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

MCU failure rate poll; please vote *whether or not* you've had a failure

Have you had a failure with your MCU that required replacement? If so, when did it happen first?

  • Yes; car was <1 year old

    Votes: 5 2.7%
  • Yes; car was 1-2 years old

    Votes: 6 3.3%
  • Yes; car was 2-3 years old

    Votes: 1 0.5%
  • Yes; car was 3-4 years old

    Votes: 12 6.6%
  • Yes; car was 4-5 years old

    Votes: 10 5.5%
  • Yes; car was 5+ years old

    Votes: 9 4.9%
  • No; car is <1 year old

    Votes: 11 6.0%
  • No; car is 1-2 years old

    Votes: 40 22.0%
  • No; car is 2-3 years old

    Votes: 35 19.2%
  • No; car is 3-4 years old

    Votes: 23 12.6%
  • No; car is 4-5 years old

    Votes: 29 15.9%
  • No; car is 5+ years old

    Votes: 24 13.2%

  • Total voters
    182
This site may earn commission on affiliate links.
I think that although this poll is not going to get to the depths of the issue (ie: heat effect vs write cycles vs available space in eMMC for optimized writing) it certainly shows likely systematic failure.

I think you would need hundred's of respondants for each vehicle type / use profile / climate type..... in order to filter out each possible cause and compare failures with non-failures.

What is very difficult is that eMMC failures do not have single, consistent manifestations. The point at which an MCU finally quits is not an indication of an absolute failure of the eMMC, just that error levels became unworkable and the whole system could not keep running.
 
I just changed my vote. Originally (as in a week ago) the Service center told me my MCU needed replacement, because a software installation was stalled and could not be completed, and the center screen would not function. But later, the shop determined the cause was a faulty or corrupted USB memory stick, of all things. Car now functioning normal after USB stick removed.
 
Does the Model S have a time counter (or good proxy) that is user accessible?
Mileage maybe the best we have but I suspect my mileage in NC is a lot less time than mileage in CA. I mean I drive in traffic but stop and go is probably 3 hours total out of 79k.
Also I have highish mileage because I had 2 years of long commutes - at 80 mph for 95% of the commute. And most of my roadtrip mileage is at 80 mph.
If time is the primary factor, then my estimated time to failure might be 7-8 years - in which case it isn't worth $400 now. 5/2015 build.
I guess the next question is how hard is it to check the eMMC - like will someone do it for free or nominal charge? Can it be done remotely?
 
the next question is how hard is it to check the eMMC - like will someone do it for free or nominal charge? Can it be done remotely?

Since the chip is soldered to the board, the only way to get an indication of condition would be if there is anything like SMART statistics recorded anywhere. You'd think with so much useless data being logged, they would also log something more useful like error counts somewhere! Unless you have had remote diagnostic level access given to a third party (they would need physical access to the MCU / eMMC to do this) then such data as may exisit is only available to Tesla.... and I cant imagine them handing that out.
 
I read about people having problems with their MCU they can't make go away. Like stuttering Spotify playback (which afaik cache songs on the eMMC) and frequent reboots. Could this be symptoms of a failing eMMC?

Some of these people have identical specs on their cars as I have on mine. And I don't have these problems. A flaky eMMC could be the difference between our cars.
 
Since the chip is soldered to the board, the only way to get an indication of condition would be if there is anything like SMART statistics recorded anywhere. You'd think with so much useless data being logged, they would also log something more useful like error counts somewhere! Unless you have had remote diagnostic level access given to a third party (they would need physical access to the MCU / eMMC to do this) then such data as may exisit is only available to Tesla.... and I cant imagine them handing that out.

Along the lines of SMART:
The wear leveling is internal to the emmc. The manufacturer could make that data (bad blocks, available usuble space, write counts) accessible via out of range memory addresses. If that were implemented, the software could check the part condition.
Who does this? Shrugg.
 
I just changed my vote. Originally (as in a week ago) the Service center told me my MCU needed replacement, because a software installation was stalled and could not be completed, and the center screen would not function. But later, the shop determined the cause was a faulty or corrupted USB memory stick, of all things. Car now functioning normal after USB stick removed.
I had an issue initially where my flash drive was causing my MCU to constantly reboot. Once I did take it out, the constant rebooting stopped but my firmware was still corrupted. Tesla updated my firmware with engineering help and it was working fine even with the same flash drive for about 5wks. Then I started having problems reading from the flash drive again and took it out and plugged back in and that worked for a fews days. It happened again a few days ago and it stopped reading after removing and plugging back in. So I did the next logical step which was to reboot the MCU. Well it never came back up.
 
