EMRFD Message Archive 15371

Message Date From Subject
15371 2019-08-14 11:54:04 Leinweber, Glen crystal oscillator tempco
Pulled out the standard circuit G3UUR crystal-characterizing oscillator to
measure some old 10MHz HC-49 pulled from very old hard drive interfaces.

I had built G3UUR on a small PCB, 2N3904's. Used two 470pf polystyrene
capacitors between base-emitter, and emitter-gnd.
Stuck in my crystal, and left it on overnight. Seemed that frequency was
remarkably stable while outside temps dropped from 29C to 19C.

Curious, seemed far more stable than a cheap canned Xtal osc.

Those polystyrenes supposedly have about -150ppm/C tempco. Could
this G3UUR be fortuitously temperature-compensated or nearly so?

So made a crude oven (far more crude than one built by Wes in the fog
of times past). A tin can wrapped in foam packing material except for
its bottom end, in a pot of water hot-to-cold. An old mercury thermometer
to measure temperature.
Starting with hot water, took it down from 45C to 10C, probably too fast.
Through that range, frequency varied 10 Hertz (nominal 9997228).
One turnover temp was found at 35C. Another was approx. 13C.
That's one stable oscillator: could use more negative tempco.

Checked a large parts supplier. No polystyrenes. Any caps with negative
tempco? Only 50kV and $$$. All others marked "obsolete", "no stock". Sigh.
Harbour any you can find.
15372 2019-08-14 13:42:25 Jim Kortge Re: crystal oscillator tempco
15376 2019-08-24 08:41:02 w0gk Re: crystal oscillator tempco
Re: NTC caps, I've bought polystyrene caps from West Florida Components.

Digikey has some NTC ceramics in useful values. Use their component search feature to find them. You may need to figure out the EIA letter code equivalents for N150, N220, etc.

Also, Dan's used to have some NTC ceramic trimmers. These are useful for dialing in temperature compensation. Don't know if they are still available. As I recall, they weren't advertised as NTC, but were marked "N300" and seem to behave as NTC.

The problem I've run into with polystyrene caps is that the available values are often too big to be useful for temperature compensation. Generally, values less than than 100pF are what I've needed.

You mentioned Wes, W7ZOI. His article in QST a number of years ago is a great resource for tweaking oscillator temperature compensation in an organized, orderly manner. From your mention, I assume you've seen that article, if not it's worth tracking down.