Let the hacking begin... (Model S parts on the bench)

[stopcrazypp]

Heatsink is not really considered active cooling. Shared heat sink heats up the EMMC to the same temperature as the rest of the chips (SoC, RAM, couple of others), so the EMMC chips warm up faster than in the Tesla but also don't reach the same highs. Additionally, I'm only starting to dig into this (just picked up my Taycan 3 weeks ago, then drove it home for 9 days to get to know it), but I don't think Porsche is writing much to the OTA module EMMC at runtime, other than during OTA.

The video says a shared heat sink does not warm up the NAND faster. Leaving it unsinked actually lets the NAND heat itself up faster and retain that heat better. It's probably because of the large thermal mass of the heat sink which sucks out all the heat from the NAND. I guess it does depend on the specific temp target of the SOC (vs how hot the NAND gets without a heat sink), but the rest of the components I don't imagine would contribute to help the NAND, given the limited amount of heat they put out.

And as you say it limits the peak temp, which is not what you want in a NAND chip. If you look at the chart I linked, there are life expectancy gains up to 80C (when they ended testing).

 

[Krash]

It’s definitely under the heat sink and you have to remove it. Been there done that.

View attachment 913902

Actually that is the eMMC. The heatsink is on the left above but the right side above is a removable heatsink shroud. Here’s a shot with the shroud removed…

JWC68 is likely the same part in the Intel 3/Y that tops out at 64G. Maybe there is a higher density part in another line that is BGA compatible and will work with the same controller.

 

[whitex]

The video says a shared heat sink does not warm up the NAND faster. Leaving it unsinked actually lets the NAND heat itself up faster and retain that heat better. It's probably because of the large thermal mass of the heat sink which sucks out all the heat from the NAND. I guess it does depend on the specific temp target of the SOC (vs how hot the NAND gets without a heat sink), but the rest of the components I don't imagine would contribute to help the NAND, given the limited amount of heat they put out.

And as you say it limits the peak temp, which is not what you want in a NAND chip. If you look at the chart I linked, there are life expectancy gains up to 80C (when they ended testing).

Are you implying the hotter the better for the NAND? Why not mount them on engine radiators then?
As a side note, automotive electronics are speced up to 90C ambient temperature for normal operation. They have to survive higher, but can offer limited operation. And yes, cars do get above 100C inside in some places.

 

[PaulF2020]

Are you implying the hotter the better for the NAND? Why not mount them on engine radiators then?
As a side note, automotive electronics are speced up to 90C ambient temperature for normal operation. They have to survive higher, but can offer limited operation. And yes, cars do get above 100C inside in some places.

I wouldn't worry about it in this specific case, eMMC has controller in the same chip as NAND, you can't separate them to cool one or another. Also, I would disregard this video completely as it is not relevant to real world use of NAND.

 

[Zeemo]

Has anyone successfully attempted to read the SPANSION chip on the Tegra board, without removing the chip from the board?

 

[cloneguru]

Has anyone successfully attempted to read the SPANSION chip on the Tegra board, without removing the chip from the board?

I've done it.

 

[cloneguru]

Preventive eMMC replacement on MCU1

Just had the MCU2 upgrade done to 2014 Dec issue S85 as after may visits and calls regarding MCU1 failing issues I just thought that I would like my heating to work, internet connection to come back on and the car to stop rebooting if I looked at it the wrong way. Very happy with the screen...

teslamotorsclub.com

 

[Synthax]

Preventive eMMC replacement on MCU1

Just had the MCU2 upgrade done to 2014 Dec issue S85 as after may visits and calls regarding MCU1 failing issues I just thought that I would like my heating to work, internet connection to come back on and the car to stop rebooting if I looked at it the wrong way. Very happy with the screen...

teslamotorsclub.com

that's eMMC, not Spansion flash. It can only be read via the kernel or U-boot.

 

[Zeemo]

that's eMMC, not Spansion flash. It can only be read via the kernel or U-boot.

I have done it using dataman with a programming adapter

 

[Synthax]

I have done it using dataman with a programming adapter

Technically there are other ways for sure, but those two are the easiest. JTAG will do it as well.

 

[Zeemo]

Technically there are other ways for sure, but those two are the easiest. JTAG will do it as well.

how can that be done? I spoke with a tech from infineon and this is what he said:

Thank you for contacting Infineon Technologies.

Assuming that the RESET#, WE# and WP# are at logic HIGH (deasserted), VCC and VI/O have a 0.1uF decoupling capacitor with clean (undisturbed) input signals, and that no DNU input signals are connected to the board, the minimum pins required to initiate a successful READ operation are as follows:

Chip Select (CE#) : one input pin
Output Enable (OE#) : one input pin
Address (A24 - A0) : Twenty-five address pins
Data Bus (DQ15 - DQ0) : Sixteen data (I/O) pins

D = Fortif. BGA package (LAE064) 9 x 9mm

Total pins : 43 pins (minimum) to initiate the READ operation

and for some reason i burnt one chip and could read the other one..

 

[Synthax]

how can that be done? I spoke with a tech from infineon and this is what he said:

Thank you for contacting Infineon Technologies.

