So... No advantage of faster MCU2 ? Or is the replacement Tegra faster too?
-
Want to remove ads? Register an account and login to see fewer ads, and become a Supporting Member to remove almost all ads.
Tesla now repairing MCU1 with a new Tegra card for less than $500
- Thread starter michaelrap
- Start date
I got an automated email from Tesla service this evening showing the ordered part. I'll call my SA tomorrow to ask about the length of warranty for this repair. I'll also post the full repair invoice when I get a copy.
'Provisioned' implies to me that the card has been programmed with the proper software.
It will probably not contain your certificates, but it contains a basic MCU operating system which allows the MCU to boot. The Service Center can then upload your certificates through the diagnostics port and that's it.
The part #: 2728211-S0-A is simply a production new Tegra board, BUT with the Hynix eMMC that will fail again after a few years.... I have seen a picture of one of these cards after a Tesla repair that was done last week. The Tesla invoice was roughly $500 for the board + labor, a total of $735.
Fine that it is a cheaper replacement, but the initial problem still remains, and as far as I know, only 3rd party shops actually upgrades the eMMC to a better version (Industrial grade).
When I say a production new Tegra board, it is a board from old stock, the above mentioned board did have a PCB date code from 2013.
Sorry problem don't remain Tesla update car software to reduce writting logs on the emmc. If you new repair last more than 10 years I think it's not a problem !!!
Yes, the problem remains, the extensive logging is only a small part of the problem.Check another of my posts here.
Also, the crazy logging is only reduced a little on the latest firmware releases.
My initial thought when I saw this news: somebody has gone through old storage at Fremont, and found a bunch of Tegra boards
It's anyway kind of good news for us Europeans that cannot get a MCU2 upgrade, provided it reaches the continent, of course.
It was completely nonsensical to design it the way they did, using a memory module with ~2000ish read/write cycles for such extensive logging.
A few things.
1 Tesla should have no problems installing this. It’s not much different than their process or replacing MCU.
2 3rd party repair will not be likely to drop their prices to match this. If anything. Most would probably stop doing them and if anyone wants to continue they would probably offer other services along as at 150 your basically at or above costs.
3 still being Hynix would not be a surprise. I’m still trying to find a newer mcu or this board. Actually I’ll bring in a car for this board so I can get one the check. Price is not bad at all. I’ll even just buy it if it can. If they will sell it to me. Yes. I’ll know what to do with it and get it to work in my car
1 Tesla should have no problems installing this. It’s not much different than their process or replacing MCU.
2 3rd party repair will not be likely to drop their prices to match this. If anything. Most would probably stop doing them and if anyone wants to continue they would probably offer other services along as at 150 your basically at or above costs.
3 still being Hynix would not be a surprise. I’m still trying to find a newer mcu or this board. Actually I’ll bring in a car for this board so I can get one the check. Price is not bad at all. I’ll even just buy it if it can. If they will sell it to me. Yes. I’ll know what to do with it and get it to work in my car
It is listed as a "restricted" part in the catalog, so I doubt they would sell it to you.
I guess I have to say this until I'm blue in the face; it is not the extensive logging, 'wearing out' the cells. It's just a cheap chip and the controller circuitry eventually develops problems, especially when warmer.
I've tested several bad chips and can't get multiple repeated dumps to sha256sum the same from -any- partition. This is an internal controller problem. Spray cold it and you can often recover the image, as per my wiki. That is where all these third parties learned to do this stuff. And it was my decision to specify Swissbit.
To reduce logging I use a script which shuts down the cid-updater daemon (the worst offender) except when I need to use it. It's not ever a good idea to turn off -all- logging, I think.
I've tested several bad chips and can't get multiple repeated dumps to sha256sum the same from -any- partition. This is an internal controller problem. Spray cold it and you can often recover the image, as per my wiki. That is where all these third parties learned to do this stuff. And it was my decision to specify Swissbit.
To reduce logging I use a script which shuts down the cid-updater daemon (the worst offender) except when I need to use it. It's not ever a good idea to turn off -all- logging, I think.
Last edited:
This is a much better solution vs full MCU replacement. Yes, having a larger eMMC would be the perfect solution, but unless the orig Tegra board had an option for a larger part, there’s no easy path for mass volume rework without taking time to put in quality/validation checks.
I’m assuming Tesla is making a last buy of the Tegra boards or asking for another build run.
I only have a little bit of warranty time left so will most likely we paying out of pocket when my MCU dies. I’ll gladly take $500 for another 4-5yrs vs $2K.
I’m assuming Tesla is making a last buy of the Tegra boards or asking for another build run.
I only have a little bit of warranty time left so will most likely we paying out of pocket when my MCU dies. I’ll gladly take $500 for another 4-5yrs vs $2K.
Great point, however the only metric Tesla has is to track the write cycles, outside of rooted cars it is hard to see any info about a controller failure outside of the symptoms of corruption.
Shutting down updater daemon is not a viable solution for owners unless tesla implements.
a DIY guide would likely result in many bricked cars, I have no proof, just wild guessing.
jelloslug
Active Member
Considering the $2000 fix was just an entire new (old) MCU, how is this much different other than being a fraction of the price?
