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

Re: Wire Length (fwd)



Moderated and approved by: Gerry Reynolds <greynolds@xxxxxxxxxx>



---------- Forwarded message ----------
Date: Tue, 26 Dec 2006 21:23:33 -0800
From: Barton B. Anderson <bartb@xxxxxxxxxxxxxxxx>
To: Tesla list <tesla@xxxxxxxxxx>
Subject: Re: Wire Length (fwd)

Hi Paul,

Not to bad. Javatc when I configure for rim grounded case with topload 
at dimensions originally stated give me a 463 kHz, so really close. Les 
is 4.43 mH which a little off from what you indicated for f1, but I am 
finding Les (or Ces) and deriving one from the other. At the moment, I 
don't remember which I'm using (this is for exact back calculation from 
Fr, one to other, without slight differences. Some folks just had issues 
with that. In retrospect, it speeds up the process).

The waveforms don't appear to drastic. Something I can play with. 
Measurements will help sort out the data. BTW, I ran your waveforms with 
IE6 and yes, they stumbled along slowly, even after I was sure they were 
fully loaded. I use Netscape Email so when I click any link, my Netscape 
browser opens the link. Netscape worked just fine (nice fluid movement). 
I also tested with my Firefox which was also ok. IE displays each frame 
slowly, thus it appears to stagger. Not the same with other gif images 
with various frames. Something odd about your particular simulations 
that IE doesn't handle well.

Also, for those interested, Firefox works pretty good with Javatc and 
Fantc! It's just as fast as IE6 and there is even a way to get rid of 
that annoying message screen forever (just ask offline if interested). 
Netscape still sucks! Opera isn't all that great in this region document 
object handling either.

We need measurements so we can get real data for a case such as this. 
The build is more difficult than it may appear. I imagine Tesla ran into 
the same difficulties I'm running into. His approach will different than 
my own.

I ran into a slight stumbling block today with the build and will have 
to go another route (form that replaces Tesla's wooden slats behind the 
coil). What I had in mind was runners between each long runner to attach 
the intermediate runners, but no good. I didn't feel comfortable with so 
little material at the joints (their strength). So, I decided to pick up 
a big slab and simply cut circle runners. i.e., 1" width circles with a 
51" diameter and another with a lesser diameter. I can then attach these 
via dowels to the main runners and then attach the smaller inner runners 
to them. No one probably knows what I'm talking about here. Anyway, I 
need to modify the design (only mechanically thus far).

So Paul, how did it feel getting back into a little "code investigation" 
again? I've been doing a little of that myself this week (it sucked!). I 
decided to figure out "point to a line segment regardless of line 
segment angular dimensions". Not as easy as it sounds. Had to re-visit 
trig in some detail. I simply wanted to update Javatc's pri to sec 
clearance number so that the nearest point of "any" possible 
configuration was known and using the same equation to minimize coding. 
It took a bit of work, but I finally got it nailed down. I then decided 
to code the algorithm. I made the mistake of defaulting a value of 1 and 
also introducing a value of 1. Took me a full day to figure out that the 
formula was actually working! (Of course, I had to feel out every part 
of the code before I thought of it).

Anyway, thanks for running the numbers and for figuring out the reason 
for the error. Measurements will come when the coil gets built.

Take care,
Bart





Tesla list wrote:

>Moderated and approved by: Gerry Reynolds <greynolds@xxxxxxxxxx>
>
>
>
>---------- Forwarded message ----------
>Date: Tue, 26 Dec 2006 07:49:55 GMT
>From: Paul Nicholson <paul@xxxxxxxxxxxxxxxxxxx>
>To: tesla@xxxxxxxxxx
>Subject: Re: Wire Length (fwd)
>
>Bart, All,
>
>With the primary and center load added, the rim-grounded planar
>spiral model gives:-
>
>f1          461 kHz
>f3         1228 kHz  
>f5         1960 kHz
>
>Les @ f1    4.61 mH
>Lee @ f1    5.35 mH
>
>Cdc        92.3 pF
>Ces @ f1   26.3 pF
>
>Lpri     16.5 uH
>
>Addition of the center load has pulled the frequency down by
>only about 4 or 5 percent.  But it does seem to stabilise the
>calculations quite a bit - the variation of f1 with different
>segmentation plans is about +/- 5 kHz.
>
>Highest surface field gradient on the topload is about 0.11
>kV/cm per kV of center voltage, occuring on the top of the
>sphere.  So if breakout occurs at 24kV/cm, that's a center
>voltage of 24/0.11 ~= 220kV.  
>
>I ran a time domain model using a primary cap of 7.22nF and
>a firing voltage of 10kV.    The 'split' frequencies come out
>at 382 kHz and 613 khz, giving an effective k of 0.44.
>
>Here is an animation spanning about 3.5 full beats (pri-sec
>transfer in about 2.1uS).
>
> http://www.abelian.demon.co.uk/tmp/baps-rg3.anim.gif
>
>There is incomplete energy transfer (not surprising - no
>attempt is made to fine-tune the model), but the primary energy
>residue doesn't look too bad at all.  Not too much should be
>read into the fine detail of these waveforms: small changes
>in the modelling produce relative phase and frequency shifts
>of the various components, altering the detail significantly.
>
>But the voltage gradient looks ok, doesn't it? No nasty steps
>or spikes. There is an HF transient rolling back and forth,
>but this shows up mainly in the current waveform, indicating
>it is of quite low impedance.
>
>Peak center volts for a 10kV firing is around 145kV, so no
>breakout predicted if a simple 24kV/cm surface threshold
>applies.
>
>BTW, I had a play for the first time with the IE browser on
>windows.   It didn't make a very good job of playing these
>animated gifs - very slow and jerky.   I hope that's not
>generally the case with IE.
>
>--
>Paul Nicholson
>--
>
>
>
>
>
>
>  
>