BigNick
Infamous Fat Sweaty Guy
To Elon Musk, a billionaire nearly 200x over, buying a new Tesla is like buying a cup of coffee.
Not Kopi Luwak, not Starbucks, it's like the convenience store $1.09-with-free-refills cup.
-
Want to remove ads? Register an account and login to see fewer ads, and become a Supporting Member to remove almost all ads.
Tesla infotainment system upgradeable from MCU1 to MCU2
- Thread starter MurrayJimW
- Start date
Not Kopi Luwak, not Starbucks, it's like the convenience store $1.09-with-free-refills cup.
mspohr
Well-Known Member
Hey Boomer... You listen to radio?No, $2,500 won't even get you a Tesla MCU2 "fix" with FM radio.
I'm still looking for my Creedence tapes, man
More like the socks he wears in his running shoes.
I've just come to accept that at some point my MCU2 will (on some future software release) perform about as well as MCU1 cars perform under the current software stack.
There will be a time when the MCU1 codebase will be frozen, and eventually there will be a time when the same will happen for my MCU2 car. I'm not saying I like this inevitable fact, but I do acknowledge that it is inevitable.
There will be a time when the MCU1 codebase will be frozen, and eventually there will be a time when the same will happen for my MCU2 car. I'm not saying I like this inevitable fact, but I do acknowledge that it is inevitable.
SucreTease
Teslarian
I don't think that is clear, at all. The problem could have been due to bad batches of parts, or parts that didn't meet their spec. Or the design may have been fine for the original expected use, but later software changes invalidated the assumptions upon which the design was based. There are a variety of possible scenarios, only some of which entail a design flaw.
mspohr
Well-Known Member
My understanding is that it was excessive logging which wore out the memory. They fixed that some time ago.
Akikiki
A'-Lo-HA ! y'all
There's long conversations scattered in many of these threads among the third party fixers on what the primary contributions to the eMMC failed problem are. (According to really smart people working on these, its not simply leveling or excessive logging.) And there are some very competent fixers like Jason Hughes that firmly believe and told Elon the problem is excessive logging and Elon fixed that, two years ago. Not so many posts from them in this thread, because this is not about failures, its about the MCU2 upgrade. Most people would not notice the names/who the third party fixers are without reading and saving some of that content. There's two people on TMC that have reported doing serious work (studying the operation) of the eMMC. (I am not going to mention their names, they will, one regularly does speak up). These two people have reported here that they believe the cause of the failure is less often leveling or bad spots and more a controller failure on the eMMC. Here's an excellent source for eMMC problems, fixes and what's going on Unofficial Tesla Tech Wiki. I think right now, something's wrong with the site. However save the url. Nuff said about that part.
I agree with @SucreTease said above "Or the design may have been fine for the original expected use, but later software changes invalidated the assumptions upon which the design was based. There are a variety of possible scenarios, only some of which entail a design flaw." .
We all know that Tesla is very tight lipped, so we have to depend on smart non-Tesla people to share what they have learned. And we piece together info from many places. Here's a couple of things that I have gleaned from TMC, TM, reading articles, emailing/PM third part fixers and some other places.
Personally I think the wrong hardware was used. However, that's easy for me to say. Its easy to look back at what has happened and make that judgement. Was there other hardware from another company that could have been used? I have not heard anyone from any other company - including Tesla say that they had more than one possible source for a eMMC like daughterboard that they choose to use. I bet most people don't know that the Tegra daughterboard was not designed for Tesla's use. It was an nVidia off-the-shelf product already being used in another manufacturer's car - Audi. Audi used the a version of the Tegra daughterboard in their Multi Media Interface. I don't know when they started using it, but were still using it as recent as 2012. Here's a link to the user manual and some pictures. I found this info and this link here on TMC in a NL thread a couple of years ago and saved it. The link does not work all the time. After you select an item on the left column, scroll down. If its working there content lists there. Let's not forget, the Tegra daughterboard is not the CPU, its storage. In another place I read that the Tegra board was modified by nVidia for Tesla to make it compatible with power requirements, and that made it slightly different from the one in the Audi. We've all heard stories how thin the resources were at Tesla when they were designing the Model S. If this Tegra board was all there was in 2010-2012 when Tesla was designing the Model S, there was little to choose from to get the job done. I think its easy now to say, they should have found an alternative to the Tegra board before 2015 when the worse of the failures have occurred.