Assuming that the RESET#, WE# and WP# are at logic HIGH (deasserted), VCC and VI/O have a 0.1uF decoupling capacitor with clean (undisturbed) input signals, and that no DNU input signals are connected to the board, the minimum pins required to initiate a successful READ operation are as follows:

Chip Select (CE#) : one input pin
Output Enable (OE#) : one input pin
Address (A24 - A0) : Twenty-five address pins
Data Bus (DQ15 - DQ0) : Sixteen data (I/O) pins

D = Fortif. BGA package (LAE064) 9 x 9mm

Total pins : 43 pins (minimum) to initiate the READ operation

and for some reason i burnt one chip and could read the other one..

JTAG is much more difficult. Your best bet to safely read/write Spansion is to modify the U-boot defconfig for a similar device. You could start with Cardhu for instance. Use Tesla's Linux device tree source as a reference to add Spansion to your new device tree and pinmux. You can reference the U-boot implementation for Surface RT as well. You will need a mini USB cable and UART. You can load U-boot with nvflash for Tegra.

 

[Domih]

Hey! I just found out about this thread so I decided to create an account and participate! Hope any of you can point me in the right direction for my problem

I own 2 Model S with MCU 1(2014 P85 and 2016 P90DL). I wanted to swap the display, because one of them had it come loose.

After changing it, one of the MCU's wouldn't boot anymore, even after swapping back the display.

I connected it to the testbench but there was no way to ping the CID(192.168.90.100). I can ping the GW without a problem.

I also wanted to see if it's possible to drive without MCU, so I removed MCU while car was in drive. I think that messed up the car and put it in some kind of recovery mode.

I've been messing around with it for 2-3 weeks now but I have no idea what to do now. Anyone can point me in the right direction?

With the other MCU I can ping the CID for about 4-5 seconds during boot and then it times out, I think I miss the wake signal. I only connected 12V and ethernet.

Thanks in advance, and happy tinkering

 

[cloneguru]

Hey! I just found out about this thread so I decided to create an account and participate! Hope any of you can point me in the right direction for my problem

I own 2 Model S with MCU 1(2014 P85 and 2016 P90DL). I wanted to swap the display, because one of them had it come loose.

After changing it, one of the MCU's wouldn't boot anymore, even after swapping back the display.

I connected it to the testbench but there was no way to ping the CID(192.168.90.100). I can ping the GW without a problem.

I also wanted to see if it's possible to drive without MCU, so I removed MCU while car was in drive. I think that messed up the car and put it in some kind of recovery mode.

I've been messing around with it for 2-3 weeks now but I have no idea what to do now. Anyone can point me in the right direction?

With the other MCU I can ping the CID for about 4-5 seconds during boot and then it times out, I think I miss the wake signal. I only connected 12V and ethernet.

Thanks in advance, and happy tinkering

Sure sounds to me like the eMMC failed and needs to be replaced.

 

[Domih]

Sure sounds to me like the eMMC failed and needs to be replaced.

Tesla will just replace the whole MCU, right?

 

[cloneguru]

Tesla will just replace the whole MCU, right?

If it’s under warranty… they’d either replace the tegra board or the entire MCU … I don’t think they are tooled or trained in replacing just the eMMC.

If it’s not under warranty… myself and many others are capable of removing the eMMC and recovering the critical files (assuming the chip responds when placed in a programmer), provisioning a new eMMC, and reinstalling it back on the tegra board and putting it all back together.

 

[islandbayy]

If it’s under warranty… they’d either replace the tegra board or the entire MCU … I don’t think they are tooled or trained in replacing just the eMMC.

If it’s not under warranty… myself and many others are capable of removing the eMMC and recovering the critical files (assuming the chip responds when placed in a programmer), provisioning a new eMMC, and reinstalling it back on the tegra board and putting it all back together.

Yes, they are. It's part of the EMMC RECALL. Not sure if they will touch it now that he messed with it though.

 

[M0rpH3x]

If the MCUs are not rooted, the CIDs won’t respond to ping (maybe they will just a few seconds during boot).
Did you exchange just the displays? Maybe you need to check the connections. You can do it on the bench, the CID should boot fine.
Did you accidentally break something when removing the displays? Those wires are connected with latches that sometimes are easy to miss.
If it comes to that, I have a rooted MCU1 I could part of, with US LTE modem. I also have a Tegra board to spare with a good chip that will last forever. Although both of them are rooted, we would need someone to work on it and on your car (I normally work with @LayZ, who is fast, great, fair, reliable, honest, and anything you can think of). But I hope it’s just some silly sloppy connection.

 

[cloneguru]

Yes, they are. It's part of the EMMC RECALL. Not sure if they will touch it now that he messed with it though.

I’m aware of the recall. Do you have supporting information stating they replace the eMMC on the Tegra board as opposed to replacing the tegra or MCU?

I’m sorry but I don’t believe service centers do intricate solder work when they can just replace a board.

 

[M0rpH3x]

I’m aware of the recall. Do you have supporting information stating they replace the eMMC on the Tegra board as opposed to replacing the tegra or MCU?

I’m sorry but I don’t believe service centers do intricate solder work when they can just replace a board.

They replace the whole tegra board. Tesla might be doing fine resoldering work because they might be running out of tegra boards, but certainly not at the SC level. They might take your broken board somewhere to fix it, and offer it to the next guy.