[time-nuts] Some data from my LPRO GPSDO

[Matthias Welwarsky]

Hi,

I've been running my prototype LPRO GPSDO for a couple of days now. I've 
attached some plots of key data extracted from the GPSDO log.

RAW GPS TIC , this plot is the the raw measurements of the TIC (phase relation 
between GPS 1PPS and 10MHz clock).

Phase error is the basically the same as RAW, but filtered with an EMA with a 
200s time constant. It's the error input to a PI control loop.

Frequency offset is the second-to-second differences of the phase error. Not 
sure about the merit of this plot.

Temperature is the on-board temp sensor, it basically measures the PCB 
temperature.

DAC / DAC bias is the DAC output. The blue trace is the actual DAC output. The 
red trace is the state of the integrator part of the PI loop.

DAC MDEV is the mdev calculated from the DAC output scaled by (4ppb / 65536).

There's an interesting aspect shown in this data:
- the DAC output has a distinct downwards trend. It shows the LPRO drift 
against the GPS. The DAC MDEV shows it as well, see the upwards tick starting 
at tau= 3 * 1e4.
- the temperature plot has a similar trend.

As the PCB temperature is largely dominated by the heat from the LPRO, I 
reason that this LPRO has a problem with the lamp or cell heater: The 
regulation is drifting, causing a much higher frequency drift than normal.

I know that this LPRO is problematic. I obtained it as "broken" and was able 
to get it back to life with meticulous adjustment of the synthesizer and 
cavity resonance, but just barely. A broken heater would be an appropriate 
explanation.

BR,
Matthias
-------------- next part --------------
A non-text attachment was scrubbed...
Name: lpro-gpsdo.png
Type: image/png
Size: 54411 bytes
Desc: not available
URL: <http://febo.com/pipermail/time-nuts_lists.febo.com/attachments/20200305/e2a9f916/attachment.png>