Post by thread:[PIC]: GPS frequency standard

I've appended the material I posted yesterday re a PIC based GPS frequency standard. I've been thinking about this and wondering if it might be possible to produce a very very simple albeit nowhere near as good result moderately simply using any GPS with NMEA or other serial output. I'll almost certainly not get to investigating the following but it may give someone some ideas. The original product is not especially expensive, using mainly standard ICs and a VCXO BUT the 1 Hz pulse output GPS is unusual. PERHAPS: Using the leading edge of the NMEA serial stream which occurs once a second would be good enough ??? This necessarily has an excellent long term stability but short term may be excessively dependent on internal processing in the GPS. At an absolute minimum the PIC could modify its own clock with a DAC and varicap. "On frequency" decisions could be based on whether the NMEA edge occurred prior to or before a program event which should take exactly one second (or a submultiple thereof). This need not necessarily even be IRQ driven - sampling an internal comparator at the desired time to determine if the NMEA edge has arrived may suffice. While this may seem to be a "blunt axe" approach with a quantisation only of the order of 4 cycles of the PIC oscillator, a clever implementation which fine tuned the oscillator so over the long term it "hovered" on the edge of the phase transition should be able to have increasing accuracy the longer it was turned on. Initial accuracy should be able to be in the order of 0.2 ppm (bout 5 seconds error per year) soon after turn on, improving by several orders of magnitude with time. Nowhere near the "proper" implementation in accuracy BUT amenable to cheap implementation with little more than a PIC and, more importantly, an input when required from any NMEA GPS. Even the initial accuracy should be several orders of magnitude better than for an out of the box crystal oscillator based product. . Russell McMahon ========================== PRIOR POST: Build your own PIC controlled GPS freq std.Brooks Shera W5OJM, QST July 1998 page 37 (how's that for an encyclopaedic memory?)(thanks Google). The original seems to have vanished but Google's cache of it is here http://www.realhamradio.com/GPS_Frequency_Standard.htm Ah - this seems to be from the man himself. Info on design aspects and sources of PCB etc Also PIC code http://www.rt66.com/~shera/index_fs.htm More importantly, here is a copy of the original article http://www.rt66.com/~shera/QST_GPS.pdf FAROUT !!!!!!!!!!!!!! (as they say) A quick look at the circuit shows that the design is essentially trivial - although there is still of course work involved. Roughly it works like this (see article for how HE says it works :-) : The object is to lock a VCXO to the 1 second pulse. He uses a 1 second pulse from a commercial GPS unit. This may not be available externally on all units. He uses this pulse edge to set an RS flip flop which is then reset by the VCXO. A digital phase lock loop using a hardware counter external to the PIC is used along with a 1second PIC interrupt from the GPS to phase lock the PIC 6 MHz clock to the 1 second pulse. The PIC produces a digital error word to a serial DAC to produce an error voltage to drive the VCXO to maintain phase lock. _____________________ 1 Hz and 100 Hz output Motorola GPS board here (and related links to above article) http://www.tapr.org/tapr/html/Foncoreut.html Russell McMahon -- http://www.piclist.com hint: PICList Posts must start with ONE topic: [PIC]:,[SX]:,[AVR]: ->uP ONLY! [EE]:,[OT]: ->Other [BUY]:,[AD]: ->Ads

Russell McMahon <spam_OUTapptechTakeThisOuTPARADISE.NET.NZ> wrote: > The original product is not especially expensive, using mainly standard > ICs and a VCXO BUT the 1 Hz pulse output GPS is unusual. Not at all; most OEM modules include it. The Motorola UT+ modules are optimized for this use -- their spec on the 1pps is +/- 50 ns, IIRC. > PERHAPS: Using the leading edge of the NMEA serial stream which occurs > once a second would be good enough ??? This necessarily has an excellent > long term stability but short term may be excessively dependent on internal > processing in the GPS. Sure; just make sure you verify the actual time in the NMEA message in case it skips a second every now and then. As you say, the jitter will be much higher than that of the 1pps output. > At an absolute minimum the PIC could modify its own clock with a DAC and > varicap. "On frequency" decisions could be based on whether the NMEA edge > occurred prior to or before a program event which should take exactly one > second (or a submultiple thereof). You can also use DDS techniques to create a disciplined oscillator without modifying the CPU's clock, assuming the CPU's clock is inherently stable enough. I outlined the technique in a Circuit Cellar article: http://www.dtweed.com/circuitcellar/caj00084.htm#1266 (Unfortunately, this one isn't available online. If you have trouble tracking down a copy, let me know and I'll see what I can dig up.) -- Dave Tweed -- http://www.piclist.com hint: PICList Posts must start with ONE topic: [PIC]:,[SX]:,[AVR]: ->uP ONLY! [EE]:,[OT]: ->Other [BUY]:,[AD]: ->Ads

