Welcome to Tesla Motors Club
Discuss Tesla's Model S, Model 3, Model X, Model Y, Cybertruck, Roadster and More.
Register

Autopilot, GPS, and Nav broken. Tesla can't fix it. Advice?

This site may earn commission on affiliate links.
You said they replaced the "center console screen". Did that include the entire computer or just the screen?

Have you tried disconnecting the Dashcam? It feels like I have read someone having problems with their GPS after a dash cam install.

Who installed the audio system and what system? Changing the wiring around the MCU maybe related.
 
Why wouldn't lemon law apply here? The 15 minutes of reading I just did, which clearly makes me an expert, makes it sound like it applies. This is a safety issue that Tesla had tried to fix multiple times without success.
I seriously don't want to deal with a lawsuit. I just want my car to work :(
My legal expertise here is as deep as yours but from what I read the California law is limited to 18 months or 18,000 miles and from your timeline you are past that. Now a sharp lawyer might try to argue that the problem arose before that and defendant would counter that you took no action until Nov 2018, past the 18 months.

But the car isn't a lemon. This is a fixable problem. If I can look at your picture and tell that's it's probably a screwed up Kalman filter someone who knows the car should know where the Kalman filter is (though he may not know it by that name) and replace the unit that contains it. Or it may be a bad GPS module or loose connector or corrupt firmware. You just need to get to a knowledgeable tech. You are still under warranty and have a failure in an item covered by warranty. Make them honor the warranty. The fact that there is a potential safety issue should give you some ammunition. This will, given the communications problem within Tesla, probably be a big PITA but much much less than the one which would attend a lawsuit.
 
  • Like
Reactions: CyberShy
I had the same problem in my 2013 P85+. Tesla service center first replaced the MCU, that didn't fix the problem. They then replaced the GPS antenna and that fixed the problem.
To replace GPS antenna, they also need to remove and replace rear hatch window!
Good luck!
 
I have a dashcam and audio system installed. Why do you ask?

Have you tried disconnecting the Dashcam? It feels like I have read someone having problems with their GPS after a dash cam install.

Who installed the audio system and what system? Changing the wiring around the MCU maybe related.

You might be on to something here, but I would have hoped that Tesla would disconnect all 3rd party items when verifying the issue.

But we have seen several people that reported bad GPS location after they had an aftermarket dashcam installed especially if the installer installed it incorrectly by cutting the shielded cable and splicing it back together to make their life easier.
 
  • Informative
Reactions: rickyjb
The "antenna" referred to here is not just the antenna. The GPS module is mounted as close to the antenna as possible. PunchIT had a bad module and it looks as if cryptyk does too. But replacing it once didn't fix the problem. So that focuses attention on the harness between the GPS module and the computer up front (as has been suggested here). This is the kind of thinking that should be done by the SC - not by a bunch of guys on the internet.
 
I took the car back to a different service center. They let me know that they can't find any issues and asked me to continue keeping logs and let them know if it keeps happening.

I feel like they've given up and just expect me to deal with the issue, which I'm unwilling to do. Do I have any options left?
 
Yeah they defiantly need to be looking at the antenna/harness at this point. Probably a massive PIA for them but if it’s not the computers themselves, that’s the next thing to check. They should also be able to remove the MCU, put it on a bench and see if the GPS shows the correct location data. If it gives correct data then clearly there is something else wrong that they can locate... Don’t give up...
 
  • Like
Reactions: croman
There are two boots routing cables to the rear hatchback - the drivers side boot contains the GPS cables and should not be used to route a rear dashcam cable. The passenger side boot is just electrical and is fine to route a dashcam cable through. Also, as mentioned above, cutting and splcing a dashcam cable can cause issues. Good luck!
 
  • Like
  • Informative
Reactions: rickyjb and MP3Mike
Something very similar happened on my brother’s Benz. Main GPS bearing housing or whatever. It’s definitely something that happens, albeit seemingly rarely, with GPS systems.

you need a fresh set of eyes on the problem. I’m sorry if I missed why you can’t, but try a new service center. I have way better luck with the one in downtown Chicago than in the suburbs, for instance. The older the better in my experience.
 
The "antenna" referred to here is not just the antenna. The GPS module is mounted as close to the antenna as possible. PunchIT had a bad module and it looks as if cryptyk does too. But replacing it once didn't fix the problem. So that focuses attention on the harness between the GPS module and the computer up front (as has been suggested here). This is the kind of thinking that should be done by the SC - not by a bunch of guys on the internet.
I Mostly agree! But the folks at SC are incompetent.
A mobile tech caused damaged to dashpad and passenger side middle vent in my P100D, I sent pictures of the vent but they still ordered the wrong part! And they told me this after I waited for 2 hrs at the SC, when they promised that total wait will be only one hour for both fixes!!
DUH!!! :mad:
 
Just noticed where you are posting from. Which is the one to avoid? Rockville or Tysons?

