[time-nuts] Improving the stability of crystal oscillators
John Ackermann N8UR
jra at febo.com
Wed Oct 10 14:11:34 EDT 2007
This is getting off topic (because it's far to practical!) but hanging
heaters on crystals is a time-honored activity for hams. Many of the
surplus FM rigs for VHF and UHF that we obtained from the likes of
Motorola, GE, RCA, etc. that were made B.S. (before synthesizers) had
ovens, or later "channel elements" that were actually TCXOs. This was
particularly necessary for UHF systems that operated in rugged
environments. But most ham-grade gear didn't have anything like this.
I spent quite a bit of time 15 years ago building a packet radio network
that linked Columbus, Dayton, and Cincinnati Ohio using UHF (440 MHz)
19.2 kbps data radios made by Kantronics, a company with limited RF
experience. The data transmission scheme we used required a frequency
tolerance of about +/- 2 kHz at 440 MHz (i.e., about 5x10e-6). The
standard crystals could almost maintain that tolerance in a nicely
regulated room, but we had these radios out in transmitter shacks in the
middle of corn fields, and with wide temperature swings the radio links
would die completely.
We first built a small proportional heater circuit that we could shove
in next to the crystal, and that worked OK. We finally found a source
for thermistors spot welded to a spring clip that would mount snugly
directly to the crystal case. With 12 volts applied, these heated the
crystals very nicely. With the heaters, we were able to keep the
systems running year round with about 1/10 the service calls.
The original crystals (designed for room temperature) worked OK and we
could warp them onto frequency even with the heaters installed, but we
noticed what seemed to be accelerated aging; after a couple of years,
the crystals could no longer be brought onto frequency. Later, we
ordered some crystals designed for higher temperature (I think it was
something like 50 degrees C) and those were both easier to trim, and
didn't seem to suffer the rapid aging.
That network is now dead and gone, victim of single-source components
(Kantronics stopped making or supporting the radios, and some of the
transistors became Unobtanium) as well as the death of traditional
packet radio in the face of the internet. Sigh...
Brooke Clarke said the following on 10/10/2007 01:52 PM:
> Hi Tom:
> That's a cleaver improvement to a non ovenized crystal. Usually crystals are
> ground to have a temp vs freq turnover point occur at a specified temperature.
> The common one is to be on frequency at room temperature. So if a room
> temperature crystal is heated it's no longer operating at the turnover
> temperature and so there is a steep delta frequency vs. temperature slope. Did
> you get a crystal cut for your heater temperature or somehow match the
> heaters temperature to a turnover point on your crystal?
> I spent many many hours adjusting the oven temperature of a Gibbs frequency
> standard to get it right on the turnover point.
> Since the improvement in frequency stability depends on minimizing the
> temperature excursions it's good to have as much gain as possible in the heater
> feedback circuitry. This was done for the HP E1938 10 MHz oscillator, see:
> and the paper: The Theory Of Zero Gradient Crystal Ovens, R.K. Karlquist, L.S.
> Cutler, E.M. Ingman, J.L. Johnson, T. Parisek
> Have Fun,
> Brooke Clarke, N6GCE
> http://www.prc68.com/I/WebCam2.shtml 24/7 Sky-Weather-Astronomy Cam
> Tom Clifton wrote:
>> ); SAEximRunCond expanded to false
>> Errors-To: time-nuts-bounces+brooke=pacific.net at febo.com RETRY
>> While the website is not explicitly about frequency
>> measurement, there is an interesting bit on improving
>> the stability of crystal oscillators with external
>> heater controllers. While it may not be suitable for
>> long term high stability control it might be suitable
>> for many other purposes.
>> The main page is also a good read:
More information about the time-nuts