[time-nuts] Pulling oscillators
Murray.Greenman at rakon.com
Sun Mar 6 13:31:02 EST 2011
You haven't explained how much you want to pull the oscillator, but assuming it's less than 1ppm, what you suggest is quite viable. Replace one of the caps in the Pierce oscillator PI network (preferably the output one) with an NPO one of slightly smaller value and add across it a varactor. If your intention is to externally lock to (say) GPS, the smaller the range you can get away with, the better.
How much range you allow depends on the intended application, including how much thermal compensation is required (defined by crystal cut angle and therefore pullability) and how much ageing compensation is to be expected.
You can do some pretty subtle things for very small adjustment range. In one design I employed the small change in oscillator chip input capacitance with supply voltage - I removed the zener on it's power supply and drove it from a D-A converter. In another I heated the crystal with PWM from a phase detector as a means of holding the frequency (operated the AT cut crystal on the slope of its Bechmann curve at around 40°C).
In an old SSB transceiver I locked the 7.3333MHz reference (similar oscillator to the one you describe) to GPS by phase comparison every third second (as though the reference was 1/3 Hz). The micro doing the locking simply ignore two out of three phase measurements.
In short, yes, it can be done.
More information about the time-nuts