  • Informative
Reactions: David29
the next logical step which was to reboot the MCU. Well it never came back up.

It is a good thing that not every glitch or malfunction of the MCU is going to be an eMMC failure, but there is more than a good chance that it could be.

I can't believe how reminiscent the failure symptoms of the eMMC are to early hard drives (when 10mb was a huge hdd!). We used to run seamingly endless surface test / analysis cycles that would construct the bad block table, and you could tell when a manufacturer was going through a bad patch because the stat's would really drift off. When 'lower grade' units had been out in the wild for about a year, you would start getting random and inexplicable system failures. Software teams spent ages trying to track down and fix 'bugs' that they thought they were seeing evidence of across multiple systems, when in fact they were seeing the increased liklihood of failure at specific system pinch points where there was heavy disk traffic.

There was the same statistically predictable 'no issues' systems that threw you off the scent, because it made you thing that the hardware was capable of running just fine.... which it was, until it failed. The slower failing systems often (may be counter intuitively) seemed more likely to exhibit more extreme / less predictable failures and having often struggled on longer, turned out to be much harder to recover when they bit the dust. Even regular backups were not always a fix as they could struggle on too, backing up gibberish.

With the Tesla eMMC the annoying facts are that a) it is so hard to get any early warning that your system has an eMMC issue because of course your car doesn't go wrong and b) once you are forced to believe you have an issue, it will be a big problem with no chance for mitigation measures. If only Tesla tracked (and / or disclosed) error stats / condition inf for the eMMC.
 
  • Informative
Reactions: David29
With the Tesla eMMC the annoying facts are that a) it is so hard to get any early warning that your system has an eMMC issue because of course your car doesn't go wrong and b) once you are forced to believe you have an issue, it will be a big problem with no chance for mitigation measures. If only Tesla tracked (and / or disclosed) error stats / condition inf for the eMMC.
I don't get why it's a big problem. Failures are seldom and at long intervals, replacement is fast and not particularly expensive as automotive repairs go. With the relatively new cabin overheat protection, failures should be reduced dramatically. In my instance it was 120K miles, six years, most of the years having no cabin overheat protection and the car sitting outside in the Texas sun. That's a pretty brutal environment.
 
  • Informative
Reactions: David29
I don't get why it's a big problem. Failures are seldom and at long intervals, replacement is fast and not particularly expensive as automotive repairs go. With the relatively new cabin overheat protection, failures should be reduced dramatically. In my instance it was 120K miles, six years, most of the years having no cabin overheat protection and the car sitting outside in the Texas sun. That's a pretty brutal environment.

That is defineately a positive experience, and based on that, I would agree - 2.5k per 100k+ miles is no biggy. And if heat is a contibuting factor (generally it is where electronic components are concerned esp. over 50deg C) then Texas would put you at higher risk, but the point is that the way it fails (over weeks / months of resetting and system updates not going smoothly) makes it hard to know when / if you have an issue and at what time to pay out your 2.5k. With some failing much earlier than others, that makes it harder to respond to appropriately. On top of that, if your unit (for argument's sake) failed after 3 years rather than 6 because of excessive data being written by Tesla's logging system, then that's potentially a big unknown and further unknown with regards to if / when the same thing will happen again.

Doubtless when compared with the miles covered by Tesla cars, this is a tiny issue compared with ICE ownership issues as a whole, but that doesn't help when it goes wrong and you are out of wty.

A lot of comment is speculative, but not without some basis / evidence.
 
  • Helpful
Reactions: David29
My data point: dead MCU1 at 3 years 8 months, 44k miles. Car could be driven but charged at less than 0.5 kW — less than Level 1 speed — so it really couldn't be used in any meaningful way.

Car was transported 300 miles to the nearest service center for a new LTE MCU1 unit under warranty. I guess I have another year or two before the current screen fails. A prophylactic eMMC chip replacement seems like a good idea although I suppose I'd have to drive to California to have it done.


I'm curious to know if a car with a dead MCU can be Supercharged, has anyone every tried it?
 
I don't get why it's a big problem. Failures are seldom and at long intervals, replacement is fast and not particularly expensive as automotive repairs go. With the relatively new cabin overheat protection, failures should be reduced dramatically. In my instance it was 120K miles, six years, most of the years having no cabin overheat protection and the car sitting outside in the Texas sun. That's a pretty brutal environment.
But it’s the cost of an automatic transmission in an ICE. And a simple design change and the problem would not exist.

