Thunderbolt® E GPS Disciplined Clock

[branadic]

Hi to the community,

I have a Thunderbolt® E GPS Disciplined Clock at lab and I'm confused by what I get.
The first module that I received obviously had some defect as amplitude was of compared to datasheet. After some weeks I got a replacement last friday, with amplitude okay right now (1.32Vpp into 50?).

I had run the gpsdo over the last weekend together with Lady Heather's Disciplined Oscillator Control Program. It turned out, that parameters and specifications on real device seems to differ from datasheet and also from what is shown on the web.

First of all the parameters

Parameter		My Device		User’s Guide (Factory Default)		web
TC:		60sec		10sec		100sec.
DAMP:		1.000		1.000		1.200
GAIN:		1.707Hz/V		8.830Hz/V		-5.000Hz/V
INIT:		2.500V		2.000V		0.000V

Second, after the weekend the stability is also different from what the datasheet suggests.

Maybe someone can help me to understand how to set the oscillator in the right way to get good stability results. Thanks.

branadic

[Dr. Frank]

Hi branadic,

I got the Thunderbolt, it contains FW 3.0 and the better OCXO, labelled originally from Trimble.
Description of versions here: http://www.ke5fx.com/tbolt.htm

I have chosen:
TC = 1000sec (before used 500sec)
Damp = 0.7
Gain = -5Hz/V (for the OCXO!!)
Init = 0.787V (dependant on OCXO, and I stored that parameter recently into the TB, to have a faster startup)

Concerning TC and damp, there were long discussions on the time-nuts lists, but less than 100sec was not recommended, afaik. My settings work quite well.

Additionally, related to the location of our house and the surrounding, I have chosen EL (Elevation) = 20°, and AMU = 5.50


The 10MHz Signal is 2.56Vpp @ 50 Ohm, and 5.08V open.

As you see half of my amplitude only, make sure, that there is no hidden 50 Ohm, already..

1pps has TTL level, i.e. 4.96V, 10µs pulse.

Have you already determined and fixed your position by a 24h survey?
That is required, otherwise, you'll get no stable time at all.
(Currently, your coordinates are on the Stuttgart Campus, obviously your lab @ work, nice glass building )

The gain parameter has to fit your type of OCXO, that may also be the root cause of your problems.
You should open your unit and have a look inside, maybe a photo, what's contained.

Frank

- sent from my basement lab -

[rf-loop]

Hi to the community,

I have a Thunderbolt® E GPS Disciplined Clock at lab and I'm confused by what I get.
The first module that I received obviously had some defect as amplitude was of compared to datasheet. After some weeks I got a replacement last friday, with amplitude okay right now (1.32Vpp into 50?).

Bit low but...

In my unit +11dBm (Version 3.0  T.Thunderbolt)


For best result it is best to keep it running example 2-4  weeks continuously. (Inside OCXO XTAL retract and running it back to slow ageing curve take some time)

After then you can adjust better values for discipline.

Also for good result it is best to find your individual unit V/Hz value.
For this you need good frequency counter (abd accuracy not important) then you manually adjust Trimple OCXO frequency control voltage and measure how many Hz it change when you change.

Initial voltage is just voltage what is set for start. It can be what ever but startup/stabilization is more fast if it is near right value. This also change when OCXO ageing go forward.

In my case I use TC 750s and damp 1.1  but I have quite good well aged OCXO and also thermal sensitivity is not so bad.

These default values, example short TC and so on, they are for perhaps better for new units and when purpose is more like time accuracy than frequency.  With original settings it accept more frequency error for drive time error out quite fast. 

[Dr. Frank]

branadic,

I just scanned through time-nuts, and there is a discussion about the new ROHS E type of Thunderbolt.

The new type should be compatible to the old Thunderbolt, but it was a cheaper type, maybe with a different OCXO or piezo.

There was also an old Thunderbolt with FW labelled "E", which may not be mixed up.
So, if you got the new E version directly from Trimble, and not an old TB, that would explain the different TC, damp and gain parameters.

All these are valid and stable, when the GPSDO has a fixed position, and did run for at least 2-3 days.
That's completely sufficient for the OCXO to stabilize, and for the GPS processor to "learn" the behavior of the whole set, concerning hold-over mode, i.e. T.C. (temperature coefficient) of the OCXO, and DAC value.
I think, whenever I let my unit power up again, it will meet the stability / Alan parameters after about 2 days.
It also depends on, whether you monitor the 10MHz, or the 1pps signal. Latter one is more stable, in most cases.

Ah, one further hint: LH probably does not measure the Allan statistics of the TB correctly. That's because she can only compare the OCXO against the jittered GPS signal.

If you want to reconstruct the Allan diagrams coming from Trimble, or from the other Time-Nuts, you better have to measure that, using an external T.I. counter, and a similar or better reference, like another free-running OCXO (in the counter), or a Rb clock. I recommend TimeLab for these measurements, also available from KE5FX.

