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.

Problem when connecting ARM9 of OMAPL137

Hi,

I'm using CCS 3.3 and ICETEK 560USBPLUS Emulator to develop application on OMAPL137.  DSP 674x core can be connected correctly, and we can download/debug application on it. However, when i try to connect the ARM core, it show an error of "SC_ERR_ACCESS_SUBPATH".

According to the suggestion of FAE of ICETEK, I opened CCS for DSP firstly, load an evmomapl137_dsp.gel file, and "Enable_ARM" under GEL menu. Then I tried to connect ARM on "parallel debug manager",  it still failed, reported an error of  "SC_ERR_ACCESS_SUBPATH" and the connection with DSP also got disconnected.

The L137 board is self-designed, maybe there is something error with the design? Or it is because of the emulator?

Thanks.

  • Do you have an EVM that you can test your connection?

  • No, I don't have an EVM.

    This moring I found a difference between my board and EVM. The crystal used on my board is 25MHz while which on EVM is 24MHz.

    I suspect the frequency for ARM after PLL is a little bit higher (312.5MHz), so I changed the PLLM value in GEL file.  After the change, the ARM core was connected successfully for the first time, and all the four parts (ICEPIK, DSP, ARM and ETB) all show that they were connected well.

    However, the connection got broken after a re-power for the board. And, it remains the state that DSP core can be connected while ARM core cann't what ever I changed in the GEL file.

    Also I tried to lower the TCK frequency from 21.25MHz to 5MHz, no effect.

  • When connecting ARM after DSP connected, CCS for DSP show the following error information:

    "Power Failure on Target CPU: Error 0xA0002020/-1044 Error during: Execution, Control,  Device driver: Cannot acquire emulator process Possible conflict for device driver usage."

    I tested waveform in 1.2V Vcore and 3.3V Vio by oscilliscope and found some noise increase on Vcore (from ~60mVp-p to ~80mVp-p) when connecting DSP core. And there's nothing changed on the waveform when connecting with ARM.

  • I think the problem is solved up to now.

    The reason is that due to size limitation of PCB, we use an old JTAG header converter board for TI's DSPs. On this board, TCK and RTCK signals are tied together. it's no problem when used for DSP emulation.

    However, the SPRA439d.dpf told us:

    "However, on devices that include an ARM core of version ARM926 and above, the TCK_RET signal must
    actually come from a dedicated pin of the chip itself. This is because ARM Ltd varies from the IEEE
    1149.1 specification and makes the JTAG clock dependent on the CPU clock. Therefore, the RTCK signal
    from the ARM core is actually in internal synchronization with both the TCK signal from the JTAG header
    as well as the CPU clock."

    That is the real reason.

    After spliting the two signals, conneting RTCK correctly and lower TCK frequency (RTCK's signal quality is very poor due to long jumper wire) to about 5MHz, all the cores can connected weill.

    -Bradley