[time-nuts] LPRO-GPSDO

[Magnus Danielson]

John Miles wrote:
>> I control my lpro101 with trimble thunderbolt .
>> now is more then one month , work good very stable .
>> But I am not happy , sure long time is more stable of original
>> thunderbolt with ocxo , sure is more stable in holdover ( we need in
>> holdover ? i never lost gps signal ) .
>> But short time is worst then original ocxo . I have 4 lpro101 , all four
>> have short time as datasheet near 2.5 E-11 .
>> Maybe I am not lucky and my all lpro are not the best in short time ,
>> but because my primary interest is short time , I  need to drive my
>> spectrum analyzer and my microwave generator with clean and stable
>> reference , the lpro not respond to this needs .
> 
> GPSDOs and atomic standards like the LPRO are not that different inside.
> There are different bells and whistles, but they are both just simple PLLs
> at heart with a more-or-less noisy reference that contributes long-term
> stability to a quartz oscillator, of varying quality, that in turn is
> responsible for short-term stability.
> 
> The LPRO is about 10-12 dB cleaner at 1 Hz than my older Thunderbolt, but
> about 20 dB worse at 1 Hz than the newer Thunderbolts that most people have.
> Why?  Open the LPRO up and you'll see that the 10 MHz crystal is a small,
> ordinary-looking part soldered to the board and exposed to the environment.
> It's reasonable to expect worse short-term performance than the OCXOs in
> GPSDOs, which tend to be better-than-average parts.
> 
> It probably wouldn't be too tough to replace the 10 MHz oscillator in the
> LPRO with an external OCXO, as can be done with a GPSDO.  It would be
> necessary to measure the V/Hz figure of the existing oscillator and
> construct an attenuator or amplifier to reconcile it with the new
> oscillator.  Or you could leave the LPRO intact and use its existing 10 MHz
> output to discipline an external OCXO in a slow (> 100s) loop.

On that note...

I am currently reverse-enginering the FEI FE-5607E Rubidium, which is a 
variant of the FE-5607A, but with the internal OCXO removed such that a 
larger and better OCXO may be used. Interestingly enought, the EFC is 
always available on the 13-pin circular connector. The variations with 
10 MHz output (FE-5607A), 5 MHz input (FE-5607D) or 10 MHz input 
(FE-5607E) is just what SMB connector hooks in where and wither an OCXO, 
Doubler or nothing is in the OCXO location. The unit I have is partly 
damaged, but I've got the basic plot figured out fairly well.

If someone do have the service manual for this one, I would be greatful 
ofcourse.

I agree that some care would need to be taken in ensuring similar V/Hz 
figure, as it will not only control the PLL loop bandwidth, it will also 
affect the Q of the resonance, and you do not want to raise it to insanity.

> The latter is basically what will happen anyway in many test instruments
> that accept an external 10 MHz reference, so it may not actually be a
> problem.  You should try to get a feel for the loop bandwidth of any
> instruments you want to drive from an external reference, so you can avoid
> wasting time making improvements that won't be noticeable.

You mid-tau properties should improve by implanting an oscillator, if 
done correctly. If you don't need it, then it certainly is unnecessary.

Cheers,
Magnus