Keithley 775A Frequency Counter TCXO Repair 

Thanks for the suggestion. I would probably eventually replace it with another TCXO or a low-powered OCXO (since the onboard PS is pretty weak)...

The power supply on the 775A is pretty weak though, so probably wouldn't provide enough juice for the initial current needed by the OCXO...

Agreed that modding a TCXO like this will likely break the TC part. It could be that the crystal has been damaged by shock or something, but my first guess was maybe the temperature sensor wasn't working right. Still, 2 kHz off is a huge error. I've had very good luck using eBay TCXOs in the DIP-14 4 pin package, all the ones I've ordered were well within spec and seem to be stable long and short term. For occasional use equipment like this I wouldn't bother doing a OCXO, a TCXO will be much more accurate and convenient for a device that's moved around a lot and not left alone and powered all the time. A big equipment rack like that should have a system 10 MHz that's used when readings must be accurate and consistent between gear.

I agree, it is better as nothing but this way you degrade it. Besides that, it is so far of there must be some other problem.

For any precision instrument the moment you open it you technically "ruined" the calibration and added some unknown factor to the characteristics. The main point of this video is that a crystal oscillator's oscillation frequency can be fine tuned via external capacitors. Yes, most likely the temperature compensation wouldn't work properly as it was tailored to the original crystal's curve.

@@KerryWongBlog Only today I discover this video. I agree with other gentlemen who stated that you did not repaired the TCXO. Sorry to be rude but this is the truth!! Adding a capacitor to the xtal modifies the Q factor. You tuned the frequency widening the bell shaped response, let me say worsening the sharpness of resonance curve. Furtheremore, when you tune variable capacitor it is better to use a PLASTIC screwdriver not a metal one...Do not be offended, a sincere criticism is better than a false praise. Regards.

A further demonstration is that adding a variable capacitor you compensate 1800Hz. On the contrary, the TCXO manufacturer allows only 4.7 Hz...this is not possible, you see. Another proof could be obtained from a Phase Noise measurement. You will see a big degradation on the performance due to a lower Q. But measuring xtal oscillator Phase Noise requires very expensive equipment.