[time-nuts] -hp- 10811 repair (Thermistor)

[Rick Karlquist]

Bruce Griffiths wrote:
>>
> Dan
>
> The real reasons that this part is labelled non replaceable are
>
> 1) its epoxied to the oven.
>
> 2) Setting R20 to the correct value actually requires plotting the OCXO
> frequency vs R20 and selecting R20 so that the OCXO frequency is located
> at the stationary point on the frequency vs R20 value plot. This
> requires a high resolution frequency measurement setup with a highly
> stable reference and takes considerable time as one has to allow the
> oven temperature to settle between adjustments to R20. Drift due to
> aging and ambient temperature changes limit the accuracy with which the
> OCXO turnover temperature can be achieved.
>
> Bruce

That's news to me.  AFAIK, the individual crystal has a known oven
set point that is determined in the crystal fab.  R20 is selected
to achieve this set point.  What Dan did is exactly right, IMHO.
BTW, many 10811 crystals did not have a turnover temperature as
you imply.  They simply had a point of minimum (not zero) tempco
where there was an inflection point (2nd derivative is zero)
but 1st derivative is not zero.  This is well established SC
cut theory ("true" SC cuts, that is).  The stability of the oven
set point was quite adequate in terms of the overall 10811 error
budget.  Most the tempco is due to the electronics pulling the
crystal, as detailed in my paper on the E1938 oscillator.

When you install the new thermistor, try to duplicate the lead
dress of the old one.  That was supposed to mitigate against
heat running up the leads and corrupting the thermistor.

Rick Karlquist, N6RK
R&D engineer, HP Santa Clara Division 1979-1998