Tesla eMMC memory chip MCU problems
This article explains the causes around the eMMC memory chip problem and related defects in older Tesla MCU computers. The MCU (Media Control Unit) is the computer that is necessary to operate the vehicle from the screen. This computer has nothing in common with the autopilot computer.
Already in the first Model S an MCU with a 17-inch screen was installed and over the years the system developed into a versatile multimedia platform with Spotify music streaming, Internet browser and meanwhile even Netflix and YouTube video streaming. As a result, the demands on the processing power of the MCU computer increased more and more. So from March 2018 Tesla started to install a newer and better MCU (the MCU2) with a larger eMMC chip in its vehicles. Unfortunately, meanwhile it has become apparent that the older MCUs have a weakness that will lead to a defect sooner or later.
The two predecessors of the MCU2 introduced in 2018 are MCU0 and MCU1. As with any computer, the MCU must store program code and data somewhere. Tesla uses embedded memory chips for this purpose, so-called “Embedded Multi Media Card”, or eMMC for short. This is similar like a permanently installed USB stick in the computer. For example, the vehicle settings and the driver profile are stored, but also various log file protocols with status parameters of various vehicle systems are continuously logged. This enables the Tesla Service Center to evaluate these logs for diagnosis if a defect in the vehicle. Unfortunately, Tesla has implemented a decisive disadvantage in the design of the two older MCU variants. This is because every memory chip in today’s electronic devices wears out with every erase/write cycle. This means that with each write access the chip wears out a bit more. Until after a certain number of write cycles it is defective and no longer readable.
Normally, such a problem should only occur after many years. The SSD flash memory of a laptop or PC usually lasts more than 30 years with average use. But with the MCU, various negative circumstances coincide because the chip of MCU0 and MCU1 is only 8 GB in size. With a large memory, write accesses are distributed over many memory cells. A single cell is therefore written much less often than a small chip like the 8 GB model of the old MCU’s. Each cell can only handle a limited number of write cycles quickly memory before it is no longer readable. If too many cells are no longer readable, the whole chip becomes unusable and the stored data is lost. The smaller the chip and the larger the amount of data written over time, the faster the defect occurs.
Eventually, the MCU will be dead
With such a defective eMMC chip the MCU computer will not work anymore and the screen in the Tesla will remain black. Of course, access to all functions such as climate control, vehicle settings, navigation or changing the set maximum charge level of the battery is then no longer possible. But at least the car can still be driven (as long as the pin2drive feature was not activated before the defect. Without a screen the pin cannot be entered and you can’t drive).
Tesla had unfortunately made the problem even worse in the past. Irrespective of whether the car was driving or not, since software version V7 in 2015, a vast amount of data was suddenly being continuously logged day and night and written to the memory chip. With V8, the problem has become even worse as the software occupies more and more memory on the chip due to the increasing number of functions. Only in November 2019, these extreme log activities were reduced with the software update 2019.32.12.7. Since then, the vehicles only log when they are in operation (i.e. when driving or charging). This is also not good news for owners who like to charge for hours at the household socket.
So, the older an MCU is, the longer it was exposed to these intensive write cycles. Since about mid-2019, more and more defective MCUs have been appearing in vehicles built in 2015 and older.
Which vehicles are affected?
All Model S and Model X with a MCU0 or MCU1 will sooner or later be affected by the problem and the chip will stop working. The Model 3 has been built with MCU2 only since production started. Here the problem is massively reduced, but more about this later. The old MCU0 and MCU1 were used in Model S and X of the following years:
MCU0 (produced until Dec. 2015)
MCU1 (Dec 2015 – March 2018)
MCU2 (March 2018 – today)
To find out which type of MCU has been installed in a vehicle, go to the vehicle settings in the “Software” menu and select “Additional vehicle information”. The “Infotainment Processor” displayed there is either “NVIDIA Tegra” (MCU0 and MCU1) or “Intel Atom” (MCU2). Learn more about the MCU.
How can you tell that an eMMC chip in the Tesla is about to fail?
In the worst case, if the chip is already defective, the screen simply remains black. It is also possible that the MCU reboots unsuccessfully several times (Tesla logo appears on the screen every few minutes). Signs of a chip that is about to become defective are:
- It takes an extremely long time to start the MCU.
- The MCU often gets stuck and does not react anymore.
- The MCU restarts automatically and repeatedly.
- Operation becomes slower and slower (for example route calculation or scrolling and zooming in the map material).
- Strong picture disturbances on the display, which occur more and more frequently.
- Tesla drivers also report: If the trip data figures are no longer displayed at all, the memory chip is about to fail completely. It is then not far to a permanently black screen and it is time to act!
As soon as the MCU is dead, you can still drive (if the Pin2Drive function is not activated). But all settings on the vehicle cannot be changed anymore.
