[time-nuts] Re: GPSDO - EFC and fear

[Bob Camp]

Hi

There are multiple ways the OCXO in your GPSDO heads off frequency:

1) The crystal ages. This is normally the first assumption. It actually is the least likely.

2) The oven stops working. It either goes too hot (and everything inside fries) or it stops heating. 

3) The EFC biasing setup faults out. It might short to something. A resistor could come loose / crack. Lots of possibilities ….

Are there other things that can happen? Sure there are. The list above covers about 99.9% of the failures you might see. 

How to tell what’s what? 

If it’s crystal aging, it’s a slow process. This year you are at 70% of EFC range. A year later you are at 75%. In a few years after that, you hit the 90% limit. An error flag comes up in the status register. Typically the GPSDO still locks. 

If the oven melts, everything is dead. If it stops heating, you can spot that by a hand on the case. If the case is room temp, the oven is not heating. 

One clue that you have an EFC issue is that things happen pretty quickly. Last summer, your device was at 15%, now it won’t tune at all. 

So case 2 and case 3 *are* theoretically repairable:

1) Very carefully open up the can. Don’t get things to hot or break anything while doing it. 

2) Disassemble what’s inside. (again with some care)

3) Trace out the circuits for each assembly. (there aren’t going to be published schematics to follow).

4) Troubleshoot from that schematic.

5) Figure out what went wrong and (somehow) the value of that unmarked broken resistor / open transistor  … (or whatever).

6) If something significant is replaced, re-calibrate the internals of the OCXO. This may involve interfacing with the programable parts inside the device. 

7) Re-assemble everything.

So yes, it can be done. On a practical basis …. hmmm …. maybe not so much.

Fun !!

Bob

> On Jan 12, 2024, at 4:15 PM, ordnit via time-nuts <time-nuts at lists.febo.com> wrote:
> 
> Dear list,
> I own a couple of UCCM GPSDO which have been running flawlessy for a couple of years. On January the 4th I decided to switch them off for scheduled maintenance and noticed that one unit was not locking properly. Thanks to the log, I realized that the EFC was near its maximum and, an hour later, had exceeded its maximum.
> 
> I think that OCXO (Samsung STP2945LF Datecode 1245) has reached a drift level which is no longer recoverable by the internal DAC. I still have to desolder the OCXO and test it with counter + power supply, in order to see the amount of voltage required by the unit in order to reach (or approx 10MHz).
> 
> Is there a way to fix it? I think that using a bunch a opamps it should be possible to fix it, maybe degrading the performances.
> 
> In the junk kox I found an old trimble 10MHz OCXO which was removed by a dead GPSDO. So I powered it up ( 30+ hours) and mesured its output frequency with a HP50131A reading 9.999991 Hz (time gate 1s). By varying the Vref voltage (N6075A power supply), I managed to bring the OCXO back to 10MHz but... applying 5V. So I think that thix OCXO (dated back to 2015) has aged out too.
> 
> My questions:
> - Are my assumption reasonable?
> - Is there an OCXO which can replace the STP2495LF having the same Hz per Volt sensibility?
> 
> Thank you in advance for your help.
> 
> Have a nice 2024.
> 
> 
> adelmo
> _______________________________________________
> time-nuts mailing list -- time-nuts at lists.febo.com
> To unsubscribe send an email to time-nuts-leave at lists.febo.com