[time-nuts] Thought experiment on a low cost timing board

Steve Bennett swbenn at gmail.com
Thu Feb 23 13:42:54 EST 2006


Go for it! You can do this.

I have been working on a similar project using a Altera FPGA for
counters and latches and my favorite microprocessor (AVR). I use the
micro for monitoring the GPS, reading the counters and doing the calcs
to adjust the  OCXO. I divide the 10MHZ by a power of ten and latch
the value of the counter with the PPS. The micro compares the current
value of the counter with a previous value of the counter and makes
adjustments as needed. One could use the counters in a micro, as in
this project: http://www.rt66.com/~shera/index_fs.htm

I let the counter roll over and adjust the sensitivity of my control
loop by adjusting the number of seconds I let the counter run between
samples. Short sample intervals let me do coarse tuning and longer
intervals let me filter out short term deviations. The micro has a
uart that will talk with the GPS or with a host computer. Sounds
simple, but as usual, the devil is in the details.

Not to use a PC for the controller was mostly a decision made out of
ignorance. I do not know all the timing issues in Windows or Linux. I
have my hands full with stuff I think I know.  If I use an FPGA and /
or  a micro, I have a good chance of figuring out what went wrong when
the gadget starts giving me funny results. One of the advantages of
using a PC for the controller is all the cheap or free tools
available. With the AVR micros I use an open source compiler and IDE (
Winavr) If you can program C on a PC, you can do so on a micro with a
reasonable learning curve.

Using a micro also has the advantage of lower power and less noise.
Mostly, the clock will be soaking up all that time without any
operator intervention, no need for a screen, hard disk or all the
rest. For comunication to the outside world the AVR's UART will work
or it can be equiped with Ethernet, USB or just about any kind of
interface out there, but as you do that, those nasty details creep
back in.

For myself, this type of thing is a project where the benefit is in
the doing rather than the finishing. It's an excuse to feed my brain
all kinds of fun stuff. I may not be the best guy to opine on the
likelihood of profitably selling something like this, but I feel the
market would be a little small. I could be wrong though..


> Message: 2
> Date: Thu, 23 Feb 2006 08:07:41 -0800
> From: John Pettitt <jpp at cloudview.com>
> Subject: [time-nuts] Thought experiment on a low cost timing board
> To: time-nuts at febo.com
> Message-ID: <43FDDDCD.8080002 at cloudview.com>
> Content-Type: text/plain; charset=ISO-8859-1
> I asked a couple of weeks back about getting an ICS525 with an eye to
> making one of my Soekris 4801's more stable by using the 5Mhz from my
> GPS-DC to generate the clock.  However I have a "better idea" that I'd
> like to run past the group for comment.
> I read the work phk did (http://phk.freebsd.dk/soekris/pps/) and came up
> with the following product idea.
> The spec:
> PCI 3.3v board with:
>     10Mhz OCXO (provision for external clock source?)
>     Uart (serial is an endangered on many PC's)
>     Free running counter driven from the OCXO and readable by PC
>     inputs to latch the counter (how many?) with the latched result also
> readable (for PPS)
> This will let me turn any PC into a stable clock using cheap gps (eg
> GPS18LVC) that has a PPS output.   However the real win comes if I can
> figure out  how to discipline the OCXO using an external 1PPS input - if
> I can do that then I can provide a stable X MHz out (where X may be
> programmable using the ICS525) making the board very useful for lots of
> things.
> So does the basic idea make sense?  More importantly can somebody opine
> on if it's possible to use a 1PPS to discipline an OCXO (I've seen
> designs to it from a 10KHz signal for sale on eBay)) -
> Now if this actually looks feasible I may pay somebody to design it and
> build a a few  to sell on ebay.
> John
> ------------------------------
> _______________________________________________
> time-nuts mailing list
> time-nuts at febo.com
> https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
> End of time-nuts Digest, Vol 19, Issue 17
> *****************************************

More information about the time-nuts mailing list