This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

IWR6843: Intermittent hang at startup in SOC_waitBSSpowerUp

Part Number: IWR6843

We are using IWR6843 ES1.0 samples on custom PCBs very closely following the IWR6843ISK layout & BOM, running the lab0015 firmware from mmWave industrial toolkit 3.6.2.

Startup is not reliable, on some boards the firmware only sometimes reaches the point of starting the CLI.

I have now built the firmware from source in CCS, and using the XDS110 debugger I have identified that when this happens, both the DSS and MSS cores are hanging in SOC_waitBSSpowerUp at the following line:

while (CSL_FEXTR(ptrSOCDriverMCB->ptrTopRCMRegs->SPARE0, 18U, 16U) != 0x7)

with call stacks as shown below:

I have found two previous reports of BSS startup problems:

IWR6843: Problem with BSS power up

IWR6843ISK: MMWAVEICBOOST and IWR6843ISK Unreliable

The TI answer to the second of these mentions a problem with SDK 3.0 which is fixed in SDK 3.1:

There was a firmware issue identified in SDK 3.0.0.8 (and hence all binaries built on top of it) which could cause the power up sequence to hang in the loop you mentioned. EVMs shipped from the assembly house are flashed with the SDK 3.0.0.8 OOB demo and need to be updated with the corresponding demo from SDK 3.1.1.2 in which the firmware issue was resolved.

In fact we had similar intermittent problems with earlier versions of the lab0015 firmware running on TI's IWR6843ISK boards, and those problems went away when we used the new prebuilt image from mmWave industrial toolkit 3.6.2.

However it seems we now we are getting the same symptoms on our own boards with the 3.6.2 version of the code.

My questions are:

1. Do you have a solution?

2. What was the firmware issue fixed in SDK 3.1.1.2 which could cause a hang in this loop?

3. Are you sure the 3.1.1.2 change definitely fixed the problem?

4. What factors could be causing the BSS to sometimes fail to start up?

  • Hi Martin,

    I am discussing this with other engineers on my team, we will have a response soon.

    Regards,

    Justin

  • Hi Justin, many thanks for your help on this  - I work with Martin on this project - Martin is currently on leave.  Could you please provide me with an update as soon as you have on eon this issue as we are at a critical point in our project.

  • Hi Martin,

    The firmware update in SDK 3.1.1.2 improved the robustness of the APLL clock calibration procedure. Prior to this version, the APLL calibration, which is done by BSS during power-up could fail on certain IWR6843 ES1.0 devices, causing the BSS power-up to hang-up in the above loop. The firmware update made the calibration procedure deterministic.

    When you say you are now seeing the same issue on your boards with demos with the updated firmware (i.e. SDK 3.1.1 and above), did these specific boards work previously and only now showing the problem with nothing else changed in software? Or are these new boards that show the problem?

    Regards

    -Nitin

  • Hi Nitin,

    Nitin Sakhuja said:
    When you say you are now seeing the same issue on your boards with demos with the updated firmware (i.e. SDK 3.1.1 and above), did these specific boards work previously and only now showing the problem with nothing else changed in software? Or are these new boards that show the problem?

    These are new boards we have had fabricated, populated with ES1.0 ICs. The layout and BOM closely follows the IWR6843ISK design.

    Earlier, on IWR6843ISK boards we received from TI (also with ES1.0 ICs), we had startup problems with earlier versions of the firmware. These problems were solved when we updated to the pre-built lab0015 firmware from version 3.6.2 of the mmWave industrial toolkit.

    On our own boards, we have seen the intermittent startup on 3 out of 5 of our PCBs, using the same lab0015 firmware from mmWave industrial toolkit 3.6.2. The other 2 start up reliably. We have seen this early in testing, it is not something that has developed over time.

    Have we missed some step? Is there any calibration/configuration that needs to be completed or flashed to the IWR6843, separate to the firmware image, which would have been done by TI for the IWR6843ISK but which we have not done on our boards?

    Are there any changes in the ES2.0 silicon that may be relevant, or resolve this issue?

    Regards,

    Martin

  • Hi Martin,

    Thanks for the confirmation. 

    Could you please follow the debug instructions provided in the following response and provide the requested register values from the non-working board(s)? Please flash the lab0015 binary from Toolbox version 3.6.2 as you have been using so far:

    https://e2e.ti.com/support/sensors/f/1023/p/844780/3124873#3124873

    Thanks

    -Nitin

  • Hi Nitin,

    I get the same values as reported by Lola, see below:

    Regards,

    Martin

  • Hi Martin,

    Thanks for providing the requested debug information. The values you provided indicate that this IWR6843 ES 1.0 device has an irrecoverable failure during boot-up. We have identified that this issue affects some IWR6843 ES1.0 devices and it has been resolved in ES2.0 devices.

    I would suggest you to verify the problem signature on non-working boards and if it matches the above, please replace those devices with ES2.0 silicon which is now available on TI.com. 

    Regards

    -Nitin

  • Hi Nitin,

    I'm surprised that this fault doesn't seem to be mentioned in the IWR6843 errata (SWRZ087A) given that apparently it is a known issue fixed in ES2.0, and appears to affect a large fraction of our samples.

    I see that you have offered to replace Lola's IWR6843ISK board with the ES2.0 version, will TI do the same for our remaining XI6843QGABL ICs? We have some that are still in unopened packaging which were awaiting placement on further trial devices. Without a way to reliably detect which are affected by this issue we will not be able to use these.

    Regards,

    Martin
  • Hi Martin,

    This issue was discovered recently and found to be affecting some ES1.0 devices. Due to the inherent improvements in ES2.0 and the devices following the full production quality cycle, it is known that the issue cannot happen in ES2.0 which is the production version of IWR6843.

    ES1.0 devices are Pre-RTM/experimental devices meant to support early evaluation and development. Pre-RTM devices may not be representative of the specifications of the final production silicon and may not use production assembly flow. Please refer to the following links for TI's standard policy for Pre-RTM devices.

    TI's Product Life Cycle
    www.ti.com/.../product-life-cycle.html

    IWR6843 Device Datasheet (Section 8.1 Device Nomenclature)

    To avoid the possibility of this issue, we recommend that you migrate to the ES2.0 device.

    Regards
    -Nitin