[time-nuts] Ancient OCXO in scope calibrator.

[Bruce Griffiths]

David C. Partridge wrote:
> I'm having some problems with an ancient Bulova OCXO in a Tektronix 184
> 'scope calibrator.   
>
> This is an crystal mounted on an octal valve base with a heater winding
> wrapped round it, and a common or garden variety thermostat which is
> supposed to switch at 75 celsius.   I've not measured the switching
> temperature, as I don't have the means but the outer case does get fairly
> warm (40 plus degrees?).   The oven is turning on and off.
>
> It is used in a circuit with a 7587 Nuvistor tetrode (yes it's a valve/tube
> circuit).  The crystal is connected cathode to G1 with a trimmer capacitor
> of 3-12pF.   The signal at the cathode is supposed to be about 70V p-p.
>
> I'm measuring the frequency of the signal after the transformer stage which
> couples it to the first stage of a countdown board.
>
> For the first minute or so after turn on from cold, it sits below 10MHz and
> is fairly stable and climbing as the oven warms up, and I can adjust the
> frequency up towards 10MHz with the capacitor (but not all the way), then
> suddenly, at the stage where it is starting to look as if it will soon
> stabilise at about the right frequency, it jumps to way over 10MHz and the
> lowest I can get it down to with the capacitor is about 10.0003xxx MHz where
> xxx is not very stable at all - in fact it can vary up to to 10.0005xxx and
> down to 10.0002xxx.
>
> I've tried freeze spray on most of the components round there to no effect.
>
> If I try to probe the signal at the cathode of the nuvistor even with a high
> impedance active probe with a P6201 with a 100x attenuator (about 1pF
> loading IIRC) the oscillation just drops dead.
>
> Now for calibrating 'scopes, it doesn't need to be any more accurate than it
> is (30ppm) - but ...
>
> Do any of the collected mavens have an explanation for the behaviour, and
> recommendations for fixing things?
>
> Cheers
> Dave
>
>   
Dave

Need a few more circuit details.
The crystal current wont be very high unless their are other components 
connected to G1.

Bruce