crystals have to meet only one criterion: when plugged into the receiver, they
must operate at the [design frequency] plus-or-minus tolerance. The total allowed
frequency "error" (tolerance) is determined by the design of the receiver
if filters, tolerances in oscillator components, frequency [in]accuracy of the
crystal itself, frequency drift of the crystal with temperature variations,
and aging of the receiver crystal -- plus errors in the center frequency of
the transmitter (also determined by a crystal with its own tolerances as for
the receiver crystal) -- and in the setup (tuning) of the transmitter's shift
Now, the transmitter crystal has an added requirement to that of the receiver
crystal. First, it is of much lower frequency than the channel frequency, its
signal being multiplied a number of times. But (very important!) the transmitter
crystal must also meet certain "pullability" requirements; i.e., the
frequency change per change in load capacitance. This is how the frequency shift
or FM is achieved in the transmitter. In the transmitter, the shift points are
individully set by means of two potentiometers or variable capacitors or a combination
It should be noted that these adjustments will be different for every crystal
unless this "pullability" spec is tightly controlled by the manufacturer,
which costs money.
From the above, it should be clear that you should always use the manufacturer's
crystal so that you will stay within your radio's design limits. Don't forget:
all these tolerances build up.
It is amazing that our radios work this well with all these tolerances, but
the precision of the quartz crystal helps us here -- you can typically expect
a max error in the order of between 3 and 10 ppm (parts per million), depending
on brand, which means that your frequency error will be between 3 and 10 times
"one millionth times the frequency." For 72 MHz, this would be a total
error of 360 Hz (for each crystal) when the tolerance is 5 ppm for each; total
frequency error can be as much as 720 Hz in that case.
It will be easily seen that, since tolerance is expressed in ppm, the real (absolute)
error at 72 MHz is about twice as great as the absolute error one sees at 35
MHz, with the same crystal specs.