Russell McMahon <.....apptechKILLspam@spam@PARADISE.CounterSpam.NET.NZ> wrote; > >Build your own PIC controlled GPS freq std.Brooks Shera W5OJM, QST July >1998 >page 37 (how's that for an encyclopaedic memory?)(thanks Google). >The original seems to have vanished but Google's cache of it is here > > www.realhamradio.com/GPS_Frequency_Standard.htm > http://www.rt66.com/~shera/index_fs.htm > http://www.rt66.com/~shera/QST_GPS.pdf > Funny, I had just looked at the Shera article before stumbling across this thread. Great minds must think alike:^) Something about the way he takes the 1PPS signal and uses it as a PIC interrupt kinda grosses me out though. He then throws a lot of software at it but I don't know if that is the best solution. There must be a better way. The 1PPS edge is where the accuracy is. It should go to a high speed phase detector, not a PIC input. Generate your pristine oscillator signal and then run it into the PIC clock input or a frequency synthesizer. With all these low-cost GPS boards available (on Ebay and elsewhere) it would be really cool to come up with a low-cost programmable signal generator that could boast NIST class accuracy. _________________________________________________________________ Protect your PC - get McAfee.com VirusScan Online http://clinic.mcafee.com/clinic/ibuy/campaign.asp?cid=3963 -- http://www.piclist.com hint: PICList Posts must start with ONE topic: [PIC]:,[SX]:,[AVR]: ->uP ONLY! [EE]:,[OT]: ->Other [BUY]:,[AD]: ->Ads

j galt wrote: > > Russell McMahon <apptechKILLspamPARADISE.CounterSpam.NET.NZ> wrote; > >Build your own PIC controlled GPS freq std.Brooks Shera W5OJM, QST July > >1998 > >page 37 (how's that for an encyclopaedic memory?)(thanks Google). > >The original seems to have vanished but Google's cache of it is here > > > > www.realhamradio.com/GPS_Frequency_Standard.htm > > http://www.rt66.com/~shera/index_fs.htm > > http://www.rt66.com/~shera/QST_GPS.pdf > > > > Funny, I had just looked at the Shera article before stumbling across this > thread. Great minds must think alike:^) Something about the way he takes the > 1PPS signal and uses it as a PIC interrupt kinda grosses me out though. He > then throws a lot of software at it but I don't know if that is the best > solution. There must be a better way. Not really. Brooks is pretty much an 'expert' on this topic. If you'd browsed around his web site you'd see that he continues to work on ways of making a simpler and cheaper version of his precision reference. > The 1PPS edge is where the accuracy is. Not exactly. The 1PPS edge has significant jitter because it is generated by software (in most cases) and because there is still an intrinsic inaccuracy in the satellite measurement process. SA (selective availability) affects time a well as position (since SA is achieved by diddling the timing in the first place). > It should go to a high speed phase detector, not a PIC input. Generate He explains in the article/website why he does it the way he does. Remember the word 'inexpensive'. That also means 'readily available parts'. > your pristine oscillator signal and then run it into the PIC clock input or > a frequency synthesizer. Sure, if your 1PPS is accurately 1PPS to 1E13 or so. It isn't. Hence the software filtering to discipline the VCXO so that it's frequency is made more accurate over time, by creating a running average with a low pass filter with an incredibly long time constant (and not mentioning the correction needed for crystal aging). > With all these low-cost GPS boards available (on Ebay and elsewhere) it > would be really cool to come up with a low-cost programmable signal > generator that could boast NIST class accuracy. Sure it would, but most such GPS boards don't care about the precision of their 1PPS signal (if they even have one). Just look at the jitter spec to see the difference between a 'position' (+/- 250ns) receiver, and a 'timing' receiver (+/-10). I've been following Brook's (and others) work with great interest because of my interest in long baseline amateur radio astronomy. Getting an oscillator to NIST class accuracy is not cheap or easy or it would have already been done. Robert -- http://www.piclist.com hint: The list server can filter out subtopics (like ads or off topics) for you. See http://www.piclist.com/#topics

