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

Tesla now repairing MCU1 with a new Tegra card for less than $500

This site may earn commission on affiliate links.
View attachment 575979 I got a surprise text message from Tesla service saying my car was ready to be picked up this afternoon. The bill was $490 and change as promised. I have attached a screenshot from the invoice to this post.

I asked my SA how long the warranty would be for this repair and he seemed to think it was one year unlimited miles. If this has an upgraded chip on it hopefully it will last several years.

Controls of the MCU do seem more snappy. Noticed I have AM radio as I was going through the controls. Don’t remember having AM radio? Anyway so glad to be back in the Model S.

Do you know if you got the 64 gb chip?
 
Most likely did/is. The part number being reported from the most recent Invoice information being posted has shown in the description that its a 64 GB chip. And I just read on a Tesla club site, that the separate Tegra board Tesla has been offering/using is with the 64 GB chip.
 
My 2013 is still on its original MCU so Friday 8/14 I submitted a service request for replacing the Tegra as preventive maintenance. Got an email last night (8/16) that parts are ordered:

MSX MCU PC TEGRA DAUGHTERBOARD-PROVISIONED (64GB)
Part #: 2728211-S0-A
Order Date: August 16, 2020

They haven't given a cost estimate yet. Wondering if it will have the $175 diagnostic fee? I think it shouldn't since there's no diagnostics to be done.

Appointment is 8/26. I'll update if there's any interest.
 
My 2013 is still on its original MCU so Friday 8/14 I submitted a service request for replacing the Tegra as preventive maintenance. Got an email last night (8/16) that parts are ordered:

MSX MCU PC TEGRA DAUGHTERBOARD-PROVISIONED (64GB)
Part #: 2728211-S0-A
Order Date: August 16, 2020

They haven't given a cost estimate yet. Wondering if it will have the $175 diagnostic fee? I think it shouldn't since there's no diagnostics to be done.

Appointment is 8/26. I'll update if there's any interest.

My model S has been dead for 7 weeks now. If they supply your 'preventative maintenance' before my repair of a dead car, that will be where I draw the line on legal action.
 
Go 3rd party. Do it now.

Contact TonyT/EV-Fixme. Contact info and how to remove the Tegra board in this thread: Consolidated eMMC Thread (MCU repair) (Black Center Screen)

I'm under the extended warranty, otherwise I would have already done this. Tesla has been stringing me along that they are working super hard to get the part. If I find that somebody whose part hasn't even failed can waltz in and get it replaced in no time, that is where I say enough with the BS. They have set a repair date for 8/27 , I can hold on for 10 days before launching missiles.
 
I find myself speculating this morning on whether or not this is really a new part with a 64 gigabyte chip? That seems like an odd number. Why would they all of a sudden start replacing these parts with a chip 8x larger for no apparent reason or functionality?

On the other hand, storage components like this are generally spec'd/referenced by their manufacturers in bits, not bytes. a 64 gigaBIT chip is an 8 gigaBYTE chip. Throw in Tesla's all-caps parts catalog descriptions and you've got some ready-made confusion.
 
I find myself speculating this morning on whether or not this is really a new part with a 64 gigabyte chip? That seems like an odd number. Why would they all of a sudden start replacing these parts with a chip 8x larger for no apparent reason or functionality?

On the other hand, storage components like this are generally spec'd/referenced by their manufacturers in bits, not bytes. a 64 gigaBIT chip is an 8 gigaBYTE chip. Throw in Tesla's all-caps parts catalog descriptions and you've got some ready-made confusion.

If I can get a photo of the board when my Ranger comes to do the replacement, I'll do that. They owe me big time, and I can't imagine they'd have an issue with that.
 
I find myself speculating this morning on whether or not this is really a new part with a 64 gigabyte chip? That seems like an odd number. Why would they all of a sudden start replacing these parts with a chip 8x larger for no apparent reason or functionality?

On the other hand, storage components like this are generally spec'd/referenced by their manufacturers in bits, not bytes. a 64 gigaBIT chip is an 8 gigaBYTE chip. Throw in Tesla's all-caps parts catalog descriptions and you've got some ready-made confusion.

Memory chips are typically marketed/listed in gigabytes, not gigabits. And 64 is not an odd number at all, they typically increase size in multiples of 2. Search for eMMC chips and they are always 1GB, 2GB, 4GB, 8GB, 16GB, 32GB, 64GB, etc. It seems a pretty safe assumption that it is an upgraded chip, especially considering the low price of chips and that an insufficient size is part of the root cause of failure.
 
I find myself speculating this morning on whether or not this is really a new part with a 64 gigabyte chip? That seems like an odd number. Why would they all of a sudden start replacing these parts with a chip 8x larger for no apparent reason or functionality?

On the other hand, storage components like this are generally spec'd/referenced by their manufacturers in bits, not bytes. a 64 gigaBIT chip is an 8 gigaBYTE chip. Throw in Tesla's all-caps parts catalog descriptions and you've got some ready-made confusion.
They are future proofing it. They know they will lose the NHTSA investigation and they have to prove how they will remedy it. A 64gb chip should never go bad.
 
Memory chips are typically marketed/listed in gigabytes, not gigabits. And 64 is not an odd number at all, they typically increase size in multiples of 2. Search for eMMC chips and they are always 1GB, 2GB, 4GB, 8GB, 16GB, 32GB, 64GB, etc. It seems a pretty safe assumption that it is an upgraded chip, especially considering the low price of chips and that an insufficient size is part of the root cause of failure.

