Hello all,

I did some examinations with old air-variable-capacitors
from old (tubes) radio's.

Tnx's Jerry, K0CQ @ W0AK.#CIA.IA.USA
for info about Radiotron Tube Designer's Handbook.
Unfortunately i do not have acces to that book or cd-rom.

I foundout that the angle of rotation can be be:

Capacity-linear
Wavelength-linear
Frequence-linear
Log-Linear

C=a1*phi + b1          capacity linear
C=(a2*phi + b2)^2      wave     liner 
C=1/(a3*phi + b3)^2    freq     linear
C=a4*e^b4*phi          log      linear

The shape of the turning plates can be quite different.

In principe a1 = the shape of curve form factor and b1 the zero capacitance.
phi is the rotate angle.

At the low capacity edge of the 180 degrees turning the curve is not
following exact the formula because of the zero capacitance.

Most radio C's I found where wavelength-linear for old MW C's.

Many C's for FM rx where frequence-linear in practice.

While mica c's from (old) small battery 6 transistor radios where
almost always capacity linear.

I am searching to find out the a and b factors so i can reconstruct the
C= formula from measured data with angle and capacity.

The C has to be recalced to frequency C=(2^2*pi^2)^-1/(F^2*L) 
or C =.0253303/(F^2*L)


I have made a big MW-loop(1 meter) for MW with an ordinairy capacitor 2 x 500pF
found in junk box.

I did mount to it a 200 steps stepper engine from old HD equipent with 
teethweels to turn it remote by an interface connected to a LPT port
and my Yaesu FT840 connected via cat control rs232 so i recieve the
frequence for recalculating to steps to turn teh capacitor to.


Now i wrote some software to adjust the Medium Wave magloop c with 
just 2 set points already quite good.

I use the linear interpolating methode with about 30 calibration points.
With a so called wavelength linear capacitor.

I think 540KHz at 0 degrees and 1600KHz at 180 degrees.
In stepper positions it looks like 0 steps for 540KHz and 620 steps for
1600KHz.

For the wavelength-linear C it should be easy. 
Just recalc the frequence to wavelenght (f=lightspeed/wavelength) from the 
2 calibrated points and then linear interpolate.

Example of my MW-tabel-list.
----------------------------
Versie 1.07f 2
 0             620   'steps 0 to 620
 408           24    'freq  KHz    steps 
 540           182
 621           237
 648           252
 747           304
 1008          398
 1215          461
 1272          474
 1296          489
 1512          525
 1584          538
 1645          549
 1900          597
-----------------------------

For the frequence-linear C its just linear interpolating.
But using a special table with much calibration points.

I have written a small program to calc the tabel from just 
freq-low and  steps-low
freq-high and steps-high
Cmax example 120pF 

You may also enter just 1 for C because it doesn't
affect the angle versus frequence only the calculated L value will differ.

Example run of my program.
------------------------------------------
Filename: 80loop1.txt
pi= 3.141592653589793
Input Steps low 0
Low Freq= 3304 kHz
Input Steps high 545
High Freq= 6083 kHz
Input Cmax= 120 pF
Steps between calculations 50
Cmin= 35.4018279329482 pF
Delta C= .1552260037927556 pF
L= 19.3 uH
------------------------------------------
Freq= 3304 kHz              Steps= 0
Freq= 3416 kHz              Steps= 50
Freq= 3541 kHz              Steps= 100
Freq= 3680 kHz              Steps= 150
Freq= 3837 kHz              Steps= 200
Freq= 4017 kHz              Steps= 250
Freq= 4224 kHz              Steps= 300
Freq= 4466 kHz              Steps= 350
Freq= 4756 kHz              Steps= 400
Freq= 5111 kHz              Steps= 450
Freq= 5559 kHz              Steps= 500
Freq= 6083 kHz              Steps= 545
'-----------------------------------------

It's amazing thats thats fits quite good already.


I measured with a scoop at the Loop couple winding the output and 
1 winding very lose coupled where i feed the loop with an RF signal generator.
Then set the C on max capacity and look for max amplitude on scoop and read
the freq and steps-low also so for C at Cmin and steps-high.

I tried to show in ascii how the shapes of vanes of variable capicitors are
as used in many old radio's.

Liniar vanes (x=center knob axe).
Half a circel symmetrical around the axe.
C=a1*phi + b1
                     - 
                   /   \
                 /       \
----------------/----x----\---

Liniar Wavelength vanes. 
Half a circle asymmetrical around the axe.
C=(a2*phi + b2)^2

                   -
                 /   \
               /       \
              /         \
---------------------x---------

Log linear vanes.
Asymettrical and flattend top. 
C=a4*e^b4*phi  

             ^^^^^^^^
           ^          \
         /              \
---------------------x-------

Frequence liniar vanes
Asymmetrical and very long stretched left end.
C=1/(a3*phi + b3)^2


   ^ ^^^^^^^^^^^^^
 ^                  \
/                      \
---------------------x-------


Okay i hope this is clear enough to find out what kind of variable C
you have in your junk box hi......

My Stapjes.exe and Calc_c.exe program are written in QuickBasic
in Dutch and if intrested you can get the source or exe.


73's from  Lodewijk 
PA3BNX@PI8SHB

 