HP53132A not holding calibration

[tkamiya]

I have an HP53132A that works very well except for one thing.  It won't hold calibration.

This is an older machine with firmware 3403.  OCXO is an older but original HP board.  What's weird about this board is there is no HP11801 socket but there are holes on board that will take different OCXO other than Isotemp that medium standard.  By tracing the line, I've determined it is compatible with MV89A.  I stuck one in, performed a cal, it cal'd correctly. 

In the man time, I have a splitter on house standard and an independent GPSDO that acted as a standard for calibration purpose.  They were fed to an oscilloscope.  When calibration ended, they were correctly in sync.  Traces are stable.  Nothing wrong.  Now, I press FREQUENCY 1.  Suddenly, it loses calibration and I can see traces on scope losing sync.

I suspected EEPROM has a dead spot at the exact location where OCXO EFC data are kept.  But calibration count is only 30.  While it is not impossible, it would be hard to imagine someone went through trouble to reset this.

Are there any LOGICAL reason how else this may happen? 

[TheSteve]

How do you know the OCXO is stable?

[tkamiya]

I don't.  But the fact that frequency jumps right at the moment FREQUNCY1 is pressed every time says stability isn't the issue.  If it were, it will be drifting while it finished calibration but still in CAL mode.

But I did think about it.  To satisfy my curiosity, I'll replace it with another one and see if the problem remains.  I happen to have ten OCXO burning in for over a month.

[tkamiya]

The result is inconclusive.  I'll have to wait til tomorrow morning to let the new OCXO completely warm up and acclimate in new environment before I can say conclusively.  But one thing appears to be certain.  The counter itself cannot save the frequency calibration parameter.  This is evidenced by sudden jump when getting it out of cal mode.

I took the OCXO and OCXO board to a second HP53132A at same firmware level, the drift doesn't happen the moment I get out of cal routine.  Cal parameter itself appears to be saved.  Drift is at much slower rate and irrespective of the mode. 

Both counters work equally fine with external reference. 

I have too many moving target to make certain though.  I'll probably step back and measure the OCXO independently before proceeding further. 

[tkamiya]

Today, the new 53132A and old OCXO board with NEW OCXO is behaving exactly like the first one.  Drift starts immediately after going to FREQUENCY 1 mode.  As long as I stay in CAL mode, it's good.

It is plausible that, due to the fact that only common item is the OCXO board, that is the defective component.  I was not able to measure VFC voltage and see any change.  It is likely in low millivolt to microvolt range change.  Noise basically overwhelmed any attempt at getting a useful measurement.  New board is en-route.

[tkamiya]

I just got a brand new (to me) board and brand new (to me) OCXO.  Installed it into a box that previously did not hold calibration.  It works.  It holds calibration.  So this is quite not logical way to approach this but the failure point has to be the board I was using.