Figure 2
Not only do we like to look at nicely refinished old radios; many of us like to listen to them as well. We here in Moorpark California love to tune in the BBC, Australia, or Tokyo early in the morning to hear the world news. Unfortunately, these old superheterdynes often drift significantly as they warm up, necessitating a tricky adjustment of the dial. Newer models from the 50s and 60s can also experience drift, a real problem when listening to CW or a SSB station. My old tube style Swan 500, one of the first truly successful SSB transceivers for radio amateurs, began to experience drift and instability, not only during the warm up period, but well after its temperature had stabilized. The problem was a temperature compensating capacitor in the VFO.
The purpose of this article is to help those who desire better temperature compensation within an old radio. We offer two options: approximate compensation for AM listening, and exact compensation for CW or SSB listening, and for transmitting.
Do You Really Have a Problem?
To determine whether you have a problem, measure the frequency drift
experienced by the dial
of your radio. For example, tune in WWV when the radio is cold, and let
it warm up for about 1.5 hours. Then re-tune again to record the amount
of frequency drift. If the drift is more than about 500 Hz, your radio
has a basic drift problem. Now, tune in a stable carrier and put on the
BFO to obtain a low pitched tone. If the tone indicates instability of
any kind, your radio has a stability problem that could ruin CW and SSB
operation.
Here is a word of caution. If an old radio has been unused for a long
time, leave it on for many
hours before assessing the quality of its VFO. My Collins R390, for
example, sat unused for about a year. When first put into operation, it
would randomly change frequency by about 500 Hz, every few seconds.
After about a week, however, the problem disappeared entirely.
A similar instability occurred in the VFO of my Heathkit DX100, which
had been in a chicken
coop unused for years. Unfortunately, the DX100 did not fix itself. Its
main problem was traced to the temperature compensating capacitors.
Another type of instability is a random wavering drift of a few tens
of hertz that never goes
away. Such drifts are especially irritating in CW communications. They
often are the result of poor quality temperature compensating capacitors
in the VFO.
The best way to really determine if the temperature compensating
capacitor is defective is to
carefully open the VFO, and to temporarily replace the appropriate
temperature compensating capacitor with new silver mica of the same
value. If the chirping or wavering instability vanishes, you have
located the problem. Of course, the powers of compensation within a
silver mica are limited, so expect a slow drift for the first hour or so,
usually to a lower frequency as various inductors and capacitors increase
in value slightly.
Instead of just guessing the amount of drift from the dial as above,
you might borrow a
frequency counter, or a digital receiver that is known to be completely
free of drift. It should be easy enough currently to measure the
frequency of the subject VFO to an accuracy of 1 Hz.
Sources of Temperature Compensating Capacitors
The VFO is a critical circuit in any radio, while the temperature
compensating capacitor is
critical to correct VFO operation. Yet, it is not easy to buy such
capacitors off the shelf. Currently, the author knows of only two
possibilities: XICON CD Series Class 1 temperature compensating
capacitors, available from Mouser Electronics. In particular, the
temperature coefficient SL is available, specified to be N330 ?500
ppm/°C. The coefficient ?330 parts per million, or ?0.000330 means
?0.033 %/°C. A variation between units of up to 500 ppm indicates that
these units are not very accurate. My tests found that the SL type can
have a slight waver in the capacitance value, meaning a random noise of a
few 10s of hertz in the frequency of oscillation. They are fine for AM,
although not for CW. XICON has better capacitors, although not in small
quantities.
A better type is the Dogbone Ceramic, available from Surplus Sales of
Nebraska. The author
found these quite stable, and available in several values with
appropriate ratings. Another common type of capacitor is the NP0,
meaning negative and positive coefficient is zero, available almost
anywhere today.
Important Technicalities
If you are lucky, you will have a schematic with a parts list for your
radio. Moreover, the radio
might have existing temperature compensation capacitors in its design.