Robert Rolf <.....Robert.RolfKILLspam.....UALBERTA.Spammers.Are.Scum.CA> wrote in message news:<EraseME3E6436D5.179973ACspam_OUTTakeThisOuTUALBERTA.ca>... {Quote hidden}>j galt (Dave) wrote: > > Russell McMahon wrote; > >>Build your own PIC controlled GPS freq std.Brooks Shera W5OJM, > >>QST July 1998 page 37 (how's that for an encyclopaedic memory?) > >>(thanks Google). The original seems to have vanished but >>Google's >cache of it is here > >> > >> www.realhamradio.com/GPS_Frequency_Standard.htm > >> http://www.rt66.com/~shera/index_fs.htm > >> http://www.rt66.com/~shera/QST_GPS.pdf > >> > > > >Something about the way he takes the 1PPS signal and uses >it as a PIC >interrupt kinda grosses me out though. He > >then throws a lot of software at it but I don't know if that > >is the best solution. There must be a better way. > >Not really. Brooks is pretty much an 'expert' on this topic. If you'd >browsed around his web site you'd see that he continues >to work on ways of making a simpler and cheaper version of his >precision reference. I did not see this, I also did not notice any error analysis or discussion. > > The 1PPS edge is where the accuracy is. > >Not exactly. The 1PPS edge has significant jitter because it is generated >by software (in most cases) and because there is still an intrinsic >inaccuracy in the satellite measurement process. SA (selective >availability) affects time a well as position (since SA is achieved by >diddling the timing in the first place). Yes there will be jitter but you don't improve things by adding even coarser amounts of jitter. SA has been turned off for years, or do you watch the news? > > It should go to a high speed phase detector, not a PIC input. > >He explains in the article/website why he does it the way he does. Remember >the word 'inexpensive'. That also means 'readily >available parts'. I will read it more closely. There are parts available in Digikey that are quite available and quite fast. > > > With all these low-cost GPS boards available (on Ebay and > > elsewhere) it would be really cool to come up with a low-cost > >programmable signal generator that could boast NIST class > accuracy. >Sure it would, but most such GPS boards don't care about the >precision of their 1PPS signal (if they even have one). Nearly all have 1PPS, but you are right. >Just look at the jitter spec to see the difference between a >'position' (+/- 250ns) receiver, and a 'timing' receiver (+/-10). > >I've been following Brook's (and others) work with great interest >because of my interest in long baseline amateur radio astronomy. >Getting an oscillator to NIST class accuracy is not cheap or easy >or it would have already been done. Well, no, but how much accuracy does the application require? And how do Brooks and others determine whether their GPSDO's are working properly? Who has NIST class reference sources available for testing? _________________________________________________________________ STOP MORE SPAM with the new MSN 8 and get 2 months FREE* http://join.msn.com/?page=features/junkmail -- http://www.piclist.com hint: The list server can filter out subtopics (like ads or off topics) for you. See http://www.piclist.com/#topics

j galt <galt_57spam_OUTHOTMAIL.COM> wrote: > Yes there will be jitter but you don't improve things by adding even > coarser amounts of jitter. Have you ever heard of the concept called dithering? It doesn't harm things, and it often improves the statistics of the signal considerably, making it easier to achieve precision locking with a proper loop filter. > And how do Brooks and others determine whether their GPSDO's are working > properly? Who has NIST class reference sources available for testing? Anyone with a shortwave receiver. Even though the signals from WWV et al have lots of short-term jitter and wander, and even an unknown constant time offset, the long-term frequency accuracy is directly traceable to NIST. And of course, you can always compare your disciplined clock with the raw 1 pps pulses from the GPS receiver. Seriously, commercial GPS-disciplined oscillators are quite common in the telecom and broadcast industries. Most of the cost and complexity in these units is associated with having a local oscillator that has low inherent phase noise and good long-term holdover characteristics for when the GPS signal is unavailable (e.g., measuring and predicting the oscillator aging characteristics). The actual frequency locking is quite straightforward. -- Dave Tweed -- http://www.piclist.com hint: The list server can filter out subtopics (like ads or off topics) for you. See http://www.piclist.com/#topics

