GPS pi4 incorrect date

[mbalkham]

Hi all. First time post from novice user.

I've installed gspd and chrony on a pi4 with the aim of syncing date, time and location for astronony.

I ran the same process on pi3 b+ all works fine. But on the pi4 gpsmon reads date of 2000-01-23T08:04 and chrony syncs the pi date and time to that date.

Any ideas?

Thanks

[topguy]

What is the GPS ? Are you using the same GPS with the Pi3 as the Pi4 ?

It might look like your GPS is rather old and has a bug that makes it report the date wrong.

If you can look at raw gps data you might find lines like this:

Code: Select all

$GPRMC,072404.051,A,6325.298234,N,01023.692289,E,0.700,32.08,170100,,,A*6C

My example are from a few days ago, so the date is 170100 ( 17.Jan 2000 ).

This problem started 6th of April this year when a 10bit counter for week number in GPS satellite data flipped from max value 1023 to 0.
Those GPS that haven't taken this into account in their firmware are now showing a date 1024 weeks ( 7168 days ) too old.

[mbalkham]

The GPS is the same physical usb device. I plug it into the pi3, boot and check gpsmon all ok. Then switch to pi4, remove usb from pi3 and connect to pi4 boot and wrong date is displayed in gpsmon.

I think all the settings/ config files are the same (at least I've made the same changes).

I'll look at the raw data later. I can't understand why if it is a 6th of April thing why I get a different result on each pi?

[topguy]

I can't understand why if it is a 6th of April thing why I get a different result on each pi?

No, that is definitely not obvious, but it fits quite well.

Are you using different SD cards ?
Is it the same version of Raspbian on both SD cards ?

Ideas:
- GPSD has probably some functionality to handle this problem, question is if its a config thing or if it can autodetect it.
- Maybe GPSD on the Pi3 has some way to detect that the date is wrong ? It has (had) internet connection and can ntp sync ?

[mbalkham]

Different sd cards with files from Stellarmate (as part of the 1.4.2 Stellarmate OS for the Pi3 and 1.4.3 for the Pi4). I'll message them to see if they have any ideas. I'll check which version of Rasabian is running on each when I get home.

Both have been connected to the internet recently but have been powered down since (I noticed the clocks were both initially wrong and then updated using the gps data the pi3 to the right time and the pi 4 to the wrong date and time) and for my test were accessed without network connections (as that is how I will be using them at remote astronomy sites).

[mbalkham]

So the date is always 2000-01-30 and do its not the same number of weeks out its always the same date. Any ideas?

[DougieLawson]

https://www.youtube.com/watch?v=iyz7dSnZItw Matt Parker (the Standup Mathematician) explains the problem in exquisite detail.

[mbalkham]

Thanks Dougie but I'm pretty sure this isn't a roll over thing given the date is not a different amount out each day (every day the gps thinks is 30-01-2000).

Also the gps dongle is only weeks old and the chrony and gpsd latest versions.

Any other ideas?

[Ernst]

Thanks Dougie but I'm pretty sure this isn't a roll over thing given the date is not a different amount out each day (every day the gps thinks is 30-01-2000).

Also the gps dongle is only weeks old and the chrony and gpsd latest versions.

Any other ideas?

What do you see when you look up, do you see a ceiling or do you see a clear sky ?

[DougieLawson]

Try the GPS dongle on another computer. Buy another dongle from eBay.

https://www.ebay.co.uk/itm/GPS-Smart-An ... 3140614746
Note: that one doesn't do PPS but it does work with my 2014 RPiB+.

[topguy]

Thanks Dougie but I'm pretty sure this isn't a roll over thing given the date is not a different amount out each day (every day the gps thinks is 30-01-2000).

In your first post ( 8/9 ) the date was 23/1/2000 and now a week later (15/9) its 30/1/2000.. do you not see the pattern ?

As I said in the first post, you can verify this by checking the raw NMEA data for the GPRMC messages. You can try "cat /dev/ttyUSB0" or whatever the device is, but you may need to stop GPSD first.

[mbalkham]

Thanks topguy no I didn't notice the pattern! In my sleep deprived state I thought birth dates were 30th Jan!

I'll have another look at the raw nmea data as you suggest. Still makes no sense to me why the discrepancy between pi3 and pi4 though!

Dougie yes I'm outside, yes the GPS works fine on pi3 running same software (and on mac and PC). Just reports a different result on pi4.

[mbalkham]

Removed Chrony and I get the correct date. No idea why but it works so that's all good.

Thanks for the suggestions.