[time-nuts] GPSDO Alternatives

Mark Spencer mspencer12345 at yahoo.ca
Wed Dec 5 23:26:34 UTC 2012

Over the last two years or so I've put some thought into various home brew alternatives to purchasing a surplus or new GPSDO.  The goal for me would be to have a reference source that combined the short to medium term stability of one of my best OCXO's with the long term stability of a GPSDO.  
The application would be to serve as a house standard for my test gear and a time server.

I've envisioned a scheme whereby I'd use an off the shelf time interval counter (probably a surplus HP5370) to continuously compare the OCXO output to the raw 1pps output from a suitable GPS receiver.  (This project would likely give me the excuse I've been looking for to purchase a CNS II GPS receiver that I believe are one of the better choices for raw 1pps accuracy.)  The counter would be connected to a PC via GPIB.
I'd then need to write the necessary code to periodically steer the OCXO via a to be determined digital to analog converter which in turn would then drive the EFC input on the OCXO.   Rather than implement a software PLL scheme I'd likely start by simply computing the average drift over each day and then simply adjust the OCXO every day or so but eventually I'd expect to implement a PLL scheme in software.    I’m hopeful that at first I could implement this in EZGPIB or something similar.   I expect eventually I’d end up coding this in C.

The main missing piece in the puzzle for me is a suitable DAC that can commanded by a PC (either by RS 232 or GPIB.)  I leave PC's and various pieces of test gear on all the time currently (they help heat my basement lab in the winter) so I'm not worried about dedicating a PC and TIC to this.

I'd also need a low noise power supply for the DAC and I suspect the performance of the DAC and the pysical interface between the DAC and the OCXO would be the weakest link in this whole system.

After contemplating the time, effort, and expense to complete a project such as this I've settled for now on simply manually adjusting my OCXO's from time to time and if I am concerned about the drift while using once of them as a reference I simply compare the OCXO in question to a GPSDO while carrying out my other measurement.  The drift of the OCXO can then be accounted for.   

In reality I can’t imagine having the time to even properly plan let alone implement something like this until I retire and I suspect I’d be lucky if I matched the performance of my best existing GPSDO.   The other alternative that occurs to me is simply connecting a high end OCXO to a Thunderbolt board. 

Sorry if I come across as overly cynical or pessimistic here (:

> Message: 3
> Date: Wed, 05 Dec 2012 22:24:43 +0100
> From: Volker Esper <ailer2 at t-online.de>
> To: Discussion of precise time and frequency measurement
>     <time-nuts at febo.com>
> Subject: Re: [time-nuts] GPSDO Alternatives
> Message-ID: <50BFBB9B.7010105 at t-online.de>
> Content-Type: text/plain; charset=ISO-8859-1; format=flowed
> Am 05.12.2012 18:31, schrieb Bob Camp:
> > Hi
> >
> > If the intent is to come up with something in the same
> league as the TBolt
> > there are a few other things you will need:
> >
> > 1) Something to compare the two pps signals to within
> 0.1 ns....
> Following Ulrich Bangerts suggestions, that a loop time
> constant should 
> be at about 3 hours (GPS disciplining an OCXO), do I really
> need that 
> resolution? Ok, the more accurate, the better. But the
> question is: can 
> I reduce this requirement when using long time constants
> (10000s)? The 
> ratio then is 10E14...
> Volker
> ------------------------------

More information about the time-nuts mailing list