[time-nuts] My NTGS50AA failed

Thu Nov 13 06:03:55 EST 2014

Hi Ignacio

I'm sure removing the oscillator would make your faultfinding much more  
straightforward.

When I removed mine I started out quite convinced I wouldn't be able to do  
it, even with the solder mostly removed from around the pins they were 
obviously  quite a tight fit and all I had available at that time was a plunger 
type  solder sucker and desoldering braid, as the bits on my old Pace 
desoldering kit  were well past their best. 
The answer basically was lots of braid, lots of patience, and resisting  
like crazy any temptation to pull against a hole that wasn't fully released, 
but  I certainly wouldn't want to do it that way again in a hurry.
The desoldering gun I use now would make it easier but the holes are really 
 a bit small for those oscillator pins.

The good news though is that the connectors are there as a back up and  
fitting them definitely makes it more versatile.
I had considered repeating the performance on another NTGS50AA  and/or 
NTBW50AA, I really think I should but so far haven't found the  courage:-)

I think previous checks for 1PPS outputs might have been limited to the  
external connectors but it would make sense if it was available somewhere on 
the  PCB, I'll try to get one powered up later and check your findings, if I 
can find  a bit of space that us amongst the usual chaos!

Regards

Nigel
GM8PZR

In a message dated 13/11/2014 01:11:30 GMT Standard Time,  
eb4apl at cembreros.jazztel.es writes:

Hi  Nigel,

Thank you for the suggestions, I was trying to avoid the OCXO  removal but 
I think that now it must be done. 
I was also playing with the  idea of populating the connectors so an OCXO 
exchange could be easily made,  this makes possible to try various 
oscillators.
Meanwhile I has been  probing and measuring a lot of points and by chance I 
found a very interesting  thing:  Probing TP33 (which is close to the 
Trimble chip (U2) and  directly connected to pin 76) there is a 1PPS, 10 us wide 
signal.  I've  checked it and appears to be synchronous with the 1/2 PPS 
output so maybe it  can be routed to an output, probably I'll replace the 1/2 
PPS with it, using  the existing drive circuit and connector since it is very 
 straightforward.
My only concern is if this signal only is there during the  anomalous 
condition that I have now, I have to retest it after fixing it.
I  had asked several times if anybody had located a 1PPS signal on these 
units,  but the responses were negative and I had not probed systematically 
the board  before.  Also I'm taking notes of the signals found and I'll try to 
make  a partial schematic at least with the EFC circuitry.  When I fix it 
I'll  clean the notes and figures and I'll upload it to some place.

Best  regards,
Ignacio

El 12/11/2014 a las 11:47, GandalfG8--- via  time-nuts escribió:

Hi Ignacio

I have removed a "faulty" oscillator from one of these, not one  of my more 

relaxed moments and quite amazed when the board emerged totally  undamaged, 

but proof at least that it can be done:-)

With the oscillator removed there's access to two sets of connector  pads 

that will either accept SMA or SMB connectors and after  conducting tests 
on 

the original oscillator via flying leads just  soldered to the board I 

decided not to fit a replacement to the board after  all but to fit a 
couple of 

SMB connectors to make the setup more versatile, and  to avoid the 

possibility of having to remove another oscillator in the  future:-)

Another advantage of these pads is that if the board does get damaged  

removing the oscillator they can still be used as an alternative.

The connectors take the 10MHz from the oscillator into the board and  the 

EFC control voltage out from the board to the oscillator, now there's a  

surprise:-), and oscillator power can be completely external if required.

>From this it would seem that the reference supply from the  oscillator 

itself is not used in practice, certainly not in the "offboard" case  
anyway.

As others have suggested it seems likely your problem may not be the  

oscillator itself, but it still might be worth removing anyway to make 
testing  

and fault finding more straightforward.

The maximum positive excursion of the NTGS50AA should be 6 volts, not 5 as  

you're seeing, and another indication it might be worth removing the 

oscillator  to see how the board behaves stand alone.

I've not seen what seemed to be the repeated attempts at lock that you  

mentioned previously, but then I wasn't even aware for a long time that the 

control voltage could drive below 3 volts as well as above it:-)

This is my note from previous observation of my "faulty"  unit----

-------------------------------------------------------------

When first powered it brings up all LEDs and then switches to a green LED  

for a few seconds and then amber. It starts a self survey and acquisition  

process with all appearing ok, and the DAC voltage reported as 3.000002  

volts.

Sometime later, I've seen as short as 6 minutes but as long as 12 to  15 

depending on how long the oscillator has been turned off and allowed to 
cool,  

once enough satellites are being tracked, the DAC voltage starts to 

increase,  presumably seeking to drive the oscillator frequency to 10MHz, 
but the 

frequency  doesn't reach 10MHz and the DAC voltage ramps up to 6.000004 
volts 

over a period  of approx 30 seconds where it remains.

As the DAC voltage crosses approx 5.6  volts the Red "fault" LED is 

switched on, as opposed to green that would  normally be expected to 
indicate all 

was well, and Lady Heather's "OSC:" report  switches from Good to Bad and 

highlights red. Similarly "Normal OSC age" changes  to "OSC age alarm" and 
also 

highlights red.

---------------------------------------------------------------

I've attached a Lady H plot that shows this, hopefully it will get through  

OK.

The above DAC voltages were as reported by Lady Heather but I've checked  

these and, when the board is working as it should be anyway,  they're very 

close.

In my case the problem described above was an oscillator that had  aged 

beyond the upper 6 volt limit, needing approx 6.54 volts to reach  10MHz, 
and 

once removed from the board I was able to add a simple op  amp level 
shifter 

to bring it back into range just to prove all else was ok,  which it was, 

but obviously Lady H now indicated the EFC into the level  shifter rather 
than 

at the oscillator itself.

Whilst your problem sounds like it might not be quite such an obvious  fix, 

removing the oscillator would open the loop and make  testing both the 

oscillator and the board much easier, so much as it's  a pain I do feel 
that's 

probably your best next step.

Regards

Nigel

GM8PZR

In a message dated 12/11/2014 02:06:54 GMT Standard Time,  

_eb4apl at cembreros.jazztel.es_ (mailto:eb4apl at cembreros.jazztel.es)  writes:

Hi,

Removing the oscillator for testing and replacing it  with other if it 

was the culprit was my first option.  I have a spare  Trimble oscillator 

that probably came from other NTGS50AA since it still  have the foam band 

attached, but this oscillator is really aged, it needs  7.91 V to bring 

it on spot and the maximum control voltage of the NTGS50AA  is 5 V.

I was trying to avoid removing the oscillator but probably it must  be 

done to clarify things.

Thank you,

Ignacio  EB4APL

.

El 12/11/2014 a las 2:40, Mark Sims escribió:

I  have seen this caused by the oscillator not responding to the EFC  

signal.  Fixed it by swapping in a MV-89 oscillator.

The  oscillators used in these units don't output an oven temperature 

monitor  signal.                     

_______________________________________________

 time-nuts mailing list -- _time-nuts at febo.com_ (mailto:time-nuts at febo.com) 

To unsubscribe, go to  

https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts

and follow the  instructions  there.

_______________________________________________

time-nuts  mailing list -- _time-nuts at febo.com_ (mailto:time-nuts at febo.com) 

To unsubscribe, go to  

https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts

and follow the  instructions there.

_______________________________________________

time-nuts mailing list -- _time-nuts at febo.com_ (mailto:time-nuts at febo.com) 

To unsubscribe, go to 
https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts

and follow the instructions there.