If the chip becomes completely unreadable, you will lose the following information that was stored on the chip:
- The VPN connection key, which is necessary for the vehicle to establish a mobile phone connection to the Tesla server. Therefore, it is missing: Software updates, Spotify/TuneIn music streaming, video streaming via mobile phone, voice control, Google Maps and also access via the mobile phone app.
- Trip Data
- Driver profile settings
- Vehicle settings
- Homelink profiles
- Bluetooth’s connections to mobile phones.
UPDATE from 11/11/2020
Tesla acknowledges the eMMC chip failure and changes the warranty. For all S/X built before March 2018 where the chip is broken, it will now be replaced free of charge if the car is maximum 8 years old and has not run more than 100’000 miles. Important: The car will only be exchanged if the chip has failed. Not proactive. Tesla will also refund all owners who have paid for eMMC repairs at Tesla in the past. Official information from Tesla.
Before that we had the following possibilities to solve the problem preventively or with an already defective chip.
Vehicle still under warranty
Until the MCU is no longer functional, Tesla will not take any action. However, there is currently an investigation by the US authority NHTSA (National Highway Traffic Safety Administration). It is possible that there could be a recall.
If the MCU is defective, Tesla will replace it with an identical other (used!) MCU1 with an 8 GB eMMC free of charge. Furthermore, Tesla has limited the warranty for such a replacement MCU to 2 years. An upgrade to a MCU2 is not possible during this exchange when the vehicle is still under warranty. Technically, it would only be necessary to replace the fingernail sized memory chip and not the whole MCU. Tesla has been doing the same since around September 2020 for around 400 EUR/USD. The problem is also not solved permanently and will also occur again with the replacement MCU at some point. Because of the only 8 GB sized chip this will probably be the case again after a few years. Furthermore, the MCU remains similarly slow as before.
Vehicle not longer under warranty
If the warranty is no longer valid, the complete MCU can be replaced by Tesla for a fee. In the past, a new MCU costs about 3000 EUR/USD. Meanwhile, such an exchange apparently still costs 1700 EUR/USD.
A cheaper alternative is to have only the chip replaced. This is done by Tesla since about September 2020 for about 400 EUR/USD (before only the replacement of the whole MCU was possible). You get a new chip with 64 GB capacity and a warranty for 2 years / 25’000 km. For a layman it is not possible to do this repair by himself because the chip is soldered directly on the board. Only with special equipment and appropriate knowledge such a chip can be replaced. This video shows how such a repair is carried out.
ATTENTION IMPORTANT: If you have exchanged the eMMC chip with a third-party supplier, you will have to pay an extra fee of about 1000 USD if you exchange the MCU later. This applies to all actions where the old MCU comes out of the car, including the upgrade to MCU2. The reason for this is that Tesla cannot continue to use the old MCU1 because it has been repaired (sometimes old MCU1 are reinstalled in the cars as reconditioned spare parts). In such a case the old MCU1 may be kept by the customer. If you want to upgrade to the MCU2 later, you should consider if you want to have such a repair done by a third party.
In addition to Tesla, a number of third-party suppliers now also offer such repairs. For example:
|BIZTeam Automotive||Zagreb – Croatia|
|e-mobility driving solutions||Different locations in Germany|
|Forcar Concepts||Pfäffikon – Switzerland (Repair on site in about 4-6 hours)|
|GS-Design||Waldmünchen, Bavaria, Germany|
|Smartmod.de||Frankfurt – Germany|
|Laadkabel_winkel.nl||Eindhoven – Netherlands|
|TMC Forum Source||Various locations in the USA|
|Eurolane||Buckingham – UK|
|Tesla Doctor||Bucharest, Romania|
|Tesla Service Center||Worldwide|
The cost of replacing the chip varies greatly depending on the supplier and ranges from around 400 to 1000 EUR/USD. Please also note that the data on the old chip cannot always be transferred to the new chip. Depending on the condition of the old chip, it may not be readable anymore. The most important information on the old chip is the mentioned VPN key, which is necessary to establish the internet connection between the vehicle and the Tesla server. Without this key the vehicle cannot reach the server.
This means that various functions such as Spotify/TuneIn/FMRadio, voice control, navigation maps, software updates and the connection to the mobile phone app no longer work. If the repairer cannot transfer the key to the new chip, the only thing left to do after replacing the eMMC chip is to go to the Tesla Service Center. Since about May 2019, Tesla also loads the key onto an MCU whose chip has been replaced (previously this was refused). Costs: between 100 and 300 EUR/USD.
