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

Re: Where am I going askew ?



Original poster: Bart Anderson <classi6-at-classictesla-dot-com> 

Hi Ed,

>Christopher,
>
>I read Bart's reply on your questions.  I also ran some calculations on 
>your coil but for some reason, came up with different numbers than Bart did.
>
>(Bart, the primary difference seems to be in the resonant frequency of the 
>secondary.  I used 49.75 turns / inch and at a 97% winding factor - i.e. 
>we don't usually achieve a turns per inch as tight as in the wire 
>tables.  I get 1,086 turns total.

Ed, using 24 gauge wire (which is what was used), Javatc calcs 44.4 
turn/inch. 24 gauge is 0.0201' bare and 0.0012" insulation (single build). 
This is nominal magnet wire values from Phelps-Dodge tables. If anything, 
44.4 TPI is on the tight side. In the real world, a closewound coil will 
typically end up with less which can be put on the shoulders of 
human/machine error and tensioning affects. This of course will decrease 
the number of turns per unit area. Your 49.75" is extremely tight for 24 
gauge wire (unless the wire is far outside of nominal values). Possibly you 
used some other gauge?

That is one factor of of why your Fres is lower. The other factor is your 
loaded C value (your program is probably seeing 9.3 pF Cself and your 
probably using 21pF for topload, and then lumped together). What your not 
putting into the mixing bowl (and can't very well in Excel) are the affects 
in capacitance when the coil is in proximity with toroid at their positions 
above ground and the affects of current distribution at Fres (which becomes 
more uniform with addition of the toroid, but in this case, still has a way 
to go). Fantc would be the a place to take a look at current distribution 
(the VI profiles are an output of the Fres Analysis). I've also made a 
change a while back to output the VI profile length's to the input units 
(inches, meters, etc.), so it should make a little more sense to the user.

Here are a couple graphs using Fantc's VI profile data. I simply copy/paste 
the numbers to notepad, opened delimited in Excel, and plotted the graphs. 
Both Fantc and Javatc use VI profile data. I've chosen to leave out this VI 
output from Javatc as it is more inline with Fantc's analysis style as 
opposed to Javatc's design style.

The current is based on a 1 amp nominal value and are the mean values over 
the length of the corresponding element. These of course would need to be 
scaled to actual base current measurements.

http://www.classictesla-dot-com/temp/vi_loaded.gif

http://www.classictesla-dot-com/temp/vi_unloaded.gif

Take care,
Bart

>   Using a 4.5" secondary with 22.5" winding length, and a 4 x 20" toroid, 
> I get a loaded frequency of 184 khz.  Much lower than your 
> calculations.  I am sure my program is not nearly as accurate as Javatc, 
> but it usually works good for me.)
>
>With a 6.5" I.D. on the primary, I calculate 19.2 turns needed on the 
>primary.  Even if it is only 16.4 as Bart calculates, that is getting to 
>be a good sized primary.  For this reason, and that .005 ufd is just about 
>resonant with your transformer, I would suggest going to a larger 
>capacitor, like .01ufd.  The primary would then tune when tapped at 14.6 
>turns - according to my calculations.  I do agree with Bart on the 
>selection and design of the secondary.
>
>Ed Sonderman
>
>