GPS for Nixie Clock
I'm thinking about converting a Nixie clock I built years ago into using
GPS for the time base. No real NEED, just for fun.
The clock uses an FPGA for formatting and display, using the 60Hz line
frequency as the time base. The case is a single hollowed out block of
walnut.
I'm looking into a TU36-D400-020 receiver. This seems to be optimized
for timing purposes rather than navigation, it has 1PPS and 10KHz outputs.
I'd be getting it from RDR Electronics, which says it it uses the
Motorola command set. This seems fine for me, it has the information I
need, specifically UTC time so I don't have to worry about leap seconds.
I have a few questions about this receiver:
The data sheet lists two serial ports, but I don't see any information
about which to use. Are they identical, do I have to use one for some
functions and the other for other purposes?
What are the serial port parameters? 9600-8-N-1? Or something else?
Which is better to use, the 1PPS or the 10KHz? I can easily go either
way. The clock display just goes down to seconds so 1PPS would work. I
could also re-clock the 1PPS with the 10KHz.
What antenna to use? I would prefer something mounted inside the case.
It is wood so an internal antenna will hopefully work. The board comes
with a pigtail but it is not SMA.
Any other hints for using this?
I've never done a GPS interface before so I'm not sure about how I
calibrate the time coming from the message over the serial port. Is it
something like "the time is such and such at the rising edge of the next
PPS, or the previous one? Or is there some other mechanism for
calibrating when the second changes on the display to something close to
reality?
I previously toyed with the idea of using an X72 rubidium oscillator
just for the bragging rights, but I would still need the GPS to get the
time, I decided the TU36 on its own is probably just fine.
Thanks,
John S.
Well, it's late and I've nothing else to do.
I sympathize with you, but my Nixie clock is a computer display made
years ago by Jag Air Clockvault. One of these Windows OS upgrades is
going to refuse to run it.
You say it has an FPGA for the display. What provides the numbers to be
formatted?
A line frequency clock has no time info - it has to be set. Do you have
raise/lower buttons or just raise?
Good luck doing a simple conversion from GPS receiver time to numbers
the FPGA can convert. Much simpler to count 1 PPS pulses, but like line
frequency, there's no adjustment for leap seconds.
Reclocking the 1 PPS with 10 KHz seems like overkill for a clock to be
read by humans.
GPS reception and accuracy falls off quickly if the antenna can't see a
large part of the sky. It won't function if it can't see several
satellites. I know this empirically.
WWVB has no such requirement, and it's accuracy is a better match for
human precision.
Could you give us a link to a picture of your solid walnut Nixie clock?
Got any more walnut?
Bill Hawkins
-----Original Message-----
From: John Swenson
Sent: Sunday, July 10, 2016 10:55 PM
I'm thinking about converting a Nixie clock I built years ago into using
GPS for the time base. No real NEED, just for fun.
The clock uses an FPGA for formatting and display, using the 60Hz line
frequency as the time base. The case is a single hollowed out block of
walnut.
I'm looking into a TU36-D400-020 receiver. This seems to be optimized
for timing purposes rather than navigation, it has 1PPS and 10KHz
outputs.
I'd be getting it from RDR Electronics, which says it it uses the
Motorola command set. This seems fine for me, it has the information I
need, specifically UTC time so I don't have to worry about leap seconds.
I have a few questions about this receiver:
The data sheet lists two serial ports, but I don't see any information
about which to use. Are they identical, do I have to use one for some
functions and the other for other purposes?
What are the serial port parameters? 9600-8-N-1? Or something else?
Which is better to use, the 1PPS or the 10KHz? I can easily go either
way. The clock display just goes down to seconds so 1PPS would work. I
could also re-clock the 1PPS with the 10KHz.
What antenna to use? I would prefer something mounted inside the case.
It is wood so an internal antenna will hopefully work. The board comes
with a pigtail but it is not SMA.
Any other hints for using this?
I've never done a GPS interface before so I'm not sure about how I
calibrate the time coming from the message over the serial port. Is it
something like "the time is such and such at the rising edge of the next
PPS, or the previous one? Or is there some other mechanism for
calibrating when the second changes on the display to something close to
reality?
I previously toyed with the idea of using an X72 rubidium oscillator
just for the bragging rights, but I would still need the GPS to get the
time, I decided the TU36 on its own is probably just fine.