Simply replacing one of these capacitors is impractical: (1) It is
generally impossible to buy the values specified in the original parts
list, and (2) over the years, the required compensation might have
changed. It is better to design the amount of compensation to a correct
value. In particular, since the resulting temperature coefficient is
always reduced by additional circuitry, it is better to begin with more
temperature compensation than required. The circuit in Figure 1 serves
to reduce the resulting coefficient.
Figure 1 Fine Tuning a Temperature Compensating Capacitor
C1 is the temperature compensating capacitor, for example, 24 pF N220.
C2 represents either a discrete capacitor, or a variable that you may
add, for example, 440 pF NP0. C3 represents an existing padder, or it is
the main tuning variable that will restore the total capacitance, and
therefore the frequency of oscillation to its specified value. Usually
we do not know the value of C3. However, we might know the required
value of the series parallel combination of the three capacitors, defined
to be Csp. In the case of the Swan 500, it is 22 pF because the dial is
calibrated when Csp ? 22 pF.
What matters most to us is the net temperature coefficient of Csp,
referred to as Nsp. Using
the Swan example and my web calculator, it is N215. Refer to
http://www.csun.edu/~hceen006/ahome/in4.html
The equations that do the calculation are available on the web page, with
inputs of C1, N1, C2, N2 and Csp.
Case 1
Approximate compensation with C2 a discrete value
The required temperature compensation can be determined by plotting
points on the
coordinates in Figure 2. If you can assume a straight line, only two
points are needed near the ideal of 0 fF/°C (1 fF = 0.001 pF).
Figure 2 Determining temperature compensating capacitor values
Begin with a fF/C value that is slightly larger than required. The fF/C value of Csp characterizes the capacitor; it is merely the product of the temperature compensating capacitance multiplied by the coefficient, for example, 24 pF x -220/106 per degree C is -0.00528 pF/C, or -5.28 fF/C. This is going to be an attempt to teach by the example. The Swan 500 originally called for 22 pF N220. As you see in the graph, 24 pF N220 (C2 bypassed) is results in overcompensation. Therefore, using C1 = 24 pF N220, we added in series C2 = 220 pF NP0. From the web, with Csp specified to be 22 pF, the coefficient goes down to N195. This amounts to -4.29 fF/C. After installation of the 220 pF, the drift became ?2 kHz, undercompensated. As a result, going to C2 = 440 NP0 is better. If your radio has no compensation, you can begin with a point on the vertical axis where fF/C is zero. You will have to try a value for C1 (with C2 bypassed or very large), for example, 10 pF N750 to obtain another point on the graph. Then, you can easily design a correct value.
Case 2
Exact compensation with C2 a Variable
Figure 3 indicates that by using 24 pF N220, the range of C2 is rather
large at about 500 pF.
We could choose C1 to be significantly more than nominal, for example, 27
pF N450. Figure 4 shows that C2 becomes practical at about 50 pF. If
there is room in the VFO chamber, a variable C2 is a great idea. For
example, the Valiant VFO contains a spare 11 meter variable that could
serve quite well for temperature compensation.
Placement Considerations
We find that the capacitors work best if placed near the metal
chassis, since all components in
the VFO volume need to change temperatures at about the same rates. Heat
enters the capacitor mainly by radiation from the metal chassis. If
heating is uneven, the frequency will deviate excessively before
achieving its correct value. Of course, the final configuration should
have short leads so that the VFO does not respond much to mechanical
vibrations. All mechanical parts should be snug in the VFO area to help
avoid sudden changes in their positions, and therefore in the VFO
frequency as the unit warms up.
Conclusion
Temperature compensation is involved, although well worth the effort.
It only has to be done
once. Even if the compensation is imperfect, the radio warms up faster
with less drift every time you turn the radio on.
References 1) October 1996 Electric Radio, "Recompensating Old Oscillators to Minimize Drift," by John "Robert" Burger WB6VMI
Click to see the last two drawings as RTF files, viewable in MS-WORD and other
Word Processor programs.
Figure 3 Assumes Co = 24 pF N220 & that
C1 and C2 adj to give 22 pF.
Figure 4 Assumes Co = 27 pF N450 & that C1 and C2 adj to give 22 pF.