
>things to give realistic results, simpler and easy to use programs can be
>developed. There is nothing magic about how modeling programs work. They
>can be striped of all the bells and whistles into much simpler and user
>friendly dedicated tools once the basic mechanics are known. Fairly soon
>(few years), programs will be able to predict and optimize coil behavior
and
>suggest the best parameters to use in their construction given a set of
>"parts at hand" to use. The models don't have any hurtles to overcome.
>when they get good enough to spit out super coils, we will start building
>them like mad to just what the printout says. At that point, a little
magic
>will be lost forever in coiling, but our sparks will be better so we won't
>notice. Of course, there will be times when the "seat of the pants"
coilers
>finds something that our models don't predict and we can sit back and enjoy
>us all running around like wild rats trying to figure out what when wrong
>with the model. But that is just the way the modeling world is.
> We have all noticed that old knowledge get rediscovered by others all
>the time. There is no Tesla coil school so we all start from the very
>beginning without much background. No teachers or anything to drill the
>basics into us. We all start fresh at step one and climb our way up.
There
>are many books and such but the authors can't keep up with the state of the
>art (I have started mine six times now, but the table of contents keeps
>changing :-)). Lots of bleeding edge technology in this hobby but, of
>course, that's why we are all here.
>
> Terry
>
>
>