[time-nuts] Nortel GPSDO osc age alarm

Fri Jan 24 06:08:21 EST 2014

Hi Volker

One possibility, as I found with one of these recently, is that your  
oscillator has aged such that the required EFC voltage for 10MHz output is now  
outside the range provided by the board.

The EFC behaviour can be tracked if Lady Heather is  enabled from switch on 
and the DAC voltage monitored.

The EFC voltage should start at 3 volts, in my case reported by LH as  
3.000002 Volts, and will sit at this level until enough satellites are being  
tracked for the board to decide it can start the conditioning process, at 
which  point the DAC voltage will ramp upwards over the course of a few seconds 
until  the point is reached where the oscillator frequency crosses 10MHz and 
the  control loop takes over.

All standard stuff of course but, as you've discovered, if the  oscillator 
hasn't reached 10MHz by the time the EFC voltage reaches, in my  case 
anyway, approximately 5.6 volts the board's "OSC" report switches to  "BAD" and 
"Normal OSC age" switches to "OSC age alarm" and LH highlights  both in red.
The EFC voltage finishes its ramp at 6 Volts, reported as 6.000004  Volts, 
and then just sits there, all exactly as you're seeing.

Having removed my "faulty" oscillator, something that needs to be  done 
VERY carefully to avoid damage to the board, I found that it  required an EFC 
voltage of just under 6.6 Volts for the the output frequency to  reach 10MHz.

The seller, fluke.l, was very helpful as always and offered to  supply a 
replacement 34310-T oscillator, this one did have it's required  EFC voltage 
in the correct range and resolved the problem.

However, whilst waiting for the replacement to arrive I left the  original 
oscillator on test and monitoring it for a  while suggested that the only 
thing "wrong" with it was the required EFC  voltage, so I reconnected it to 
the board using a wired  lash-up on the bench with a simple 2 Volt level 
shifter inserted into the  control loop and did indeed get a locked condition 
with Lady H reporting the DAC  voltage from the board as close to 4.5 volts.

This suggests that whilst the board  design requires the oscillator EFC 
Voltage at 10 MHz  to be between 3 and 6 Volts an oscillator that falls outside 
this  range shouldn't automaticall be assumed to be "faulty" in more 
general terms,  although it's obviously getting a bit long in the tooth and that 
doesn't  help much if it happens to be soldered into your circuit board !

Whilst it is possible to remove and replace the original  oscillator there 
is an alternative to physical replacement, one which  I've now implemented, 
and this uses the mounting positions already  available beneath the 
oscillator, not usable though until it's  removed, to fit a couple of SMA or SMB 
connectors.
These two connectors couple the 10MHz signal from the oscillator into  the 
board (J9) and the EFC Voltage out to the oscillator (J10).
Although there is a  regulated supply available from the  oscillator, which 
might be expected to supply the EFC  circuitry, in practice this connection 
does not seem to be  required.
I have not investigated further as yet to determine whether the board  auto 
senses and uses this supply if it is present, or whether it always just  
ignores it, but it doesn't seem to be an issue either way and it's interesting 
 to note, at the extremes anyway, that LH does report the DAC voltage to be 
 the same in both instances.
The oscillator can be powered from the board or given its own supply,  in 
which case the only required connections between them are the two coax  leads.

Using this arrangement, and some variation of a positive or negative  level 
shifter if required, it becomes possible to use different  oscillators with 
the Nortel board and it does become an even  more interesting toy.

One limitation though when using the Nortel boards in this way is  that 
they don't seem to share the versatility of the Thunderbolt when it comes  to 
modifying the oscillator conditioning parameters.
It's suggested in the LH documentation that such commands are currently  
"undocumented" but that also leaves the possibility that any such  change 
might only be a firmware option.
If anyone has further information on this that would be much  appreciated.

I've been sufficiently impressed with the modified unit that I'm seriously  
considering modifying another one, even though it doesn't actually "need"  
it, although not over enthusiastic about repeating  the oscillator removal.
However, an added bonus with this configuration is that it's also more  
forgiving when it comes to any damage caused during removal of the original  
oscillator, since the pads for the external connectors are  separate from the 
internal oscillator pads. Obviously it needs to be  ensured that any 
necessary continuity is maintained but that could be  easier if a board mounted 
oscillator is no longer required.

The later single board unit, the NTBW50AA, has a similar external connector 
 arrangement but in this case the connector pads are available without 
removing  the original oscillator.
This hints at the possibility that perhaps a simple track interruption  
might allow fitting of an external oscillator with the original still  onboard 
but examination suggests that the onboard connections  route first to the 
internal oscillator so that would probably still need to  be removed.

Regards

Nigel
GM8PZR

In a message dated 23/01/2014 23:59:34 GMT Standard Time,  
ailer2 at t-online.de writes:

Hi!

I bought a Trimble/Nortel  GPSDO

http://www.ebay.de/itm/300933951405?ssPageName=STRK:MEWNX:IT&_trksid=p3984.m
1497.l2648

and  Lady Heather's now tells me that everything is alright - except
- DAC  6.000004V
- OSC BAD
- osc age alarm

The rectangle "10MHz" output  signal (J5) shows a signal at about 9.8MHz
(a deviation of about 200kHz),  wobbling 4Hz up and down. Oddly enough,
the direct oscillator output (J4)  shows a sine wave at a stable
10.0000004MHz (a deviation of  0.4Hz).

It seems to me it's not the oscillator that is bad but the  servo loop -
what can I do?

Thank  you

Volker
_______________________________________________
time-nuts  mailing list -- time-nuts at febo.com
To unsubscribe, go to  
https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the  instructions there.