[time-nuts] GPSDO won't lock: OCXO aged out?

[Scott Newell]

TL, DR: Trimble 73090 5V OCXO won't tune to 10 MHz (9.999 989 MHz at 
0 VDC EFC, 9.999 997 MHz at 4.88 VDC EFC). Worth fixing? Maybe an 
internal mechanical trim?


I've got one of those China special GPSDOs (thanks again, Paul!) 
built around a surplus Trimble UCCM 57964 (or 57963?) board. PPS 
driving NTP went wonky the other day--turns out the GPSDO had lost 
lock. Due to WFH I didn't have LH running, so I can't see what the 
DAC was at before it unlocked. I did examine the Trimble logs after 
the fact and noticed this:

L73:20/12/22.19:05:24 Phase over 10 microsecond
L74:20/12/22.19:16:39 Phase over 100 microsecond
L75:20/12/22.20:32:11 Phase over 1 millisecond

Late December is when I noticed the problem, so the log looks reasonable.

I now have my hands on it and can test. DC supply current and 10 MHz 
output from cold shows that the oven is working and the output 
frequency rises during warm-up (9.999 668 MHZ cold, 9.999 993 MHz 
hot), so SC cut rock, right? (Tbolt 10 MHz used as counter ref.)

With no antenna connected, LH lets me fiddle the DAC voltage:
0 DAC = -100% = 9.999 989 MHz = 0 VDC EFC
524287 DAC = 0% = 9.999 994 MHz = 2.50 VDC EFC
1048575 DAC = 100% = 9.999 997 MHz = 4.88 VDC EFC

I thought these 5V GPSDOs were using +/- 5 volt tuning, but that's 
not what I'm seeing. Not that it would help this unit... Tuning range 
from 0-5 V is about 8 Hz, or +/- 0.4 PPM. Does that seem reasonable?

UCCM-P >*IDN?
TRIMBLE,57964-80,60498844,V2.0.1.6-01

Datecode on the OCXO is 0948. Paul gave it to me back in March of 
2016, and it's run pretty much continuously since then. LH shows 
phase error accumulating, which makes sense for an off-frequency clock.

Thought or suggestions?

thanks!
newell  N5TNL