Forums

GPS module wanted for timing.

Started by Clive Arthur July 11, 2018
On 11/07/2018 15:05, Reinhardt Behm wrote:
> AT Wednesday 11 July 2018 17:02, Clive Arthur wrote: > >> Hi >> >> I need a GPS module which will give me: >> ...an accurate 10 MHz clock output >> ...an accurate 1 second pulse output >> ...a serial interface to read UTC. >> >> (Position isn't important. I know where I am.) >> >> I know there are probably dozens of modules which will do this, but I >> have no experience in this area, so any advice would be welcome. UK >> supplier preferred but not strongly. Will need only a few. >> >> Cheers > > uBlox has modules for that > <https://www.u-blox.com/en/product/neolea-m8t-series> >
These look very good, thanks. Cheers -- Clive
On Wed, 11 Jul 2018 10:02:22 +0100, Clive Arthur
<cliveta@nowaytoday.co.uk> wrote:

>I need a GPS module which will give me: >...an accurate 10 MHz clock output >...an accurate 1 second pulse output >...a serial interface to read UTC. > >(Position isn't important. I know where I am.) > >I know there are probably dozens of modules which will do this, but I >have no experience in this area, so any advice would be welcome. UK >supplier preferred but not strongly. Will need only a few. > >Cheers
The 1PPS and serial (4800 baud) outputs are easy. The 10 MHz clock output is not. For that, you will probably need to build or buy a GPSDO (GPS diciplined oscillator). There are designs for a GPSDO available that are really crude and simple if you don't need the ultimate in stability, and others that take a week to stabilize and offer more accuracy and stability than you could possibly need. I built an OCXO (oven controlled crystal oscillator) with a 1, 5, and 10 MHz outputs to run my pile of test equipment. You can buy Chinese OCXO based GPSDO bricks on eBay for about $120 and up: <https://www.ebay.com/sch/i.html?_nkw=10+mhz+gpsdo> As others have mentioned, the place to get accurate info is the Time-Nuts mailing list: <http://leapsecond.com/time-nuts.htm> Archive: <https://lists.febo.com/pipermail/time-nuts_lists.febo.com/> -- Jeff Liebermann jeffl@cruzio.com 150 Felker St #D http://www.LearnByDestroying.com Santa Cruz CA 95060 http://802.11junk.com Skype: JeffLiebermann AE6KS 831-336-2558
On 11/07/2018 17:04, Jeff Liebermann wrote:
> On Wed, 11 Jul 2018 10:02:22 +0100, Clive Arthur > <cliveta@nowaytoday.co.uk> wrote: > >> I need a GPS module which will give me: >> ...an accurate 10 MHz clock output >> ...an accurate 1 second pulse output >> ...a serial interface to read UTC. >> >> (Position isn't important. I know where I am.) >> >> I know there are probably dozens of modules which will do this, but I >> have no experience in this area, so any advice would be welcome. UK >> supplier preferred but not strongly. Will need only a few. >> >> Cheers > > The 1PPS and serial (4800 baud) outputs are easy. The 10 MHz clock > output is not. For that, you will probably need to build or buy a > GPSDO (GPS diciplined oscillator). There are designs for a GPSDO > available that are really crude and simple if you don't need the > ultimate in stability, and others that take a week to stabilize and > offer more accuracy and stability than you could possibly need. I > built an OCXO (oven controlled crystal oscillator) with a 1, 5, and 10 > MHz outputs to run my pile of test equipment. > > You can buy Chinese OCXO based GPSDO bricks on eBay for about $120 and > up: > <https://www.ebay.com/sch/i.html?_nkw=10+mhz+gpsdo> > > As others have mentioned, the place to get accurate info is the > Time-Nuts mailing list: > <http://leapsecond.com/time-nuts.htm> > Archive: > <https://lists.febo.com/pipermail/time-nuts_lists.febo.com/> >
These look like they might do it well enough for me... https://www.u-blox.com/en/product/neolea-m8t-series ...what do you think? Cheers -- Clive
On Wed, 11 Jul 2018 17:50:54 +0100, Clive Arthur
<cliveta@nowaytoday.co.uk> wrote:

