Forums

MCUv1 flash memory issue

MCUv1 flash memory issue

There has been a flurry of recent issues about 2015 and earlier TESLAs being vulnerable to complete system fail presumably in regards complete loss of capacity of flash memory storage due to excessive data logging.
Two links (among many) are provided below.
This presumably requires a not inexpensive replacement for cars out of warranty, plus the nuisance of potentially having to have a vehicle towed to a service center.
At this point, it seems difficult to sort out fact from faction on this issue, and I haven't seen it discussed in this Forum (although perhaps I missed it).

Questions I have include:
If it is an excessive data logging issue, can some of the data logging functions be remotely disabled to minimize the likelihood this occurs?
What are the factors of vehicle usage that lead to this being likely to happen, total amount of mileage driven, mileage driven in new locations, something related to multimedia usage?
Can TESLA provide a report to customers regarding available storage with warnings regarding minimum time to fail?

https://www.tomshardware.com/news/flash-memory-wear-killing-older-teslas...
https://www.zdnet.com/article/how-teslas-flash-storage-fail-may-lead-to-...

PrescottRichard | October 18, 2019

Don’t get too excited, if you read both your links you’ll see they reference the same source-

https://insideevs.com/news/376037/tesla-mcu-emmc-memory-issue/

Both are trying to get clicks, and now they have a couple of mine. Dammit. There is an echo going on here, just like everything else on the internet.

That said, while the inside EVs article had to be updated because they got the ‘M’ in MCU wrong they did quote three reputable sources. At least, I’m assuming they are reputable.

I’ve heard of (and seen posts by) Gruber, and that is a shop that has great respect in the AZ Tesla community from what I’ve seen. He points out that new MCUs have a more robust eMMC so maybe that’s because of the earlier problems, or a change in supplier because of some other reason (costs?).

All three blame logging to the chip as the reason it fails, no one really explains how they know this. I suppose it’s so simple to a tech person that pointing this out is like saying ‘sure the ground is wet but how do I know that’s related to the rain?’

So maybe there is smoke & indeed fire in this story.

I can’t accurately answer your questions, but my WAG is that the more you drive the more data gets logged. So a car with 30k miles has less wear than my 2016 90D with 80K miles.

My memory (to answer the second question) is that SOME owners reported a ‘fatal error’ or something like that on the MCU screen before it died. Quite a few didn’t get that though, so even that isn’t helpful.

I’m interested in this too, but lets keep it in perspective. Just as the Inside EVs article stated, this looks bad because Tesla is a tech company first and they didn’t get the tech perfect. My answer to that is - yup. They prolly didn’t. I know with some of the updates there are plenty of people here pointing out how not-perfect things are.

How many different places parroted articles every time a Tesla catches fire or autopilot is involved in a crash? People need clicks to feed their families.

Darthamerica | October 18, 2019

eMMC wears out as you use it. This wouldn't be the first time a company found out the hard way that a part wasn't suitable.

lingle.chris | October 18, 2019

My original comment certainly showed no excitement, but was just trying to create a Forum for additional information as it comes out. I had read the InsideEV article, too. We are all certainly aware of the fear-and-loathing TESLA misinformation spawned by oil company shills, but this issue seems to have some valid technical basis.

Software fixes to reduce what gets logged to essentials should be possible, and, if so, it wouldn't surprise me that we might hear something on that soon.

In my case, although my MS is an early 2015, my mileage is low, so I've assumed I'll be OK for a while.

Whether a tech company has gotten everything right is always an issue, but a company still has a responsibility to customers when flaws become apparent.

PrescottRichard | October 18, 2019

Gotcha, not excited.

Have you seen this thread-

https://forums.tesla.com/forum/forums/how-common-replacing-mcu?page=1#new

This one might have some interesting ramblings on the MCU as well-

https://forums.tesla.com/forum/forums/mcu-retrofit-confirmed

There is a Rich Rebuilds video mentioned in that second thread that *may* be the origin of the flash memory failure idea.

I’m thinking your MCU and the one I have are prolly the same vintage, so I’m as interested as you are! I agree with you, if there is a design flaw then that flaw should be addressed and not on the customer (paraphrasing your last paragraph).

FIngers crossed!

DonS | October 18, 2019

Everything wears out. Depending on the number of writes, the flash memory may or may not wear out sooner than other components.

The really interesting thing to me is that most components wear out and then have to replace them. That is just the way it is. This flash issue is unusual because it may be possible to change the software to extend its life. I think that is pretty neat. Of course if my mcu fails, I won't be impressed.

TeslaTap.com | October 19, 2019

The original source video, which most articles afterwards are based on, said the problem was due to Linux logs. I seriously doubt this, as there really isn't that much data to be written. Logging data should last well beyond the life of the car and hundreds of thousands of miles.

There is other activity that could increase the writes that cause the problem in some cars. Keep in mind this was designed around 2009 to 20010, long before features such as AP were envisioned. At the time, it was likely the right choice. The oldest cars also don't have AP, Sentry mode, and a host of other newer features.

With all the additional features over the last 10 years, it may have contributed to more writes and some eMMc failures over time. From Elon's tweets, seems Tesla has made some changes to reduce the wear, but they didn't explain what they were or how much effect may have.

So it's a real problem, but unclear how significant. My sense of it is less than 1% of cars have the issue in the first 5 years. Still if happens to you, it can be expensive to fix.

jordanrichard | October 19, 2019

TT +1 and if you ever had to replace a blown head gasket on an ICE car, you how expensive that can be to fix.......

Skillfulist88 | December 2, 2019

Hope this gets fixed by the time the Cybertruck arrives.

TeslaTap.com | December 2, 2019

MCU2 in all new vehicles since 2018 uses a much larger flash drive, and one from a different vendor. In theory, it should last at least 4 times longer - in the 20-year range.