[time-nuts] FE-.5680A trimming resolution

Javier Herrero jherrero at hvsistemas.es
Sun Jan 29 18:19:36 UTC 2012


El 29/01/2012 14:45, Bob Camp escribió:
> Hi
>
> Very interesting.
>
> It sounds like dithering would be needed to get down to parts in 10^-14. If we do that from an external device (PC / PIC / FPGA / whatever) it would be useful to know the delay between the serial command and the DDS update. The more variable the delay, the less accurate the dither.
>
> Bob
>
>
Since I've all the things powered on the bench table, it is easy to 
measure: 21ms from the start of the serial message, to the end of the 
last DDS programming word. I've repeated several times, the time does 
not seem to have submitted to latencies, it is very constant.

The automatic update rate of the DDS in the unit is somewhat around 28.5 
seconds (I misread an inter-word delay of 671.5us as a 671.5 seconds 
interval). I'm trying to get a long capture from the logic analyzer to 
try to plot the DDS programming variations, but then I need to 
post-process it since the analyzer has no SPI decoding so I'm gathering 
the bit streams.

My next intention is to replace the OCXO in one of my Thunderbolts with 
a Rb and use a small microcontroller to get the voltage correction from 
the Thunderbolt through the serial port and convert to a 5680A offset. I 
probably will use a Luminary since I've some small evaluation boards 
around, and I only have to add a MAX3232-class interface IC for the 
serial ports.

Now... my question is to decide which of the two Thunderbolt will be 
converted to Franken-Thunderbolt. I've bee monitoring both from some 
time, feeding them from same GPS signal, and both looks quite the same 
(I was trying to see if one of them has a worse OCXO than the other. One 
is from the past group purchase, and the other has been purchased not 
long ago - but both are very similar, with same firmware and the Trimble 
OCXO, and monitored with Lady Heather they are really very similar). 
Next... clean-up the Rb output with an OCXO (the one from the Trimble or 
a 10811A I've around). But this depends on other things I must do... 
with more priority :)

Regards,

Javier



More information about the time-nuts mailing list