[time-nuts] Controlling FEI 5680A
magnus at rubidium.dyndns.org
Sun Jan 15 19:28:20 EST 2012
On 01/16/2012 12:04 AM, bg at lysator.liu.se wrote:
> Hi Chris,
>> One other front end change. I few people have Thunderbolts and it
>> would be faster to lock the FE5680 to the 10MHz signal then to the
> Or remove the OCXO from the Thunderbolt and feed the GPS receiver with the
> FE5680 10MHz. Either modify the FE5680 for EFC och program a uC to get
> time/freq error from the receiver (bytes 6 to 9 in message 0x8F-A7). Then
> close the loop by adjusting 5680 frequency through RS232 from the same uC.
> Depending on use, you can retain the good Tbolt ocxo and lock it with a
> PLL to the 5680 10MHz.
Ages ago we talked about locking up Superstar II receivers to an
external 10 MHz source, such as a 10811. The Superstar II receivers
builds on the Zarlink chipset, which uses 10 MHz as reference for their
RF frontend chip, which then times the digital correlator chip. The
Superstar II receiver has support for external clock input, and it is a
fairly simple modification to have them accept it from the connector
which you solder onto the layout part of the PCB. However, the Superstar
II does not include the lock-up steering of the Thunderbolt, but larger
cousins has this feature. The idea was to sniff the time error reports
and let a small MCU do the PI-lockup of the OCXO of choice.
I believe that attempts to use the frequency inputs on the Superstar II
was done for a project, but as I recall it nobody had the time to follow
it through, but I recall giving telephone-guiding around the PCB to show
where to pick things up.
Essentially much the same can be done with the Thunderbolt, which is an
amazingly open platform as PLL parameters is easy to modify to fit the
Looking at the larger receivers, the Ashtech Z12 includes frequency
locking for instance. This is used in reference applications where
typically a rubidium is used and then a logger to pull RINEX data out of
them and "report home".
However, I must say that use of a proper OCXO or rubidium, as reference
to a GPS receiver will do much to lower the receivers close-in noise if
done right, and this is a step away from the PPS crazyness for a better
overall solution. This is where the Thunderbolts performance over
TCXO-GPS with PPS output comes, and locking the OCXO using a PI
regulator brings icing on that cake.
Essentially all of the more modern receivers does not lock up to 10 MHz
but has other odd frequencies. Letting a DDS generate the odd frequency
might be useful approach. However, using DDS getting correct 10 MHz
might be difficult, so either a slow phase-correction ramp or use of a
PLL with the right "gears" might solve it if true 10 MHz is needed.
More information about the time-nuts