KV serial console

[Cex]

Hi all,

I have a KV (Wifi only) that got bricked during an update (boy and tree on screen).
I've been reading the forums for a while, and I asume that the eMMC may had been destroyed due to low battery during update.

The Kindle boots directly into USB download mode (SE blank MAGREZ HID device on connection to a PC).
I connected a 1.8V converter to the serial console, but I do not get any output (I also checked with an oscilloscope, but no output on TX pin, it even looks like there is not 1.8V in that pin, which should be as it idles at high level).

Assuming the eMMC is completely black, or UBOOT damaged (no UBOOT), or damaged...there should be any output on the console port? Is UBOOT initializing the serial port? I've been checking some logs from other users and it looks like first message ouput is the UBOOT version.

Could someone provide an UBOOT image for the Voyage suitable to be used with MfgTool? I have seen a lot of images for KT, but I'm not sure about the SoC I/O connections being the same and if I could damage the HW by loading an UBOOT image from another model.

My goal is checking if the eMMC is ok or not, and trying to read the IDMEs (I only have the S/N B0130907xxxxxxxx, but have no backup of other idme parameters).

If the eMMC IC is dead I could consider replacing it as I have access to a BGA rework station.

Thanks in advance.

[knc1]

There should be output on the serial port, even with a dead eMMC chip -
That is, the first part of the initialization is run in the SoC's internal ram from code in its internal rom.

Re-check your serial port connections.

The thread about recovering a Kindle with fastboot has many logs attached to its posts, see those for what to expect. (the thread has 'fastboot' in its title, search for that).

[Cex]

Thank you for your reply.

I already had read that thread and all the logs seem to start with Uboot messages.

Although I have good soldering skills I'll check my work again, but I'm afraid somehow I toasted the TX pin in the SoC

[geekmaster]

Originally Posted by Cex

Thank you for your reply.

I already had read that thread and all the logs seem to start with Uboot messages.

Although I have good soldering skills I'll check my work again, but I'm afraid somehow I toasted the TX pin in the SoC

What serial adapter are you using? If it has a TX LED on it, you may need to remove that (or replace it with a suitable pull-up resistor) because the kindle TX pin cannot sink enough current to pull an LED-connected wire down to a logic zero (as described in my level-shifter thread).

And BTW, "SE Blank <xxx>" is typically known here is "USB downloader mode", so you may need a modern version of "MfgTool" (or other usb downloader tool) to flash your firmware. This mode does not use the serial port, as I understand it. It was probably triggered by a failed firmware update, so your firmware is now corrupted. This can be caused by updating the firmware on a kindle with a low battery (which depletes before completion).

Also, "eMMC destroyed" implies a hardware failure, when it is probably just the contents that are corrupted (due to incomplete firmware update), not the physical device.

[Cex]

Originally Posted by geekmaster

What serial adapter are you using? If it has a TX LED on it, you may need to remove that (or replace it with a suitable pull-up resistor) because the kindle TX pin cannot sink enough current to pull an LED-connected wire down to a logic zero (as described in my level-shifter thread).

And BTW, "SE Blank <xxx>" is typically known here is "USB downloader mode", so you may need a modern version of "MfgTool" (or other usb downloader tool) to flash your firmware. This mode does not to the serial port, as I understand it. It was probably triggered by a failed firmware update, so your firmware is now corrupted. This can be caused by updating the firmware on a kindle with a low battery (which depletes before completion).

Also, "eMMC destroyed" implies a hardware failure, when it is probably just the contents that are corrupted (due to incomplete firmware update), not the physical device.

I'm using a FT232RL-based adapter with a 3.3 to 1.8V supply for the Vio (not using the kindle for that). It has LEDs, but they are driven by dedicated pins of the FT232RL, not directly connected to the TX/RX lines. I have measured voltages and they are correct (1.8V) and shorting RX to TX on the adapter also works. There is also an small chance that I shorted TX to GND while trying to check for an output as the pad in the KV is very close to the metal frame, so it is also possible that I have blown the output driver of the SoC as it should rest at 1.8V when idle and I can't measure any voltage on that pin.
It is also possible that when entering USB downloader mode the SoC set all its I/O pins as inputs, or disabled, but I don't know.

I mentioned the eMMC "destroyed" as it is explained by KNC1 on another post. Depending on when the supply fails the internal indexes of the eMMC may go wrong rendering the IC unusable (although I suposse it would be possible to read back from it). As it looks like UBOOT is not present (no output on the console even if I press a key during start-up) I assumed the whole eMMC has been erased or is failing.

I have downloaded the MfgTool2 that is supposed to support my device. I intend to follow the steps on the "PW2 eMMC upgrade" thread, but I need to build the uboot image to test and were wondering if someone could provide such an image to check if at least the device is recoverable.

Thank you very much for your answer. I'll try to build the image and check with MfgTool if I get something from the serial console port or I can flash fastboot.

[knc1]

Originally Posted by Cex

Thank you for your reply.

I already had read that thread and all the logs seem to start with Uboot messages.

Although I have good soldering skills I'll check my work again, but I'm afraid somehow I toasted the TX pin in the SoC

Think it through -

If the eMMC can not be read -
The start-up code can not learn the DRAM parameters required to start the DRAM -
The SoC will generate a message on the serial port to that effect and shut down.

= = = =

Also see the thread about replacing the eMMC with a larger device.
That larger device came from the factory blank -
Which should be (almost) the same firmware path result as a dead device.
Here, "firmware" means that built into the SoC rom.

= = = =

Buy another KV - they are still being made -
Follow my thread on recovering the initial update_*.bin from it -
Follow that with making a copy of the 'diags' kernel and 'diags' system.

G.M's "get kernels' script will do the kernel recovery for you.

Note:
The 'diags' system and kernel are written to the eMMC while they are still on the tape (before being soldered to the board).

So making your own copy is the only (legal) way to go.

Note:
No hurry on that, your KV will run just fine without a 'diags' kernel and OS installed.

- - - -

What to do with the second KV?
Resell it, give it to a good cause, keep it - your choice.

Don't like the cost involved in the above procedure?
Well maybe it will remind you to make back-ups before doing anything invasive next time.

[Cex]

Originally Posted by knc1

Well maybe it will remind you to make back-ups before doing anything invasive next time.

The KV bricked itself before doing anything. The invasive serial port connection was made to try getting it back to life.

Thanks anyway for your suggestion. I'll take into account for my next device.

[knc1]

Originally Posted by Cex

The KV bricked itself before doing anything.
- - - -

Return it for exchange.