Thanks,
John S.
Hi
For $15 to $45 a number of places will sell you a uBlox receiver card that runs off of 5V and has an internal antenna. They are quite sensitive and have a timing output. They are plenty good enough for what you are trying to do. The practical issue is getting good enough GPS signals at an arbitrary location inside a house. To do that, you want a sensitive receiver.
No matter which receiver you get, they all put out fairly complex serial strings that need to be parsed to get the time information. For a wall clock the serial data “time of arrival” will be better than your eye can see. The PPS output is overkill, but this is Time Nuts. I would do the parsing with a small MCU.
I doubt that an older FPGA will have enough “stuff” in it to do the job. You will need a few hundred bytes of RAM for buffering and not a whole lot else. There are thousands of different MCU’s that can / will / could / might do the job. Getting one already built up on a board for < $10 is pretty easy. Getting one on a board for < $3 is possible.
Lots of choices.
Bob
On Jul 10, 2016, at 11:55 PM, John Swenson johnswenson1@comcast.net wrote:
I'm thinking about converting a Nixie clock I built years ago into using GPS for the time base. No real NEED, just for fun.
The clock uses an FPGA for formatting and display, using the 60Hz line frequency as the time base. The case is a single hollowed out block of walnut.
I'm looking into a TU36-D400-020 receiver. This seems to be optimized for timing purposes rather than navigation, it has 1PPS and 10KHz outputs.
I'd be getting it from RDR Electronics, which says it it uses the Motorola command set. This seems fine for me, it has the information I need, specifically UTC time so I don't have to worry about leap seconds.
I have a few questions about this receiver:
The data sheet lists two serial ports, but I don't see any information about which to use. Are they identical, do I have to use one for some functions and the other for other purposes?
What are the serial port parameters? 9600-8-N-1? Or something else?
Which is better to use, the 1PPS or the 10KHz? I can easily go either way. The clock display just goes down to seconds so 1PPS would work. I could also re-clock the 1PPS with the 10KHz.
What antenna to use? I would prefer something mounted inside the case. It is wood so an internal antenna will hopefully work. The board comes with a pigtail but it is not SMA.
Any other hints for using this?
I've never done a GPS interface before so I'm not sure about how I calibrate the time coming from the message over the serial port. Is it something like "the time is such and such at the rising edge of the next PPS, or the previous one? Or is there some other mechanism for calibrating when the second changes on the display to something close to reality?
I previously toyed with the idea of using an X72 rubidium oscillator just for the bragging rights, but I would still need the GPS to get the time, I decided the TU36 on its own is probably just fine.
Thanks,
John S.
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
Last year I created GPS-based Nixie clock for myself. After all the
considerations I decide to use GPS receiver with external antenna. Not
so aesthetic, but more reliable for me. I was using DS32xx chip as a
clock. And GPS plus 1PPS output from GPS to set/sync that DS32xx. So,
GPS module is not critical there, as clock will do its work without GPS.
Also I implemented "OR" gate to be able to do 1PPS sync with external
1PPS source. Just in case.
The DS32xx is OK for this project. Since its very unlikely it will drift
so much.
Late, I found myself thinking that NTP protocol instead of GPS will be
better solution for that project. For example, I found that sometimes
the GPS clock is set the time by 16 second off the UTC. Likely its
because GPS getting the report without of "leap seconds". Its still
puzzle for me why its happens. I saw it couple of times for the past
year. So, its rare, but it happens.
Here is my clock "during" and "after":
https://www.youtube.com/watch?v=QTRY4YPGZQc
https://www.youtube.com/watch?v=GScyoA8gCzY
On 2016-07-10 23:55, John Swenson wrote:
I'm thinking about converting a Nixie clock I built years ago into
using GPS for the time base. No real NEED, just for fun.
The clock uses an FPGA for formatting and display, using the 60Hz line
frequency as the time base. The case is a single hollowed out block of
walnut.
I'm looking into a TU36-D400-020 receiver. This seems to be optimized
for timing purposes rather than navigation, it has 1PPS and 10KHz
outputs.
I'd be getting it from RDR Electronics, which says it it uses the
Motorola command set. This seems fine for me, it has the information I
need, specifically UTC time so I don't have to worry about leap
seconds.