So you may measure <1e-10 StD @ 1s for the 10MHz already, instead of about 2e-9, what LH usually displays.
Frank

[rf-loop]

branadic,


All these are valid and stable, when the GPSDO has a fixed position, and did run for at least 2-3 days.
That's completely sufficient for the OCXO to stabilize, ...

Frank

Yes if think minimum and accept that all are not optimal for start finding optimal parameters. Also ok if think typical Trimble all other instabilities, poor temp stability and cheap OCXO. For normal use 2-3 days is enough. But, very important is that also positioning is well done.

But if talk more deep...

Yes and no. It depends...

Also different xtals are different and so on.

If think ageing curve, retrace etc bit more accurately 2-3 days is "nothing". 


But I do not want start more about interesting frequency-nut here  (time-nut is boring because after there is more than one clock, no one know exact time)

Who is more intersted can start studying example with this quick lesson:
http://www.oscilent.com/esupport/TechSupport/ReviewPapers/IntroQuartz/vigtoc.htm

There is also sketchy story about XTAL thermal hysteresis and retrace
http://www.oscilent.com/esupport/TechSupport/ReviewPapers/IntroQuartz/vighyst.htm
"Figure 24. Oven-controlled crystal oscillator (OCXO) retrace example, showing that upon restarting the oscillator after a 14 day off-period, the frequency was about 7x10-9 lower than what it was just before turn-off, and that the aging rate had increased significantly upon the restart. About a month elapsed before the pre-turn-off aging rate was reached again. (Figure shows Df/f in parts in 109 vs. time in days.) "

[branadic]

Thanks for the response.
10MHz output level is okay, datasheet specs min. 7 dBm ±2 into 50 ?, so 1.32Vpp is fine.

http://trl.trimble.com/docushare/dsweb/Get/Document-383329/022542-010B_Thunderbolt-E_DS_0807.pdf

That is also where I got the stability graph from. So maybe I understood ADEV OSC wrong? Isn't this how ADEV OSC should look like?

I also tested GAIN today, quick and dirty with a pour HP5316B (the best I have at hand):
@ 0.00005V the GPSDO gives 10.00000MHz
@ 4.99995V the GPSDO gives 10.000009MHz
Obviously the +1.707Hz/V are not that wrong. Inital voltage of 2.1V seem to match as well for starting conditions.

Self-survey was already performed and as can be seen I've a wonderful free view to the sky (antenna is mounted on the roof). So maybe I give it a 1000h to age?

[branadic]

Ah, one further hint: LH probably does not measure the Allan statistics of the TB correctly. That's because she can only compare the OCXO against the jittered GPS signal.

If you want to reconstruct the Allan diagrams coming from Trimble, or from the other Time-Nuts, you better have to measure that, using an external T.I. counter, and a similar or better reference, like another free-running OCXO (in the counter), or a Rb clock. I recommend TimeLab for these measurements, also available from KE5FX.

So you may measure <1e-10 StD @ 1s for the 10MHz already, instead of about 2e-9, what LH usually displays.
Frank

Okay, I could use the HP5316B and the OCXO of a HP4194A to measure the time interval.

[rf-loop]

Ah, one further hint: LH probably does not measure the Allan statistics of the TB correctly. That's because she can only compare the OCXO against the jittered GPS signal.

 
Like a lady who investigates himself with distorted mirror.

[uncle_bob]

Hi

If you have the new "E" version of the TBolt, it has almost nothing at all in common with the older TBolt's. The OCXO is very different and the circuits on the board are completely changed. The newer version no longer needs the same +/- 12 and +5V supplies the old one did. Much of what you have found relates to the older model and not at all to the newer one.

Which version do you have?

Bob

[branadic]

As I wrote, it's the latest Thunderbolt E, a new device from Trimble, nothing used before

[dr.diesel]

This thread, and the early morning reading it spawned; leapsecond.com/time-nuts/ke5fx really has me interested!  I bought a Thunderbolt several years ago, not sure on the type but it was originally housed in a plug-in backplane type metal enclosure.  I whacked the board into a 2U rack case, mounted the antenna outside but never did any real testing or configuration due to the LH being windows only (runs OK under wine).  However reading today I see that has changed, there is now a native Linux beta version available!  WooHoo.

Anyhow, on to further reading/tweaking   

[Dr. Frank]

on time-nuts, September, there has been a discussion about TBolt performance, regarding stability (ADEV) and absolute frequency uncertainty.
branadic, I don't know, what you're after?

This subject: Tbolt issues may fit here:
https://www.febo.com/pipermail/time-nuts/2016-September/subject.html#start

TvB made some comparisons between GPSDOs, and I assume, he uses his MASER as reference: http://www.leapsecond.com/pages/gpsdo/
He uses the Time-Pod as a T.I. comparator, and for sure the TimeLab software.

What can be extracted from his measurements, that he obviously uses  TC=100sec, or 1000s, because that's where you see a hump in the Allan diagrams.