Dave Tweed <@spam@picKILLspamDTWEED.CounterSpam.COM> wrote; >j galt <KILLspamgalt_57KILLspamHOTMAIL.Take.This.Out.COM> wrote: > > Yes there will be jitter but you don't improve things by adding > > even coarser amounts of jitter. > >Have you ever heard of the concept called dithering? It doesn't >harm things, and it often improves the statistics of the signal >considerably... Do you know of a reference which suggests dithering the input of a PLL? I think I just misunderstood the approach that this design used. The 1PPS signal does not go to the PIC interrupt. The output of the HCT4046 phase detector goes to the PIC interrupt input, which makes much more sense. > > And how do Brooks and others determine whether their GPSDO's > are >working properly? Who has NIST class reference sources... > >Anyone with a shortwave receiver. Even though the signals from WWV et al >have lots of short-term jitter and wander, and even an unknown constant >time offset, the long-term frequency accuracy is directly traceable to >NIST. Sounds interesting but the QST article dismisses this method. How about the NTSC color burst? Does anyone remember that being suggested as a calibration source? >And of course, you can always compare your disciplined clock with the raw 1 >pps pulses from the GPS receiver. From my reading of the article you can look to see if the curve of the correction voltage matches that which would be expected to correct crystal aging. Now what I would like to know is what are the most common and most interesting applications for such a precision frequency reference? _________________________________________________________________ The new MSN 8: advanced junk mail protection and 2 months FREE* http://join.msn.com/?page=features/junkmail -- http://www.piclist.com hint: The list server can filter out subtopics (like ads or off topics) for you. See http://www.piclist.com/#topics

Russell McMahon <RemoveMEapptechTakeThisOuTPARADISE.CounterSpam.NET.NZ> wrote; > >Build your own PIC controlled GPS freq std.Brooks Shera W5OJM, QST July >1998 page 37 (how's that for an encyclopaedic memory?) >(thanks Google). The original seems to have vanished but Google's >cache of it is here > > www.realhamradio.com/GPS_Frequency_Standard.htm > http://www.rt66.com/~shera/index_fs.htm > http://www.rt66.com/~shera/QST_GPS.pdf > Wait a minute. How does the Shera thing compare to an HP Z3801A ? _________________________________________________________________ Tired of spam? Get advanced junk mail protection with MSN 8. http://join.msn.com/?page=features/junkmail -- http://www.piclist.com hint: The list server can filter out subtopics (like ads or off topics) for you. See http://www.piclist.com/#topics

> >Build your own PIC controlled GPS freq std.Brooks Shera W5OJM, QST July > >1998 page 37 (how's that for an encyclopaedic memory?) > >(thanks Google). The original seems to have vanished but Google's > >cache of it is here > > > > www.realhamradio.com/GPS_Frequency_Standard.htm > > http://www.rt66.com/~shera/index_fs.htm > > http://www.rt66.com/~shera/QST_GPS.pdf > > > > Wait a minute. How does the Shera thing compare to an HP Z3801A ? Not as good. The HP has a superbly stable oscillator to start with which a home built unit of the complexity of Brooks' unit can only aim for. That said, there are liable to be few amateur applications where the difference would be noticed. The HP would always be obtained surplus but is probably dearer, larger, rather harder to repair if it breaks (as you didn't build it), can't be as easily adapted to other uses and can't be replicated by your friends once they see how good yours is. But the HP has "HP" on the front (and/or back) which is always a major consideration. RM -- http://www.piclist.com hint: The PICList is archived three different ways. See http://www.piclist.com/#archives for details.

Russell McMahon <spamBeGoneapptechspamBeGonePARADISE.Spammers.Are.Scum.NET.NZ> wrote; > > >Build your own PIC controlled GPS freq std.Brooks Shera W5OJM, > >QST >July 1998 page 37 (how's that for an encyclopaedic memory?) > > >(thanks Google). The original seems to have vanished but > >Google's >cache of it is here > > > > > > www.realhamradio.com/GPS_Frequency_Standard.htm > > > http://www.rt66.com/~shera/index_fs.htm > > > http://www.rt66.com/~shera/QST_GPS.pdf > > > > > > > Wait a minute. How does the Shera thing compare to an HP Z3801A ? > >Not as good. >The HP has a superbly stable oscillator to start with which a home built >unit of the complexity of Brooks' unit can only aim for. That said, there >are liable to be few amateur applications where the difference would be >noticed. Actually no, it seems that Shera's thing must be better. I bet there is considerable confusion about this. See; http://www.jrmiller.demon.co.uk/projects/freqstd/frqstd.htm _________________________________________________________________ Tired of spam? Get advanced junk mail protection with MSN 8. http://join.msn.com/?page=features/junkmail -- http://www.piclist.com hint: The PICList is archived three different ways. See http://www.piclist.com/#archives for details.