I have a few questions about this receiver:
The data sheet lists two serial ports, but I don't see any information
about which to use. Are they identical, do I have to use one for some
functions and the other for other purposes?
What are the serial port parameters? 9600-8-N-1? Or something else?
Which is better to use, the 1PPS or the 10KHz? I can easily go either
way. The clock display just goes down to seconds so 1PPS would work. I
could also re-clock the 1PPS with the 10KHz.
What antenna to use? I would prefer something mounted inside the case.
It is wood so an internal antenna will hopefully work. The board comes
with a pigtail but it is not SMA.
Any other hints for using this?
I've never done a GPS interface before so I'm not sure about how I
calibrate the time coming from the message over the serial port. Is it
something like "the time is such and such at the rising edge of the
next PPS, or the previous one? Or is there some other mechanism for
calibrating when the second changes on the display to something close
to reality?
I previously toyed with the idea of using an X72 rubidium oscillator
just for the bragging rights, but I would still need the GPS to get
the time, I decided the TU36 on its own is probably just fine.
Thanks,
John S.
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to
https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
--
WBW,
V.P.
On Sun, Jul 10, 2016 at 8:55 PM, John Swenson johnswenson1@comcast.net
wrote:
I'm thinking about converting a Nixie clock I built years ago into using
GPS for the time base. No real NEED, just for fun.
Which is better to use, the 1PPS or the 10KHz? I can easily go either way.
The clock display just goes down to seconds so 1PPS would work. I could
also re-clock the 1PPS with the 10KHz.
You have to use the 1PPS. It is in phase with the UTC second. Your
seconds display increments with every PPS pulse. If you were to use
10KHz, how would you know which of those 10,000 pulses was the UTC second
"tick"
The way it works with GPS is you first de-code the message from the serial
port and hold that data in some internal memory then wait for the PPS
pulse. When the pulse happens you move the internal memory data to the
display. I don't know what you'd do with 10KHz except divide it by 10,000
to create your own 1PPS but how to get it to "tick" on the exact UTC second?
If you were displaying tenths of seconds then you'd need the 10K output
(divided by 1,000) to drive a counter that gets reset to zero by the PPS.
Chris Albertson
Redondo Beach, California
It's fairly straight forward to build a 1 Hz NCO with a timer/counter block
using your processor's LO. Threshold - Frequency, count - Phase. You can
then DPLL to either PPS, Serial Data, or both. If your receiver drops its
NAV or timing solution you still have "time".
On Sun, Jul 10, 2016 at 11:55 PM, John Swenson johnswenson1@comcast.net
wrote:
I'm thinking about converting a Nixie clock I built years ago into using
GPS for the time base. No real NEED, just for fun.
The clock uses an FPGA for formatting and display, using the 60Hz line
frequency as the time base. The case is a single hollowed out block of
walnut.
I'm looking into a TU36-D400-020 receiver. This seems to be optimized for
timing purposes rather than navigation, it has 1PPS and 10KHz outputs.
I'd be getting it from RDR Electronics, which says it it uses the Motorola
command set. This seems fine for me, it has the information I need,
specifically UTC time so I don't have to worry about leap seconds.
I have a few questions about this receiver:
The data sheet lists two serial ports, but I don't see any information
about which to use. Are they identical, do I have to use one for some
functions and the other for other purposes?
What are the serial port parameters? 9600-8-N-1? Or something else?
Which is better to use, the 1PPS or the 10KHz? I can easily go either way.
The clock display just goes down to seconds so 1PPS would work. I could
also re-clock the 1PPS with the 10KHz.
What antenna to use? I would prefer something mounted inside the case. It
is wood so an internal antenna will hopefully work. The board comes with a
pigtail but it is not SMA.
Any other hints for using this?
I've never done a GPS interface before so I'm not sure about how I
calibrate the time coming from the message over the serial port. Is it
something like "the time is such and such at the rising edge of the next
PPS, or the previous one? Or is there some other mechanism for calibrating
when the second changes on the display to something close to reality?
I previously toyed with the idea of using an X72 rubidium oscillator just
for the bragging rights, but I would still need the GPS to get the time, I
decided the TU36 on its own is probably just fine.
Thanks,
John S.
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to
https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.