Fortunately I have had only minor problems (loose gaskets and the front passenger seat "needed to be reprogrammed). Rockville took care of the first loose gasket and Montreal mobile the others. All OK
 
Welcome to the club GPS is off position

The "antenna" referred to here is not just the antenna. The GPS module is mounted as close to the antenna as possible. PunchIT had a bad module and it looks as if cryptyk does too. But replacing it once didn't fix the problem. So that focuses attention on the harness between the GPS module and the computer up front (as has been suggested here). This is the kind of thinking that should be done by the SC - not by a bunch of guys on the internet.

Is the signal analog all the way from the GPS module to the car or something?
 
I couldn't say specifically in this case but generally a satnav system consists of a patch antenna with rf and digital electronics sitting under it. The module is supplied with power and is connected to a communications bus which runs to a host computer. The computer sends commands to the module and retrieves data from it using the bus protocol (often NMEA).

Example: Ultimate GPS Module - 66 channel w/10 Hz updates
 
The car just got back from service. The service advisor told me that there's an internal document that says this is "normal behavior" and there may be a firmware update in the future to address it. Apparently the Driver Assistance computer can get "stuck" and needs to be reset before it will acquire a GPS fix after being parked underground.

Is there a way to escalate safety issues to the engineering team or something? This has been a known problem for years and the fix clearly hasn't been a priority.
 
Our 2015 S 85D is intermittently having the GPS mis-fire, its manifestation usually displayed in one of two modes. In one mode the position of the car is locked whether the car is moving or not. In the other failure mode the indicated position is offset by a distance and direction that moves in parallel with the motion of the car. Sometimes it will revive to functionality when the car moves, and sometimes it will slowly drift off the correct position until it either locks/freezes or achieves its ultimate incorrect offset position and continues to move.
It is my understanding of GPS receivers that they, as a matter of necessary prudence, will not function without adequate satellite data. If the antenna is partially or totally dysfunctional, they announce that with the "no signal" display. GPS does not make guesses about location. I doubt very much that these GPS malfunctions are antenna-related.

More basic research on GPS led me to a very informative and understandable web page, found using the search parameter "fundamentals of GPS". The pdf document by Dan Doberstein at cdn.sparkfun.com/datasheets/sensors/gps/etc...is an illustrated, diagrammatic explanation that is both simple and complex, depending on how deeply you want to understand the whole "time is relative" thing. You should have a handle on the basics in about 6 giant-fonted, mostly-illustration pages after the table of contents.

Here's what I think, based on the info from Doberstein: GPS receivers are little clock-shops with clever tricks employed to synchronize the receivers own clocks with the satellites' clocks' signals in a way that allows calculation of a position ...tic, tic, tic, calc, you are here! It seems to me that the mis-locating or non-locating malfunction (especially the moving, mis-locating failure mode) has to be a failure of the receivers' reference clocks to sync properly, causing a slight shift of the computation of position. Perhaps the 'locked up" failure is the reference clock not operating at all, or it's so out of phase that the receiver just sits, waiting. My leaning is to the semi-functioning reference clock, because the GPS receiver is clicking along, happy as a lark, thinking everything is working properly. I have no idea how any of the GPS clocks might fail in such a manner, but the symptoms are much more indicative of that than of antenna problems.

Tesla is about to do the rear-window/antenna 'fix' to our car. I am going to try to talk with them about my concerns. As many of you may already know, getting Tesla folks to converse face-to-face about problems isn't easy or straightforward. It's all under warranty, but I am not happy with willy-nilly replacements, and as you can tell, have little faith that their fix will take, so they'll end up replacing whatever component of the GPS down in the (must-be-prohibitively-expensive-to-service) guts of the control panel/system. So far, all the effort has resulted in resets, software/firmware downloads, and Tesla replacing what they named the "connectivity board" , but the dysfunction continues.

Lastly, and again, I urge you checkout the "fundamentals of gps" referred to above, if only to see an excellent example of how graphics can demonstrate and explain a complex subject...that whole 'picture...thousand words' thing.

Not. Antennas.
 