I understand binary multiples. I'm saying going with a 64GB part is sort of a strange decision. Of course the price difference between that and 32GB or anything else is probably minuscule so maybe they really are doing the right thing and paying a couple extra bucks for the overkill to never have to think about it again. I sure hope so.
 
Just confirming that the Vancouver, BC SC is installing Tegra boards for under $500 (CAD) inc. installation and tax. Just picked up my 2014 S85 and it seems to be working fine again.

My MCU1 was replaced under warranty Fall 2019 (maybe with a refurb?), but failed again just after the vehicle warranty expired in Jan 2020. It became slow, buggy and rebooted frequently/randomly – just like the original had before the replacement. There was no additional warranty on the replacement. Anyway, it finally went completely dead about a week ago and after seeing this thread I called the SC, requesting a new Tegra daughterboard. They'd just got the boards in, and repaired it right away.
The only options offered to me the week before was a "new" MCU1 ($2000+) or an MCU2 upgrade ($3500+). Not willing to give up the AM/FM/XM radio options though since there are *many* remote and not-so-remote places in BC and the rest of Canada with no 3G/LTE coverage, where having no weather or road condition info can be a big safety issue, especially in winter.

I'm fortunate to have bga rework capability "in house" and was set to replace the 8MB eMMC chip with a 64MB chip myself, but with a $500 SC option now available, and the uncertain prospect of data recovery from a failed unit, it didn't seem worth taking the time for an uncertain outcome.
 
A friend's 2016 X got towed for a dead MCU. I'm curious to see what his paperwork says.

Turns out this wasn't a good example - they MCU had other issues. They replaced the Tegra card, but that didn't fix it, so he got a new MCU:
Concern: Touch screen and Dashboard cluster are completely blank
Verified customer concern finding the touchscreen blank. Was not able to restore MCU. Attempted Daughterboard
replacement however this would not bring the screen back online. Replaced entire MCU and confirmed the touchscreen is
functioning as expected.
Correction: Replace MCU Touchscreen Panel Model X
Parts Replaced or Added
Part Quantity

MCU,MDLX,NA PREMIUM,REMAN(1098010-99-H) 1.0

Correction: Replace Tegra Daughterboard Inside MCU (Model X)
Parts Replaced or Added
Part Quantity

SCR,M5X6.PH.TRX25.SST.W/PATCH(1010058-00-A) 12.0
SCR,M3x4.PH.TRX10.W/PATCH(1010052-00-A) 3.0

Pay Type: Basic Vehicle Limited Warranty 0.00
 
I understand binary multiples. I'm saying going with a 64GB part is sort of a strange decision. Of course the price difference between that and 32GB or anything else is probably minuscule so maybe they really are doing the right thing and paying a couple extra bucks for the overkill to never have to think about it again. I sure hope so.
I thought the obvious answer was 'wear leveling'. Spread out the writes to avoid 'wearing' out blocks as fast.
Wear leveling is a process that is programmed into flash memory devices that ensures that the blocks of memory in the flash device are used as efficiently as possible. The algorithm allows data to be written evenly across the device, which is essential for maximizing the capacity and longevity of the memory.
 
Reading and writing the data cannot be done remotely. The 3rd party guys have special tools to access the proper points on the Tegra board.
I do hope Tesla is using a current (and much larger) eMMC chip on these.
Service Centers may have a station that tge entire board plugs into for the programming.


They may not be able to, but on the other hand they might have been forced to by component availability issues.

The fresh MCU should last longer than the old one, even if the same capacity, because of two issues:.
- Tesla are more sensitive now on write-wear issues on flash chips, so perhaps a higher write-life chip
- There was apparently a logging issue which accelerated the write-wear issue on the current boards, which has been fixed.
 
They may not be able to, but on the other hand they might have been forced to by component availability issues.

Indeed. And perhaps the simplest explanation.

From MCU1 Flash Memory Analysis and Failures | TeslaTap

All current remanufactured MCU1s include a different eMMC part that has at least 8 times the longevity of the original part, as the original part is no longer made.
 
Memory chips are typically marketed/listed in gigabytes, not gigabits. And 64 is not an odd number at all, they typically increase size in multiples of 2. Search for eMMC chips and they are always 1GB, 2GB, 4GB, 8GB, 16GB, 32GB, 64GB, etc. It seems a pretty safe assumption that it is an upgraded chip, especially considering the low price of chips and that an insufficient size is part of the root cause of failure.

This is actually incorrect. NAND memory is always specified in units of bits.
See this example:
hynix Strengthens its NAND Flash Business | SK hynix Newsroom

Because Tesla part descriptions are specified in the work order using all CAPS, it is impossible to distinguish between 64GB (gigabytes) and 64Gb (gigabits).

The NAND part reference here "MSX MCU PC TEGRA DAUGHTERBOARD-PROVISIONED (64GB)" is almost definitely a 64Gb/8GB part.
 
From MCU1 Flash Memory Analysis and Failures | TeslaTap

All current remanufactured MCU1s include a different eMMC part that has at least 8 times the longevity of the original part, as the original part is no longer made.

I don't think 8x capacity = 8x longevity. That was disproven by the tesla owner who works in data recovery and did a microscope inspection on an old eMMC. He revealed that the failure was due to silicone degradation in the chip's controller and not necessarily excessive write cycles.

Preventive eMMC replacement on MCU1

Although the increased capacity should help avoid the corrupted register fault if I'm reading that correctly.
 
  • Helpful
Reactions: croman