Dr. Diesel, you live in beautiful countryside, couldn't resist to spy your coordinates.. 
Very similar to northern Germany, btw.
In LH, I think your ADEV value for the 10MHz output is an order of magnitude too high.
I achieve ADEV on the order of 2E-9 @ tau = 1.
Your GPS antenna rod is also visible, maybe the foliage of the trees might distract the signal. Perhaps you better choose > 20° elevation.


Frank

[uncle_bob]

on time-nuts, September, there has been a discussion about TBolt performance, regarding stability (ADEV) and absolute frequency uncertainty.
branadic, I don't know, what you're after?

This subject: Tbolt issues may fit here:
https://www.febo.com/pipermail/time-nuts/2016-September/subject.html#start

TvB made some comparisons between GPSDOs, and I assume, he uses his MASER as reference: http://www.leapsecond.com/pages/gpsdo/
He uses the Time-Pod as a T.I. comparator, and for sure the TimeLab software.

What can be extracted from his measurements, that he obviously uses  TC=100sec, or 1000s, because that's where you see a hump in the Allan diagrams.


Dr. Diesel, you live in beautiful countryside, couldn't resist to spy your coordinates..
Very similar to northern Germany, btw.
In LH, I think your ADEV value for the 10MHz output is an order of magnitude too high.
I achieve ADEV on the order of 2E-9 @ tau = 1.
Your GPS antenna rod is also visible, maybe the foliage of the trees might distract the signal. Perhaps you better choose > 20° elevation.


Frank

Hi

There is virtually nothing at all on Time Nuts about the version of the Thunderbolt that the OP has. Trimble made several units under the same name. The old ones come with revision codes on them and some revisions go up to E. That is not at all the same thing as a "Thunderbolt E".

Bob

[Dr. Frank]

Hi

There is virtually nothing at all on Time Nuts about the version of the Thunderbolt that the OP has. Trimble made several units under the same name. The old ones come with revision codes on them and some revisions go up to E. That is not at all the same thing as a "Thunderbolt E".

Bob

That's correct, the TB 'E' is not available on the 2nd hand market yet.

Anyhow, this discussion applies to all versions of TBs, and also to all kind of of GPSDO.

Frank

[uncle_bob]

Hi

There is virtually nothing at all on Time Nuts about the version of the Thunderbolt that the OP has. Trimble made several units under the same name. The old ones come with revision codes on them and some revisions go up to E. That is not at all the same thing as a "Thunderbolt E".

Bob

That's correct, the TB 'E' is not available on the 2nd hand market yet.

Anyhow, this discussion applies to all versions of TBs, and also to all kind of of GPSDO.

Frank

Hi

The OP was pretty specific about having an E and looking for info on it. All of the stuff that does not apply at all to that unit has not helped much ....

Bob

[branadic]

Some news...
Obviously there are two versions of TB 'E' available, a version 60333-30 with SOCXO and version 60333-50 with DOCXO. I have the latter one.
This explains also the difference in initial GAIN, for SOCXO it is 8.83Hz/V and for DOCXO it is something around 1.6Hz/V? The users giude obviously refers to the SOCXO version. A hint would have been perfect.
Currently I perform a precision survey. Up to now only a self servey was performed.

I wish software (doing all necessary steps after installing the GPSDO by a single button click or a wizzard) or users guide (step by step guide) would be more intuitive to get the best performance out of the unit.

[w9gb]

The Trimble Thunderbolt E (BLUE Label) is DIFFERENT from original Trimble Thunderbolt (RED Label) and
the Time-Nuts group purchase of Tri-voltage version of the Red Label model.

Notes on Trimble Thunderbolt (Red Label) performance and modifications
June 1, 2008: HP 10811 upgrade
by John Miles. KE5FX
http://www.ke5fx.com/tbolt.htm
==
April 13, 2022, Trimble sold its Time and Frequency products to Precisional LLC.
Those products are now sold under the Protempis brand name.
https://www.prnewswire.com/news-releases/trimble-to-divest-four-businesses-to-the-jordan-company-301524488.html

PROTEMPIS Division of Precisional LLC
They have been changing Product Labeling during later-half of 2022.
https://www.protempis.com/
==
Protempis Thunderbolt E GPS Disciplines Clock
NEW Data sheet
https://www.protempis.com/_files/ugd/cd4160_3e9b6c8faf4a4bde8777e26097ce8834.pdf
==
NOVOTECH is the North American Distributor for the Protempis products.

Novotech Canada
57 Iber Road, Unit 2
Ottawa, ON KS2 1E7
1-800-268-8628
Sales@Novotech.com

https://novotech.com/line-card/protempis-gps-and-timing-products/

Protempis Thunderbolt E GPS Disciplined Clock (Formerly Trimble Thunderbolt E)
Product Number: 60333-50
https://novotech.com/thunderbolt-e-gps-disciplined-clock-304.html