The same thing happened to me in my 2013. Here's a screenshot of the service invoice:

upload_2019-11-6_10-37-15.png
 
Here's what I think, based on the info from Doberstein: GPS receivers are little clock-shops with clever tricks employed to synchronize the receivers own clocks with the satellites' clocks' signals in a way that allows calculation of a position ...tic, tic, tic, calc, you are here!
Sort of. Each satellite has an (atomic) clock whose offset from "GPS" time is tracked by the ground segment of the system. The receiver has its own clock which is, hopefully, reasonably close to the system clock. Every so often each satellite announces "At the tone the time will be" and has sent (in the low speed Nav message) orbital parameters from which someone who knows about where he is and about what time it is can figure out about when he should receive the tick from each satellite. Given that he knows the geometry he can figure out how many nS error a meter change in latitude would make. If the tick from a satelite is off by x nS and the rate of change of delay is y nS/m then he is off in latitude by x/y m and he corrects his position by x/y and tries again. If you can remember back to high school math he is using Newton's method. It is, of course, appreciably more complicated than that because he has dozens of statellites (GPS and GNSS) and he has error in latitude, longitude, altitude and clock to deal with. This is going to sound pretty hairy but he is trying to locate himself in space-time, a 4 dimensional coordinate system. Time is, as far as the math is concerned, just another dimension. In this problem y isn't a number. It is a matrix with one row for each satellite and 4 columns, one each for x, y, z and t.

It seems to me that the mis-locating or non-locating malfunction (especially the moving, mis-locating failure mode) has to be a failure of the receivers' reference clocks to sync properly, causing a slight shift of the computation of position.
I implied above that position and clock are corrected by estimating expected time of reception of ticks, comparing to actual arrival times and multiplying the vector of errors by the inverse of the sensitivity (y) matrix and it can indeed be done this way but in the modern receiver the equivalent is done using a Kalman filtering algorithm. This takes the current estimated position (including clock offset) and operates on it to predict a new estimated position using a model of the receivers' motion (actually, x, y, z, t and their first derivatives are estimated). From this new state expected arrival times are computed as are residuals (the error between the expected and actual). These errors are weighted by the "Kalman gain matrix" and used to correct x,y,z,t and the velocities. At each iteration the filter looks at the residuals and determines the quality of the estimates. If the satellite data looks shabby more reliance is placed on the vehicle model and less on the observed residuals and conversely. Thus if you go into a tunnel where there is no satellite data the receiver will keep moving you along a straight line until it's determination that the estimate quality (no satellite info) is too poor and then it will shut down. This is what I believe is happening here.

Perhaps the 'locked up" failure is the reference clock not operating at all, or it's so out of phase that the receiver just sits, waiting.
If the clock isn't operating at all you get no position estimates at all. Remember that to the math time is just another measure of distance (the transit time from a satellite to the receiver is called its "pseudorange" by the GPS folks). A wildly wandering clock would lead to huge residuals and the RAIM (see below) function would shut everything down.

My leaning is to the semi-functioning reference clock, because the GPS receiver is clicking along, happy as a lark, thinking everything is working properly.
The receiver, if functioning properly, is not ticking along happily if the clock loop goes out of lock. It would immediately shut down. An important part of GPS receiver design is RAIM (Receiver Autonomous Integrity Monitoring).

I have no idea how any of the GPS clocks might fail in such a manner,
PLL chip fails but this would be detected by RAIM immediately.

but the symptoms are much more indicative of that than of antenna problems.
Not if you understand how these beasts work.

As I indicated in a previous post the GPS/GNSS module in the car is evidently typical of OEM parts sold for this function. It is supplied power by the car and sits on some communication bus which connects it to whichever computer handles the navigation. As the modules cost about $20 it is obvious that replacing the module is the first step to take. If the DC voltage at the module is in spec and the receiver is configured (over the bus by the host) it is very, very probably good and you can throw concerns about its clock out the window. I don't, of course, know what the actual answer is here but it is probably interference in the communications bus perhaps from a noise source, perhaps from loose or broken pins... or a problem in the host computer (S/W or hardware).


Tesla is about to do the rear-window/antenna 'fix' to our car.
That is what a prudent tech would do. Another post shows that they get enough low level diagnostics from the module (in case it isn't clear the module usually incorporates the antenna but it doesn't have to be that way) that they can, for example, see that the module is seeing satellites but can't use any. A blown clock chip could cause that! But so could interference with the Nav message reception, a faulty RAIM circuit etc., etc.




Not. Antennas.
No, it's probably not the antenna as that's a pretty simple device. It's probably bad cabling or a bad interface but it could be a bad module and certainly that's where diagnosis should start given that the modules are so cheap.