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

Re: Spaced Coil measurements.



Hi Malcolm,

	I suspect you will find that the 172.6uH value for the inductance of your
large coil is correct.  Very small inductances are hard to measure due to
the small inductance and the sensitivity of the self capacitance to the
surroundings.  In order to measure these small inductances with accuracy, I
place a 100nF cap in parallel with the inductor and find the resonant
frequency.  Since I know the value of the capacitor to high accuracy, I
simply back calculate to find the inductance.  This is basically what you
were doing by adding your shunt capacitors but I would go for a much larger
value to insure that only the inductance was being seen.  Adding a large
shunt capacitance has the following effects:

1.	It will completely swamp out the effects of the coil's self capacitance.
 The small Cself will be far down in the noise compared to a 100nF shunt
capacitor.

2.	It will completely swamp out and RFI from radio and other sources.

3.	It will lower the resonant frequency far below where wire length or
transmission line effects would cause error (although that may not be a
problem anyway).

	I also wonder if when you add small (100pF) shunts that just the added
wires going to the coil are having a significant effect on the resonant
frequency by disturbing the fields around the coil.

	I have a secondary of about the same dimensions and you 17.8pF value for
Cself seems right.  That would tend to indicate that the voltage
distribution along you space wound coil has a sine distribution along its
length.  As long as that is true, Medhurst's equation should hold valid.
In non linear wound coils, the voltage distribution is not a sine function
and Medhurst's equation fails to give the correct values.

	I suspect that wheeler's equation is giving an error of -14% as you
suspected.  It may have a problem with highly spaced coils like yours.

	In your follow up post, it appeared that the wire length again did not
seem to influence the resonant frequency.  However, do try to look for
peaks or other indications around 2.9 MHz of the wire length effects.  Of
course the velocity factor may not be 1.000 either so it gets messy.

	Perhaps you could start with a straight wire along a long form.  Then wind
it one turn, two turns, etc. using the same long wire to look at the
effects.  The long wire should look like any 1/4 wave antenna.  Slowly
winding the wire step by step, you may be able to see where the effects of
inner turn inductance take over.  Or even more strange, perhaps the
resonant frequency of a straight wire is governed more by Cself (and less
by it's length) than we suspect.

Neat work!

	Terry




At 09:44 AM 6/1/99 +1200, you wrote:
>Hi all,  
>        As promised, here are the measurements (and they do contain a 
>surprise):
>
>The Coil:   Hs = 31.5",  Ds = 12.3",  Ns = 38 turns  Dwire = 0.8mm
>            Winding pitch = 1.2 TPI,  former = thinwall PVC tubing
>
>According to Wheeler, Ls = 147.5uH
>    "      " Medhurst, Cs = 17.8pF (or thereabouts - I estimated "H")
>
>Measurement of L was rendered impossible by the stray electrical 
>signals flying round, mostly from local radio transmitters :(  I was 
>bitterly disappointed by this and will figure out some other way to 
>get a direct reading.
>      
>      However, if one takes Medhurst's Cself as gospel and adds that 
>to the shunt capacitances listed below, one gets a remarkable result:
>
>   Bare Coil,    w. 100pF shunt    w. 228pF shunt    w. 336pF shunt
>
>Fr    2.9MHz          1.113MHz         773kHz           644kHz
>Ls    169.2uH         173.6uH          172.5uH          172.6uH
>
>Now that I hadn't expected. Possible interpretations:
>
>- Medhurst works and lumped rules hold good for this coil BUT Wheeler 
>  fails
>- Suppose Wheeler is correct. Then Medhurst fails to give the 
>correct figure for Cself (if this were true, reading #1 suggests 
>Cself = 20.4pF, reading #4 suggests Cself = 78pF). This contradicts 
>badly.
>
>Until Ls can be measured directly and definitiively, no firm result 
>but something clearly doesn't work properly for this case.
>
>Malcolm
>