[time-nuts] Trimble Thunderbolt setup optimization andadding a10811 OCXO

Arnold Tibus Arnold.Tibus at gmx.de
Tue Jul 14 22:06:02 UTC 2009


Dave,

there is a clear answer:
Why shall I do several rounds when I can succeed in one? 

If you look to my oadev values achieved with the disciplined 10811 
you will see already a significant improvement over the normal TB values 
I have seen. But there are clear indications that it shall be possible 
to achieve even better results. The EFC voltage shall not move 
like the tail of a nervous cat with 400 µV in 1 to 5 minutes, which does 
have naturally direkt negative effects  to the 10 MHz oscillator. 

As long as I can see as well the exchange of some SAT signals on 
the PPS line (sometimes) as jumps of 40 and up to 80 ns, then something 
must be possible to be done better, that I cannot accept as state of the
art.

I believe that I should not see that amount of degradation of a good 
OCXO in the first minutes until I am reaching the turnover or crossing 
point with the GPS output. The GPS starts to be better not before 
somewhere between 10 to 20 minutes I understand. 

Therfore I think to encrease the time constant of TB is important. 
I would think perhaps to go for two steps as you propose after 
having set TB properly. I want not to loose th good noise behavior. 
As second stage one could think if it does make sense to connect 
a PRS10 - coupled via a signicant higher time constant....

But perhaps I am wrong, 
I am always open to be convinced by better explanations and 
to absorb that knowledge. 

regards
Arnold  



On Wed, 15 Jul 2009 08:41:21 +1200, Dave Brown wrote:

>Why not leave the TBolt original OCXO in place and run the 10811 as a 
>separate oscillator phase locked to the TBolt OCXO? Then you have full 
>control over what loop parameters you use in the external PLL, as well 
>as still having the TBolt loop parameters to tweak if necessary.
>DaveB, NZ

>----- Original Message ----- 
>From: "Arnold Tibus" <Arnold.Tibus at gmx.de>
>To: "Discussion of precise time and frequency measurement" 
><time-nuts at febo.com>
>Sent: Wednesday, July 15, 2009 5:36 AM
>Subject: Re: [time-nuts] Trimble Thunderbolt setup optimization 
>andadding a10811 OCXO


>> Don,
>>
>> welcome in the boat.
>> You describe with other words what I am fighting against.
>> There must be a logic behind which we did not find yet.
>> I want to understand what is hidden behind all the setting 
>> parameters
>> and find a way to get the 10811 running without degradation but
>> just disciplining the OCXO drift away.
>> But without knowing the logic behind and without a known reference
>> a bit difficult...
>>
>> Regards
>> Arnold
>>
>> On Mon, 13 Jul 2009 13:15:29 -0700 (PDT), Don @ True-Cal wrote:
>>
>>>Arnold,
>>
>>>I have the same question as well as observations as yours. I have 
>>>been working with both the TBolt and Fury connected to two different 
>>>10811s. With either the TBolt or Fury, I can't get the TC large 
>>>enough to only provide a nudge to the EFC every few minutes. Both 
>>>receivers get real unstable and cause large Freq and PPS swings with 
>>>large TC numbers. It seems the EFC is providing many corrections per 
>>>minute even with the TBOLT set to 300 secs. Anything larger than 
>>>300-500 sec. on the TBolt and EFCD=150 (don't know what the equates 
>>>to in secs.) on the Fury will cause instability. I have better luck 
>>>disabling desciplining on the TBolt or disconnecting the antenna on 
>>>the Fury (no command to disable that I have found) when I need to do 
>>>a Tau 0.1-10 sec. low jitter measurement.
>>> Regards...
>>>Don
>>
>>
>>
>>
>>>________________________________
>>>From: Arnold Tibus <Arnold.Tibus at gmx.de>
>>>To: Discussion of precise time and frequency measurement 
>>><time-nuts at febo.com>; Tom Van Baak <tvb at leapsecond.com>
>>>Sent: Monday, July 13, 2009 1:56:56 PM
>>>Subject: Re: [time-nuts] Trimble Thunderbolt setup optimization and 
>>>adding a10811 OCXO
>>
>>>Tom,:
>>
>>>>> Astonishing for me,
>>>>> I found that the 10811 does behave better in undisciplined
>>>>> mode for a limited time:
>>
>>>>This is to be expected. An undisciplined high-quality OCXO
>>>>like a 10811 will be hundreds or thousands of times better
>>>>than the raw output of a GPS receiver at short averaging
>>>>times. This is why a GPSDO exists; it allows the OCXO to
>>>>run almost free for minutes to hours but allows GPS to rule
>>>>for long-term.
>>
>>>yes that is what I thought, but if I do observe correct, TB's
>>>OCXO does not run at all free for minutes. With a time
>>>constant of 360s the EFC and thePPS is surfing up and
>>>down, why, I do not see the need to shift this OCXO within
>>>2 minutes 80ppbs up and pps15ns as well - with this stable
>>>oscillator....what is the driver, the SATs (I think not so)!
>>
>>>Arnold
>>






More information about the time-nuts mailing list