I do appreciate that it’s a fast/easy fix - at least for Tesla.
 
But it’s the cost of an automatic transmission in an ICE. And a simple design change and the problem would not exist.

I do appreciate that it’s a fast/easy fix - at least for Tesla.

I agree. It is a significant cost. I don't think I spent any money on my pickup in the first 100 kmiles. I had to do the half axes in the front at 120 kmiles but that was only $1200 or so. I had to do it again at about 220 kmiles and I paid a neighbor $100 and the parts were $300. Shows how much inflation a shop adds.

I am hating my model X because the service center won't/can't fix the many intermittent problems I have now. I'm really going to hate it when I have to start paying them a small fortune for regular repairs. It makes the savings from no oil changes seem pretty trivial.
 
My data point: dead MCU1 at 3 years 8 months, 44k miles. Car could be driven but charged at less than 0.5 kW — less than Level 1 speed — so it really couldn't be used in any meaningful way.

Car was transported 300 miles to the nearest service center for a new LTE MCU1 unit under warranty. I guess I have another year or two before the current screen fails. A prophylactic eMMC chip replacement seems like a good idea although I suppose I'd have to drive to California to have it done.


I'm curious to know if a car with a dead MCU can be Supercharged, has anyone every tried it?
I have been told it could be, but when I tried yesterday it gave the same message as charging from home. Charging scheduled on the IC only. Perhaps if scheduled charging is turned off it would work?
 
I have had 1 MCU replacement on my 2013 model s, lately the screen goes blank and reboots itself. Tesla says it probably is a corrupted usb stick with my music. So my question is what are my options do I buy a new stick and download all my music again.
First: Take the drive out for a few days and see if it happens. I had them tell me that it was the .mov files. However, there were no .mov files, there were some song files with mov in the title (because they were from a movie).
 
I have had 1 MCU replacement on my 2013 model s, lately the screen goes blank and reboots itself. Tesla says it probably is a corrupted usb stick with my music. So my question is what are my options do I buy a new stick and download all my music again.
When the SC told me that, I checked the USB memory stick by plugging it into my desktop PC. Sure enough, Windows said the stick was "damaged" and offered to fix it. I said yes, and it did something for a second or two. Out of curiosity, I checked the most recent music I had added to the collection, and found an unplayable song. I deleted that song from the collection in my master file on the PC, then re-ripped it from the CD.
At the suggestion of someone in an earlier post, I bought a new memory card. (My old one was 4 years old.) This person suggested getting memory cards made for security cameras and supposedly more rugged, suitable for outdoor use and hopefully better able to withstand the harsh conditions in the car. I bought one, loaded it, and put it in the car, where it has performed OK so far.

The memory card he suggested and which I bought was this one:

https://smile.amazon.com/gp/product/B07B98GXQT/ref=ppx_yo_dt_b_asin_title_o03_s00?ie=UTF8&psc=1

I also got this adapter to allow it to plug into the car's USB port:

https://smile.amazon.com/gp/product/B07G5JV2B5/ref=ppx_yo_dt_b_asin_title_o03_s00?ie=UTF8&psc=1

It seemed like a small price to pay to avoid the problem of being unable to install the next software update and losing use of my car for a few days....
 
When the SC told me that, I checked the USB memory stick by plugging it into my desktop PC. Sure enough, Windows said the stick was "damaged" and offered to fix it. I said yes, and it did something for a second or two. Out of curiosity, I checked the most recent music I had added to the collection, and found an unplayable song. I deleted that song from the collection in my master file on the PC, then re-ripped it from the CD.
At the suggestion of someone in an earlier post, I bought a new memory card. (My old one was 4 years old.) This person suggested getting memory cards made for security cameras and supposedly more rugged, suitable for outdoor use and hopefully better able to withstand the harsh conditions in the car. I bought one, loaded it, and put it in the car, where it has performed OK so far.

The memory card he suggested and which I bought was this one:

https://smile.amazon.com/gp/product/B07B98GXQT/ref=ppx_yo_dt_b_asin_title_o03_s00?ie=UTF8&psc=1

I also got this adapter to allow it to plug into the car's USB port:

https://smile.amazon.com/gp/product/B07G5JV2B5/ref=ppx_yo_dt_b_asin_title_o03_s00?ie=UTF8&psc=1

It seemed like a small price to pay to avoid the problem of being unable to install the next software update and losing use of my car for a few days....
Thanks for the info I’m not very computer savvy, I have an iMac is it hard to take my usb stick and transfer the music to the above scan disc.