NvFlash 2.6 hangs at bootloader downloaded

Dear Support-Team,

I have one Apalis T30 V1.1b, which suddendly stopped working at a customer.

When trying to flash it NvFlash tool hangs at this position:

Nvflash version 2.6 started
rcm version 0X30001
System Information:
chip name: unknown
chip id: 0x30 major: 1 minor: 3
chip sku: 0xb1
chip uid: 0x0000000000000000015c62d68728260b
macrovision: disabled
hdcp: enabled
jtag: enabled
sbk burned: false
dk burned: false
boot device: emmc
operating mode: 3
device config strap: 0
device config fuse: 0
sdram config strap: 0

Sending generated BCT…533Mhz Done
downloading bootloader – load address: 0x88300000 entry point: 0x88300000
sending file: loader.nb0

  • 65536/467360 bytes sent
    \ 467360/467360 bytes sent
    loader.nb0 sent successfully
    waiting for bootloader to initialize
    bootloader downloaded successfully

Do you have an idea, what that could be and what could happen on customer side?

Hi, I’m the Toradex AI Assistant.

I might make mistakes or occasionally get you wrong. Our support team will step in if needed.


The NVFlash hanging issue you’re experiencing with your Apalis T30 V1.1b can be caused by several factors. Based on the information available, here are the main potential causes and solutions:

Recovery Mode Connection Issues

First, ensure proper recovery mode setup (1):

  • USB Connection: Make sure you’re using a USB 2.0 High-Speed (480Mbps) connection (1). The recovery mode only works with USB 2.0 High-Speed (1).
  • USB_DET Pin Issue: The USB_DET pin (SODIMM Pin 137) needs to be very close to 3.3V when entering recovery mode (1). If your customer board has high impedance voltage dividers, this can cause the recovery mode to fail (1).

Known NVFlash Issues

RAM Timings Problem

If you’re using NVFlash version 2.1 or 2.2, there’s a known RAM timings issue with Colibri T30 modules (2). While your issue is with Apalis T30, similar problems could occur. The issue was fixed in NVFlash version 2.3 (2).

Bootloader Initialization Hang

The fact that NVFlash hangs after “waiting for bootloader to initialize” suggests the bootloader is not responding properly after being downloaded to RAM.

Troubleshooting Steps

  1. Verify Recovery Mode Entry: Ensure the module is properly in recovery mode by checking if it appears as an NVIDIA device:

    lsusb | grep -i nvidia
    Bus 002 Device 033: ID 0955:7130 NVidia Corp.
    

    (3)

  2. Check USB Connection: Use a different USB cable and ensure you’re connected to a USB 2.0 port (1).

  3. Hardware Recovery Mode: Try entering recovery mode using the hardware mechanism by shorting the recovery pads on the module during power-on (1).

  4. Update NVFlash: Use the latest NVFlash version (2.6 appears to be what you’re using, which should be fine) (2).

Possible Hardware Issues

Given that the module “suddenly stopped working,” there could be hardware damage:

  • eMMC corruption or failure (4)
  • Power supply issues
  • Physical damage to the module

The fact that the chip information is being read correctly (chip id: 0x30, chip uid, etc.) suggests the SoC is functional, but there might be issues with the eMMC or other components.

Next Steps

  1. Try the hardware recovery mode method as described in the recovery mode documentation (1)
  2. Test with a different carrier board if available
  3. Verify the power supply is stable and within specifications
  4. If the issue persists, the module may have hardware failure and require replacement

The hanging at bootloader initialization typically indicates either a communication issue between the host and module, or internal module hardware problems.

Best regards, Toradex AI Assistant.


Did you like my answer? :+1: :-1:

If I answered your question, please tick the “Solution” checkbox.

For follow-up questions to me, please use @ToradexAI in your post.

Hello @JanP !

Just to confirm, but what is the OS being used on the module, Linux or WinCE?

This will give us some insight on how to proceed with a possible solution.

Hi @joaom.tx

WEC2013. I created wince tag to this topic.

Hi @JanP ,

I see that the loader.nb0 downloaded successfully but then does not start. This points to a problem with either the CortexA9 CPU or the DDR. Both of those are not used during the first phase when loader.nb0 is downloaded BootRom will run from ARM7TDMI CPU and internal SRAM.

Is there any visible damage on CPU and RAM chips (chips off or burn marks)? Could be also a broken Crystal (12Mhz)

If the unit is still under warranty you can start a RMA process and we will get a a replacement. Can you send the serial number so i can check?

Hi @germano.tx

Even under a magnifying glass, I can’t see anything unusual. Attached is another photo of the front and back.

I’m not sure if the device is still under warranty. We produce the devices for stock, and they might sit there for a while. Perhaps you could check using the serial number.

We replaced the device for the customer. My immediate concern is to rule out the possibility of this happening again.

Hi @JanP ,

I see in our System that you bought this Apalis T30 from DataModul, not directly from Toradex.

I’m not sure what are their warranty conditions, and when you bought it from them, you better check with them directly.

I can only tell you that this unit was manufactured in 2022 so i guess it will be out of warranty, unless you bought it only recently from DataModul

What i also see is that you are using both iMX6 and T30 modules. They come with different Heatsinks which are adjusted for the different modules, make sure that those are not swapped as it could damage the modules (by pressing on componets on the wrong place or not properly contacting the CPU to cool down.

Hi @germano.tx

Yes, that’s possible. With the end of life of the T30, we made a few changes to our procurement. I’m aware that there are different heatsinks for the two CPU platforms, and a mix-up is absolutely out of the question.

Nevertheless, were you able to see anything in the pictures? Or could you mark the areas on the chip where you suspect the damage?

I just want to determine whether it was an assembly error or simply a manufacturing defect, which, of course, can always happen.

Thank you for your understandig and patience.

Hi @JanP ,

From the pictures you sent i cannot see any damage, but it’s also hard to say from these pictures alone… would need more high resolution pictures and from divverent angles to check for bent or lifted components. Also some kind of damage might not be even visible (for example soldering contacts tear off unter the BGA components)

Hi @germano.tx

Okay, thanks!

I just saw that the customer’s system went into operation on August 24, 2023, and is therefore well outside the warranty period. Since this has never happened before and your products have been absolutely reliable so far, I’m assuming this is a one-off occurrence. You never really know what might have happened on-site at the customer’s location.

Thank you very much for your support!