Did you know that back when we received Ver 3.0 firmware it was just 300 MBs? And has grown since. Last time I heard a number without the added FSD Beta features, the size of Ver. 10 was just over 1.3 GB and still growing. That's just the firmware.
I'm not defending them now. I think now that there is enough money to fix, compensate or offer some amazing deals on MCU2 upgrades - whatever the owner (early adopters) want to do. I also think that several tens of thousands of 8GB eMMCs/MCU1 have been replaced and Tesla owes four things to owners. 1. Replace all 8 GB Tegra daughterboards that Tesla installed in refurbished MCU1 up until June 1 2020 whenever the owner request it be done Every car has earned that regardless of who owns it now - under warranty or not. 2. Offer a MCU2 upgrade at a substantial price reduction if the owner request it. 3. Compensate any owner that took their car and had the failing/failed 8 GB eMMC replaced by any 3rd party fixer - worldwide. 4. Apologize for the frustration/pain every owner with an impacted car has been through and promise nothing like this will ever happen again. But I'm not in charge, so I know my opinion does not matter.
I agree with @SucreTease said above "Or the design may have been fine for the original expected use, but later software changes invalidated the assumptions upon which the design was based. There are a variety of possible scenarios, only some of which entail a design flaw." .
We all know that Tesla is very tight lipped, so we have to depend on smart non-Tesla people to share what they have learned. And we piece together info from many places. Here's a couple of things that I have gleaned from TMC, TM, reading articles, emailing/PM third part fixers and some other places.
Personally I think the wrong hardware was used. However, that's easy for me to say. Its easy to look back at what has happened and make that judgement. Was there other hardware from another company that could have been used? I have not heard anyone from any other company - including Tesla say that they had more than one possible source for a eMMC like daughterboard that they choose to use. I bet most people don't know that the Tegra daughterboard was not designed for Tesla's use. It was an nVidia off-the-shelf product already being used in another manufacturer's car - Audi. Audi used the a version of the Tegra daughterboard in their Multi Media Interface. I don't know when they started using it, but were still using it as recent as 2012. Here's a link to the user manual and some pictures. I found this info and this link here on TMC in a NL thread a couple of years ago and saved it. The link does not work all the time. After you select an item on the left column, scroll down. If its working there content lists there. Let's not forget, the Tegra daughterboard is not the CPU, its storage. In another place I read that the Tegra board was modified by nVidia for Tesla to make it compatible with power requirements, and that made it slightly different from the one in the Audi. We've all heard stories how thin the resources were at Tesla when they were designing the Model S. If this Tegra board was all there was in 2010-2012 when Tesla was designing the Model S, there was little to choose from to get the job done. I think its easy now to say, they should have found an alternative to the Tegra board before 2015 when the worse of the failures have occurred.
Did you know that back when we received Ver 3.0 firmware it was just 300 MBs? And has grown since. Last time I heard a number without the added FSD Beta features, the size of Ver. 10 was just over 1.3 GB and still growing. That's just the firmware.
I'm not defending them now. I think now that there is enough money to fix, compensate or offer some amazing deals on MCU2 upgrades - whatever the owner (early adopters) want to do. I also think that several tens of thousands of 8GB eMMCs/MCU1 have been replaced and Tesla owes four things to owners. 1. Replace all 8 GB Tegra daughterboards that Tesla installed in refurbished MCU1 up until June 1 2020 whenever the owner request it be done Every car has earned that regardless of who owns it now - under warranty or not. 2. Offer a MCU2 upgrade at a substantial price reduction if the owner request it. 3. Compensate any owner that took their car and had the failing/failed 8 GB eMMC replaced by any 3rd party fixer - worldwide. 4. Apologize for the frustration/pain every owner with an impacted car has been through and promise nothing like this will ever happen again. But I'm not in charge, so I know my opinion does not matter.