In the ideal case the chip exchange costs less than 500 EUR/USD
The time required for a chip exchange is about 4 hours. With some providers, you can wait on site, others also accept postal delivery. To do this, the vehicle owner must remove his MCU himself and send it to the provider. Theoretically you can also just remove the Tegra board from the MCU and reinstall the “rest-MCU”. This way you can even still drive the Tesla. Just sending the Tegra-Board would also be easier. However, damage can occur when removing the Tegra board if this is not done correctly. Video instruction how to remove an MCU. And a second video on how to remove an MCU.
A big advantage of the chip exchange is besides the cost saving the possibility to have a bigger chip installed. Instead of the 8 GB chip, a double sized 16 GB chip is used. This chip will last twice as long and the MCU will be faster thanks to the larger memory. In this video you can see that the MCU1 with a 16 GB chip is much faster than with the small 8 GB chip.
Is Tesla going to fix it?
In a report from the Tesla Owners Club Norway it is mentioned that Tesla has now started to test the replacement of the Tegra board. These tests are expected to be completed in September. The Tesla Owners Club Norway estimates that the cost for a complete replacement of the Tegra board will be 4,000 Norwegian kroner. That would be the equivalent of about 380 USD. I am curious if this will happen.
UPDATEIn fact, Tesla now officially offers the replacement of the Tegra board for about 400 USD. You will receive a new chip with 64 GB capacity and a warranty for 2 years / 25’000 km.
What can I do if my chip is defective or obviously close to it?
If you decide to replace a chip in a vehicle without warranty for a cheaper repair without using the services of the Tesla Service Center, the VPN key must be transferred from the old chip to the new chip. If the old chip is no longer readable, only Tesla can help. Only Tesla has a copy of the key.
At the first signs of a defect, it can make sense to protect the chip by not writing any more data. As long as the MCU is still running, the key can still be read and transfered. With a dead MCU this might not be possible anymore. The longer the old chip is still in use, the higher is the probability that it becomes unreadable and you have to contact Tesla to get the key back. Especially if the MCU is constantly restarted automatically, the risk is very high that the MCU will finally destroy the chip due to the many reboots. Even if you suspect that the chip is about to die, you should not reset the MCU.
Prepare vehicle for MCU failure
To prevent the chip from being destroyed, many internet forums recommend removing the MCU’s fuse. For the Model S this is the fuse F51:
On the Model X it is obviously fuse F236. General information about the fuses on the Tesla. However, according to the BIZTeam, which specializes in repairing MCU’s, it is not a good idea to pull the fuse. Without this fuse, the Tesla won’t fall asleep again and thus draws power from the 12V battery and damage it. This video from BIZTeam shows what to do instead. Important: you should only do such things yourself if you know exactly what you are doing!
It is best to proceed as follows to pull the fuse:
- Disable Pin2Drive (because without MCU the pin input is no longer possible).
- Set the charge limit to 90% (without MCU this setting cannot be changed anymore, not even via the mobile phone)
- Deactivate planned charging so that the vehicle charges immediately when the cable is plugged in.
- Adjust climate settings to the desired “permanent state”. This cannot be changed afterwards without an MCU.
- For Model S: Close the panoramic roof!
- Pull the MCU fuse. (however, according to the BIZTeam, this should only be done for a maximum of 60 minutes).
Of course, except for the last step, these procedure makes sense for a vehicle under warranty, too. After all, who wants to drive all the way to the service centre with an open sunroof and a broken MCU? 🙂
Switch off Pin2Drive with a broken MCU
According to a Tesla internet forum, the following trick can be used to switch off the Pin2Drive on a dead MCU and you can still drive your car to the service center afterwards. You disconnect the 12V battery for a short time and then reconnect it. Then the vehicle should be drivable without the pin as long as the MCU is not in use.
And what about the MCU2?
The MCU2 contains a 64 GB memory chip. Technically speaking, the problem will also occur in this case at some point. But since the chip is 8 times larger than in the older MCU, one can assume that this will happen only after many years.
Tips regarding eMMC for Tesla used car purchase
For a vehicle with MCU0 or MCU1 it is interesting to know if the chip or the complete MCU has already been replaced. Unfortunately, it is not possible to determine whether the old 8 GB chip is still installed or a larger replacement chip is already installed. Here you can only believe the statement of the seller.
Replacing the chip makes sense in any case because even a new 16 GB chip lasts twice as long and the MCU is much faster. In the ideal case, the repair with the chip exchange also costs only about one third of a complete (old and slow) MCU from Tesla Service. Since Tesla has reduced the excessive logging of data meanwhile, one can assume that even a replacement MCU with a small 8 GB chip should still last longer than 5 years. But the MCU still won’t get faster. Whether Tesla will replace a defective MCU1 with a MCU2 in the future remains open. The alternative would be to do the MCU2 upgrade. But at the moment this is only available in the US.
The operator of this website assumes no liability for any damage caused by these tips.