[time-nuts] Trimble Thunderbolt setup optimization and adding a 10811 OCXO

[Arnold Tibus]

Hallo all the Time Nuts experts,

I need some help to find the understanding for a 
optimized setup with Trimble's Thunderbolts with original 
and other OCXOs like HP's 10811.

At the first glance the output of the TB does look quite 
good if I trust the values given by LH adev etc. 
But how good are the data gained without an external 
reference, may I believe them? 
(Ask a baker if his bread is good...!)
How do they judge the quality, what they refer to?

But I observe some things which makes me thinking 
and raising questions. 

If I look to the graph of Mark Sims' and John Miles' 
Lady Heather Program I see eg. up to 500µV EFC
 jumps due to SATs phasing in or out (where every µV 
does count!) resulting in PPS changes of around 7ns and 
frequency changes of 100ppb and more within several 
minutes or sudden  jumps. 
I thought that this should be eliminated by processing and 
filter routines.

I have adapted a HP-10811 oscillator (Z3801-Model, 
outer heater not connected)
to one of my TBs as shown by John Miles on his homepage. 
Was not difficult at all.

To get a base I did run it undisciplined in the TB system 
with a fixed EFC (set to get as close as possible 0.00ppb) 
and watched the output of LH. Setting different voltages 
I could easily derive the gain factor.

Astonishing for me,
I found that the 10811 does behave better in undisciplined 
mode for a limited time:

After a warming up time of several weeks this oscillator does 
actually show a very low drift with the EFC Voltage set to 
around  -80mV (-0.079724V actually, drifting about +1mV 
per week) with a gain set to -0.7425V/Hz. 
(I used LH with discipl. mode disabled 
to determine these values.)

The adev values for the undisciplined 10811:
pps oadev over 2311 pts
1 tau		5.5578e-011
2 tau		4.0426e-011
5 tau		3.6200e-011
10 tau		4.6242e-011
20 tau		5.4782e-011
50 tau		3.1047e-011
100 tau		1.9876e-011
200 tau		1.4337e-011
500 tau		8.0590e-012
1000 tau	9.0174e-012

osc oadev over 2311 pts
1 tau		3.5775e-009
2 tau		1.8014e-009
5 tau		7.1510e-010
10 tau		3.7145e-010
20 tau		1.8131e-010
50 tau		7.5577e-011
100 tau		3.6502e-011
200 tau		1.9246e-011
500 tau		8.0409e-012
1000 tau	4.5730e-012

I stopped there because the pps line began to drift away. 
The frequency curve, beside the jitter of max. +- 50ppb
 peaking, was smooth and varied along the straight center 
line about to +- 20ppb. At this point the efc should already 
have started pulling very smooth the pps back to zero 
+- a tolerance.

But when I run the TB in normal mode with disciplining on 
then the oscillator curve starts to be a bit nervous as 
the efc and the pps lines do surf slowly up and down with 
possible jumps as already mentioned. 

I tried to modify the damping factor, the time constant and
 as well the gain. 
I never got both, the pps and the osc, calm. When I run it 
with a time constant above 500s, the osc seem to be more 
stable but the pps starts to behave worse. 

The best values I could get that way 
for the disciplined TB. w. a 10811 ocxo 
after trying a lot of setup combinations are:

pps oadev over 10810 pts
1 tau		2.4865e-010
2 tau		1.7802e-010
5 tau		1.1912e-010
10 tau		9.3686e-011
20 tau		7.6403e-011
50 tau		4.9101e-011
100 tau		3.5379e-011
200 tau		2.0019e-011
500 tau		1.1316e-011
1000 tau   	7.6236e-012
2000 tau  	4.6755e-012
5000 tau   	3.2676e-012

osc oadev over 10810 pts
1 tau		6.0651e-009
2 tau		2.9834e-009
5 tau		1.2081e-009
10 tau		6.0071e-010
20 tau		3.0075e-010
50 tau		1.2337e-010
100 tau		6.3227e-011
200 tau		3.2970e-011
500 tau		1.3461e-011
1000 tau	                6.8074e-012
2000 tau	                3.4545e-012
5000 tau  	1.4172e-012

(I see just now, I should have waited for 1 step more 
to get 10811 pts.  ;-) 

If this would stay stable during days and weeks I do 
not know yet.

So, what do I wrong, what can be done better, what 
is the way to get the maximum out from the TBs?

The values gained with the original ocxo from trimble 
are even worse.What I am looking for is to get more stability 
at lower tau steps.

Why do I observe a crossing of the pps and osc oadev 
curves, which I did not see with the original oscillator?

Btw. the bottom side of the OCXO I unmounted is marked with 
'Phase Noise'.
It looks like that this series was equipped with selected 
OCXOs for low phase noise...!

I would appreciate it very much to get comments and advices.

many thanks i.a. and regards
Arnold