Maybe, maybe not, at least as far as supporting "critical car functions" -- the atom is pretty much "good enough" for most things.
The Tegra, also, may be almost good enough but it is a way more complex design in that
I'm shocked that when they designed the MCU2 they didn't include provisions for installing a variant of it in an MCU1 intended car; analog tuner, ethernet for talking to IC, etc.
I don't think tesla recognized the depth of the MCU1 problem when building the MCU2 program. I bet they do now though.
Akikiki
A'-Lo-HA ! y'all
Agree. Exactly. I've collected dates and MY info on failed eMMC/MCU1s reported on TM and TMC. There were only a few reported failures until 2019. I've found reports of 550 failed/replaced eMMCs by 3rd party fixers and MCU1 swaps by Tesla. That's not everything that failed, only those I found on TM and TMC that owners reported. So, these numbers are some fraction of the total failed around the world. But of the 550 I found only 3 were reported failed in 2015, 5 in 2016, 23 in 2017 and 34 in 2018. Let me make sure its understood what I am saying. These numbers were what was reported failed on TM and TMC in calendar years 2015-2018. The year with the largest number of reported (here) failures was MY 2013. But by Jan 2021, 39.5% of the 550 failures were MY 2015. Between 2012 and 2015 the most reported failures (twice as many) were 2015 Does that make sense? 2012-2014 were each older than 2015 but reported failures were far more in 2015. What made 2015 more prone to failure? 2015 had a different version of the Hynix chip than the two used 2012-2014. What else was different? Answer: AP1. It appears to me, AP1 stressed the eMMC more than non-AP1 cars. Teslatap has done some analysis and written an article using my data and updated his numbers each month, if you are interested in reading more. Check out his work MCU1 Flash Memory Analysis and Failures – TeslaTap
Got the tuner upgrade added to my X yesterday, and wasn't charged for it. The proforma invoice had the $500 charge, but finally it was $0. I'm wondering if cars with Premium audio are getting it for free, and others are paying the $500? I don't see any other reason for the some do and some don't, but I've not gone back and counted those cases.
Just to confirm, no way to get SiriusXM to work on a 2017 X, right? (So I should just turn the darn thing off in Sources.... ). It sits there telling me my trial expired in November, 2020, for some weird reason.... Wonder why it would think that?
Just to confirm, no way to get SiriusXM to work on a 2017 X, right? (So I should just turn the darn thing off in Sources....
). It sits there telling me my trial expired in November, 2020, for some weird reason.... Wonder why it would think that?
Got the tuner upgrade added to my X yesterday, and wasn't charged for it. The proforma invoice had the $500 charge, but finally it was $0. I'm wondering if cars with Premium audio are getting it for free, and others are paying the $500? I don't see any other reason for the some do and some don't, but I've not gone back and counted those cases.
Just to confirm, no way to get SiriusXM to work on a 2017 X, right? (So I should just turn the darn thing off in Sources....
Which service center performed the install or was it a ranger?
Do rangers do these? It was Tyson's VA. Only charge was $20 for the state inspection I had them do since I was there anyway. Car wasn't touched until second day (which is often the case there), was done in a couple hours, so, looks like the process is pretty quick.
MXFan
Member
I have a 2017 Model X, got the tuner upgrade (for free too and yes I have premium audio) and got Sirius/XM working. But only after rebooting the car after sirius enabled the XM radio to make it work.
Interesting! Didn’t reboot until I got home so I’ll give Channel 1 a shot tomorrow, thanks!
The MCU2 upgrade seems to be $1500 now instead of $2500. See link:
Infotainment Upgrade
Infotainment Upgrade
Price drop could be due to recall coming....one way to make small profit instead of repairing all emmc
Agreed, and this makes used Model S built before March 2018 more attractive now. Save $1000 on the MCU2 upgrade.
I already hit accept yesterday on the $2500 estimate, but the work isn't until tomorrow. If I cancel the appointment, will I ever be able to take advantage of the $1500 price, or will they always be able to make my price $2,500 because I had at one point accepted that number.
