RE: More Coupling...

["Tesla list" <tesla-at-pupman-dot-com>]

Original poster: "John H. Couture by way of Terry Fritz <twftesla-at-uswest-dot-net>" <couturejh-at-worldnet.att-dot-net>



-----Original Message-----
From: Tesla list [mailto:tesla-at-pupman-dot-com]
Sent: Sunday, December 24, 2000 5:18 AM
To: tesla-at-pupman-dot-com
Subject: Re: More Coupling...


Original poster: "Barton B. Anderson by way of Terry Fritz
<twftesla-at-uswest-dot-net>" <tesla123-at-pacbell-dot-net>

Hi Paul,

Tomorrow I'll take measurements for any background levels. The next K
measurements I do will be concentric amp/volt readings.

------------------------snip


I have found this "most important" when working with acmi:
"Coils measured inductances must be input correctly into acmi via dimensions
for accuracy". Since acmi uses dimensions, any coilers rough or even
relatively
precise measurements are not enough. There is human error in building and
oversites. These flow into the program if not modeled first in a TC program.
Entering coil data into a TC modeling program (JavaTC, JHCTES, Ed
Sonderman's
spreadsheet, etc..) allows the user to make slight adjustments to duplicate
measured inductances. When this is peformed, these "fine tuned" dimensions
can
then be entered into acmi. Acmi then calcs almost identical Ls and Lp values
and also becomes "sweetly" accurate.

--------------------------

Bart -

Coilers assume that measured values are "sweetly" accurate but that is not
correct. All measured values have a probable error. The problem is how do
you determine that error? It is obvious that inputing incorrect values for
the computer inputs will make the computer outputs incorrect. In the past I
have found that before making any tests it is an advantage to find the
approximate value for the measurement by computer or by calcs. I then make
several tests for the same parameter. I also use more than one test method
for the same parameter. This usually guarantees a close agreement in values
which also gives me some number differences from which I can devise an
approximate probable error.

Fortunately there is more than one test method for the coupling. Have you
compared the mutual test for finding the coupling with the coupling tests
mentioned on the List?

You may also want to try the mutual inductance test as shown in my TC Design
Manual. This is the same test as you are now doing except that it is more
accurate. A variac and lamp or resistor can be used to set the current. Non
linearity is not a factor because only one current value is used. The
accuracy then is dependent only on the digital voltmeter which is inherently
very good.

The current is adjusted to 2.653 amps and the mutual inductance in
microhenries is read directly off the digital meter. The meter is actually
showing millivolts. No calculations are required.

John Couture

---------------------------

Great job with the gradiant graphs. They open up some interesting topics of
discussion and future tests.

Take care,
Bart