>On 11/07/2018 17:04, Jeff Liebermann wrote: >> On Wed, 11 Jul 2018 10:02:22 +0100, Clive Arthur >> <cliveta@nowaytoday.co.uk> wrote: >> >>> I need a GPS module which will give me: >>> ...an accurate 10 MHz clock output >>> ...an accurate 1 second pulse output >>> ...a serial interface to read UTC. >>> >>> (Position isn't important. I know where I am.) >>> >>> I know there are probably dozens of modules which will do this, but I >>> have no experience in this area, so any advice would be welcome. UK >>> supplier preferred but not strongly. Will need only a few. >>> >>> Cheers >> >> The 1PPS and serial (4800 baud) outputs are easy. The 10 MHz clock >> output is not. For that, you will probably need to build or buy a >> GPSDO (GPS diciplined oscillator). There are designs for a GPSDO >> available that are really crude and simple if you don't need the >> ultimate in stability, and others that take a week to stabilize and >> offer more accuracy and stability than you could possibly need. I >> built an OCXO (oven controlled crystal oscillator) with a 1, 5, and 10 >> MHz outputs to run my pile of test equipment. >> >> You can buy Chinese OCXO based GPSDO bricks on eBay for about $120 and >> up: >> <https://www.ebay.com/sch/i.html?_nkw=10+mhz+gpsdo> >> >> As others have mentioned, the place to get accurate info is the >> Time-Nuts mailing list: >> <http://leapsecond.com/time-nuts.htm> >> Archive: >> <https://lists.febo.com/pipermail/time-nuts_lists.febo.com/>
>These look like they might do it well enough for me... >https://www.u-blox.com/en/product/neolea-m8t-series >...what do you think? >Cheers
uBlox makes good boards and it will probably work. However, you didn't mention if you are designing a product, or building a single unit for your own use. uBlox seems to sell modules, not complete systems. I guess their development system would be considered complete, but that probably costs more than you want to spend. I think you can buy these modules on a PCB which should be easier. I looked at the specs <https://www.u-blox.com/sites/default/files/NEO-LEA-M8T-FW3_DataSheet_%28UBX-15025193%29.pdf> and found that the NEO-M8T and LEA-M8T have 2 outputs which can be individually set to 0.25Hz and 10MHz. The diagram shows a 1PPS output, but it looks like it's shared with the two other outputs. I don't have time to dig deeper right now. I think it best read the 35 page data sheet, check for additional documents: <https://www.u-blox.com/en/product-resources> and then ask for advice on Time-Nuts mailing list for confirmation or alternatives. Also, I don't know your application for the 10MHz, but it has been my experience that when the satellite(s) disappear, the 10MHz output will tend to drift around. If can be a problem if you're using it to control test equipment clock frequency. You could add a slow PLL to stabilize it, or use a much more stable external OCXO to provide the 10MHz. If it looses lock for a few minutes, the output will not drift very far or very much. Anyway, check to see what the 10MHz output does when the satellite(s) disappear. -- Jeff Liebermann jeffl@cruzio.com 150 Felker St #D http://www.LearnByDestroying.com Santa Cruz CA 95060 http://802.11junk.com Skype: JeffLiebermann AE6KS 831-336-2558
On Wed, 11 Jul 2018 10:51:02 -0700, Jeff Liebermann <jeffl@cruzio.com>
wrote:

More:

Perhaps you should be looking for this module on a PCB.
<https://www.mouser.com/Embedded-Solutions/Engineering-Tools/RF-Wireless-Development-Tools/GPS-Development-Tools/_/N-90o44/>
<https://www.digikey.com/products/en/development-boards-kits-programmers/evaluation-boards-expansion-boards-daughter-cards/797?k=&pkeyword=&pv110=1020&FV=ffe0031d&quantity=0&ColumnSort=0&page=1&pageSize=25>


-- 
Jeff Liebermann     jeffl@cruzio.com
150 Felker St #D    http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann     AE6KS    831-336-2558
On Wed, 11 Jul 2018 08:24:01 -0700, John Larkin
<jjlarkin@highlandtechnology.com> wrote:

>On Wed, 11 Jul 2018 12:23:34 GMT, Johnny B Good ><johnny-b-good@invalid.ntlworld.com> wrote: > >>On Wed, 11 Jul 2018 02:51:02 -0700, jack4747 wrote: >> >>> On Wednesday, July 11, 2018 at 11:02:27 AM UTC+2, Clive Arthur wrote: >>>> Hi >>>> >>>> I need a GPS module which will give me: >>>> ...an accurate 10 MHz clock output ...an accurate 1 second pulse output >>> >>> What do you mean by "accurate"? >>> >> Normally, in the context of GPS, the word "accurate" would normally be >>regarded as being somewhat redundant. Presumably an accuracy measured in >>nanoseconds is being sought. :-) > >The fundamental GPS output is the 1 PPS pulse.
What is so "fundamental" about the 1PPS output. After all the GPS signal repetition rate is 1 kHz, so a 1 kHz output would be more useful for faster locking.
>It jitters and wanders >around a lot, as satellites come and go and the atmosphere changes.
As long a the receiving station remains in a fixed position, this should not really be an issue.
>It's accurate long-term, so you can lock a very good oscillator to it, >slowly, and get a good 10 MHz. The slower the lock loop and the better >the oscillator, the better 10M. A rubidium is good.
You need weeks to lock to 1PPS signals.
> >Cheap GPS receivers can have 1PPS jitter in the ballpark of a >microsecond.
Or even worse. The 1PPS is often software/firmware derived.
>I was wondering, I have several cheap ebay GPS modules with 1 Hz output >that support both Chinese, Russian, and US GPS. >30 $ or so on ebay, but no 10MHz output. >Been probing around in one of those (needed fixing) there _is_ a crystal, but not sure it is locked. >Common chips... datasheet is available, cannot remember the exact number.
I built one of the VE2ZAZ 10 MHz GPSDO boards, a few years ago... very flexible, can be configured to work with various GPS boards. I originally used a Motorola UT. When that board died, a couple of years ago, I replaced it with a uBlox board (an eBay cheapie apparently intended for use in a drone) and it's back to working just fine. I (and another guy) heavily reworked the VE2ZAZ firmware - I turned my version into a PI-type controller. It worked out quite well.
>I am thinking it should not be impossible >to build a cheap divider and xtal oscillator and then phase lock it >to the 1 pps (needs oven temperature stabilization).
That's basically what the VE2ZAZ and the earlier Brooks Shera designs do, using a PIC or similar micro plus some discrete components. https://www.sv1afn.com/gpsdo.html is an interesting approach, as it produces both a 10 MHz signal, and a second frequency which is locked to the 10 MHz. I don't think you can get 1 PPS out of it (it switches the internal GPS from 1 PPS to 10 MHz) but you could set the second output to e.g. 100 Hz and then use an external divider to give you 1 Hz.
upsidedown@downunder.com wrote in news:rkmckd1jegcpfoo75pank6p4nt8kkguieh@
4ax.com:

> What is so "fundamental" about the 1PPS output. After all the GPS > signal repetition rate is 1 kHz, so a 1 kHz output would be more > useful for faster locking.
It is fundemental inasmuch as it is the standard adopted by the scientific community and the military for time synchronisation of equipment.
On Wed, 11 Jul 2018 22:37:08 +0300, upsidedown@downunder.com wrote:

>On Wed, 11 Jul 2018 08:24:01 -0700, John Larkin ><jjlarkin@highlandtechnology.com> wrote: > >>On Wed, 11 Jul 2018 12:23:34 GMT, Johnny B Good >><johnny-b-good@invalid.ntlworld.com> wrote: >> >>>On Wed, 11 Jul 2018 02:51:02 -0700, jack4747 wrote: >>> >>>> On Wednesday, July 11, 2018 at 11:02:27 AM UTC+2, Clive Arthur wrote: >>>>> Hi >>>>> >>>>> I need a GPS module which will give me: >>>>> ...an accurate 10 MHz clock output ...an accurate 1 second pulse output >>>> >>>> What do you mean by "accurate"? >>>> >>> Normally, in the context of GPS, the word "accurate" would normally be >>>regarded as being somewhat redundant. Presumably an accuracy measured in >>>nanoseconds is being sought. :-) >> >>The fundamental GPS output is the 1 PPS pulse. > >What is so "fundamental" about the 1PPS output. After all the GPS >signal repetition rate is 1 kHz, so a 1 kHz output would be more >useful for faster locking.
There is no "repetition rate". The C/A codes are sent at rate of 1.023 Mbit/sec which is modulated by navigation messages at 50 bits/sec.
>>It jitters and wanders >>around a lot, as satellites come and go and the atmosphere changes. > >As long a the receiving station remains in a fixed position, this >should not really be an issue.
Ummm... The satellites are moving. However, even if both the satellites and receiver are stationary, atmospheric delays are quite erratic, unpredictable, and noisy. From my experience living in a forest, where foliage blockage by trees is a problem, when the data source moves from one satellite to a different satellite, there's a small but noticeable phase glitch. If my GPSDO loses lock, the OCXO clock oscillator drifts off frequency for the duration of the outage and then takes about the same amount of time as the outage to return to a stable lock condition. However, when I played with a Cesium secondary clock GPSDO, where I could theoretically get to 1 part in 10^14 accuracy (about 4 nsec/day), it did take days for things to settle down. I never got close to this level of accuracy because 60Hz power line and 120Hz power supply noise was wrecking my measurements. If it had works, it probably would have taken several days to stabilize.
>>It's accurate long-term, so you can lock a very good oscillator to it, >>slowly, and get a good 10 MHz. The slower the lock loop and the better >>the oscillator, the better 10M. A rubidium is good. > >You need weeks to lock to 1PPS signals.
Locking to 1pps was the way GPS time and freq clock receivers worked before built in higher frequency clock outputs were available. The time to gain lock was roughly proportional to how far out of phase the PLL clock was with the 1pps output. I don't recall the exact time, but I think it was less than 2 mins if the oscillator had warmed up, and maybe 20 mins if it was from a cold start. These use the 1pps output: "1pps locking module" <http://www.quartzlock.com/product/timing-modules/1pps-locking-modules/A6-1PPS> "GPS-Locked Frequency Standard" <http://www.techlib.com/electronics/GPSstandard.htm>
>>Cheap GPS receivers can have 1PPS jitter in the ballpark of a >>microsecond. > >Or even worse. The 1PPS is often software/firmware derived.
Not a problem because it has 1 full second to do the computation necessary to run whatever DSP filter algorithm (usually Kalman) is needed to reduce the noise and speed up locking by predicting the Doppler shift from each satellite. -- Jeff Liebermann jeffl@cruzio.com 150 Felker St #D http://www.LearnByDestroying.com Santa Cruz CA 95060 http://802.11junk.com Skype: JeffLiebermann AE6KS 831-336-2558
On Wed, 11 Jul 2018 15:19:37 -0700, Jeff Liebermann <jeffl@cruzio.com>
wrote:

Sigh.  I found my notes.  Looks like my memory is failing.  Some
corrections.

>>You need weeks to lock to 1PPS signals.
With a Cesium clock, possibly. With an OCXO, much less.
>I don't recall the exact time, >but I think it was less than 2 mins if the oscillator had warmed up, >and maybe 20 mins if it was from a cold start.
Faulty memory. About 14 hrs from a cold start for the OCXO oscillator using a 1pps reference. This probably varies radically depending on ambient temperature and equipment temperature, but I only did one run. About 30 minutes if the OCXO had warmed up and had been locked for a while. Sorry about the memory failure. Drivel: Visual GPS Software: <http://www.visualgps.net> NMEATime2 - PC GPS Time Synchronization <http://www.visualgps.net/#nmeatime2-content> $20.48USD, 30 day free trial. -- Jeff Liebermann jeffl@cruzio.com 150 Felker St #D http://www.LearnByDestroying.com Santa Cruz CA 95060 http://802.11junk.com Skype: JeffLiebermann AE6KS 831-336-2558