Search This Blog

Radio Guy Tees

Radio Guy Tees
Radio Guy T-Shirts

Sunday, 30 December 2018

OCXO Update


You may recall back here where I built a VE2ZAZ GPS disciplined OCXO as a frequency reference for my shack. This was back in 2013 and the unit has pretty much trundled along, doing its thing ever since.

My recent fiddling here resulted in my realisation that a more accurate 1PPS signal is achievable by putting certain GPS receivers into a fixed location mode for timing applications.

So, I decided to invest in a decent enclosure and re-build the frequency standard. I have effectively swapped the PSU, GPS receiver and the OCXO itself.

This is the internal gubbins of the new unit:

This is the main PCB - the wire you can see from a PIC pin is to a reset switch on the back panel which clears the alarm condition manually. A firmware mod included this but the PCB is prior to this change.

This is the MAX232 device and regulator which converts the PIC comms into RS232 levels for the PC.

This is the PSU board and the OCXO itself:

And finally this is the GPS receiver parked in the corner:

The config of the system is fairly complex, but for those playing along at home, these are the settings I used to get the system into a stable state. I first forced the DAC value to be close using my test gear and manually adjusting the value, then I left the device running overnight with these parameters:

Then now the unit is settled, I have changed the config to be thus:

The unit itself looks like this at the front, there are two 10MHz outputs plus a selectable 1 or 5 MHz output:

The rear of the unit has two further 10MHz outputs:

I have more than enough "bits" here to make another one of these; anyone want one?


  1. Mark,
    Good day for all of you.
    I presume the accuracy of the final pll outputs depends on factors, like differences in internal time delays of the circuitry we use.Also on the basic accuracy of the OCXO we manage.

  2. Hi Sama, good to hear from you. Yes, there are many variable factors in play - that is why the firmware is configured to take the average of many 16 second samples before deciding to increase or decrease the control voltage to the OCXO.