Now I could run the updater but after the update I could not connect again to the GPS modul.
It is ver. 3.20 so it must be the GPS 18.
Caution - Annual Service - Car power-down - GPS issue
- Thread starter jfischer
- Start date
-
- Tags
- clock gps broken gps rollover
Now I could run the updater but after the update I could not connect again to the GPS modul.
It is ver. 3.20 so it must be the GPS 18.
slcasner
Active Member
No, your photo shows GPS 18x-LVC, so it is 18x not 18.
It is common that you can't connect after the update. You need to cycle power to the sensor, then it should start up in NMEA mode and be streaming data.
It is common that you can't connect after the update. You need to cycle power to the sensor, then it should start up in NMEA mode and be streaming data.
slcasner
Active Member
Another GPS 18x that I updated was at 3.20 before the update.
The GPS 18 in my car is at 3.00. I had not seen 3.30 for GPS 18.
The GPS 18 in my car is at 3.00. I had not seen 3.30 for GPS 18.
slcasner
Active Member
You're welcome!
OK, to summarize recent activity:
- My assumption that 1.5 Roadsters have the older GPS 18 LVC sensor (that was claimed by Garmin support not to need updating) appears to be incorrect because @Dr No's Validation Prototype 1.5 has the newer GPS 18x LVC.
- The GPS 18x LVC sensors in two more 2.x Roadsters belonging to @ShawnA and @gregd were successfully updated to the new firmware to fix the incorrect date using the revised cable that I made.
- Exactly what set of circumstances leads to the date becoming incorrect is still not certain.
Possibly, but it happened to me when the car was sitting there, minding its own business. It was about a day or two after a service event, but that was one of several similar service events over the past few months, all uneventful with regard to the GPS.
The firmware on the GPS puck was pretty old (version 3.1, I think, with the updated code being 4.4). My thought is that some satellite configuration or otherwise unseen state transition occurred that triggered a bug in that version, causing a reset, and that reset tripped into the date roll-over issue when it reacquired lock.
smorgasbord
Active Member
I appreciate @slcasner's respecting of my privacy, but I'm happy to reveal that my Roadster was the guinea pig.
In retrospect, it's pretty crazy to think that people were thinking that a 1024 week rollover was reasonable - even as early as the late 1970s. Apparently, the new standard has upgraded from the 10-bit week number to a 13-bit week number, so for new devices the rollover happens every 8192 weeks. See GPS Rollover is today. Here’s why devices might get wacky – TechCrunch . Why didn't they just do a 32 bit number? This is almost the 3rd decade of the 21st century after all.
13 bits is every 157 years, so none of us will be around to see the headscratching as 22nd century consumers wonder why all their devices lost the time. Which I predict will happen because it's not like there will have been a reminder like we get every 20 years, when at least there are still people around who remembered what happened.
Someone, save this thread until the year 2177 to help those poor souls....
pharma5
Roadster F#25, Model 3 #36xx
Adding my info, if at all helpful:
Roadster shows GPS issue, date is May 2000. U.S. Founders#25, assembled 24 July 2008. Do not know what version of GPS sensor I have.
Last service was late 2018. Expect to address as part of next service TBD in 1Q. (I often go ~18 months between appointments).
Thanks for a very informative thread.
slcasner
Active Member
That's very interesting. Yours was definitely made before mine. So either the installation of 18 vs. 18x sensor is unrelated to the build order, or the older 18 sensor also will suffer the date problem but is just has no update to fix it.
Please ask Tesla to report back which version of the sensor is installed. Well, I guess if they are able to update the sensor that's already installed, we'd know it was an 18x. Or if it is an 18 they may tell you it needs to be replaced.
If Tesla says they aren't able to do the update, or if they want to charge $500 and you'd rather do the update yourself for that price, I can lend my cable to you after January 26.
slcasner
Active Member
I can now confirm that it is possible to run the Garmin GPS update utility programs under Wine on a MacBook. It would likely work on Linux as well. So it is not necessary to have Windows laptop to do the GPS firmware update with the cable that I made. This update I just did went very smoothly all the way through connecting the SNSRXCFG utility to read back the NMEA sentences and then using the Update Software function to automatically invoke the Updater.exe utility. The only hitch was that I couldn't get SNSRXCFG to connect again after the update even after cycling power to the GPS (by unplugging the cable from the laptop). I needed to exit and restart SNSRXCFG, then it connected again.
Something curious about this particular car though: The date after the update was January 22, not January 19. I don't know if it might need to collect GPS timing for a longer period to be accurate, but that is not what I would expect. The only other possible explanation that I have been able to imagine so far would be if the time offset feature in the VDS (to accommodate time zones) allows adjustment more than 24 hours.
Something curious about this particular car though: The date after the update was January 22, not January 19. I don't know if it might need to collect GPS timing for a longer period to be accurate, but that is not what I would expect. The only other possible explanation that I have been able to imagine so far would be if the time offset feature in the VDS (to accommodate time zones) allows adjustment more than 24 hours.
slcasner
Active Member
Indeed, my hunch was right. This morning the car's owner was able to bring the date and time back to the correct setting with multiple pushes of the minus button on the VDS. The reason for the large offset was an earlier failed attempt to correct the incorrect date caused by the GPS week-number wraparound.
If anyone is interested, the two GPS utilities do seem to run on Linux / Wine. I didn't have the cable, nor desire to dig into the car again, but they appeared to talk to a serial terminal as expected. I did need to create a symbolic link between a com port file and the actual port in the ~/.wine/dosdevices directory (e.g. ln -s /dev/ttyUSB0 com5). Select com5 from the utility's pull-down, and off you go. Depending on your USB serial adapter and system configuration, the /dev/ttyUSB0 name might be different. Running dmesg | tail after plugging in the dongle should reveal where the operating system put it.
pharma5
Roadster F#25, Model 3 #36xx
Thanks for the offer on this. If I were at all handy I might say yes.
Also, I must apologize in coming up empty on two counts - trying to get a tech service bulletin # to report to the group here, and the GPS sensor version in my car. I did not get to talk to the tech directly, so I think each item got lost in translation. Info coming back suggested there wasn't a TSB for this issue (?), and they were not sure how to check the sensor version (?).
Slightly OT, sometimes all of one's communication bandwidth is consumed to make sure the old roadster service basics are understood and covered (reasonable date/time, annual service, loaner). This was one of those times. I'm grateful the basics came out good. They may get a 2nd chance on the GPS in spring/summer since that little water pump started making unhappy noises for the tech, intermittently.
slcasner
Active Member
Did your service invoice indicate whether the GPS just had a firmware update or was the sensor replaced?
pharma5
Roadster F#25, Model 3 #36xx
The GPS / date issue was acknowledged on my invoice but not resolved. No bill for it. The formal statement was: "The firmware update and harness to complete this is not available at this time".
I had initially sent a link to this TMC thread when reaching out to RoadsterNA email box in late Nov, but I think that got lost in the long path to a Jan service date.
slcasner
Active Member
So that means the date is still incorrect on your car?
I would be willing to loan the cable I made to you to take to the SC to do the update if they would agree to do it and return the cable in a timely manner.
pharma5
Roadster F#25, Model 3 #36xx
That's correct, still shows Y2K.
Can I reach out to you circa March/April on the cable loan offer? I intend to get preventive replacement of the coolant pump around that time (same one over 8+ yrs in my hands), so that's the next checkpoint if TSLA is ready to field the rollover issue.
Thank you for your skills and generosity.