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.

Please help, CCSV4 Target Configuration for TMS320DM6467T using Spectrum Digital XDS510 USB

Other Parts Discussed in Thread: CDCE949

I have a new board, the design of which I know works, because I got one board totally working already.

However, now I'm having trouble getting another board to work, and part substitutions introduce uncertainties for me.

CCS: Version 4.2.3.00004

Host: Windows MCE

Target: Custom board with either TMS320DM6467ZUT7 or TMS320DM6467TZUTD1

Previously, I chose for my target configuration "Spectrum Digital XDS510USB Emulator" and "EVMDM6467".  After all, I used the Spectrum Digitial HD 1080P EVM (aka DM6467T EVM) as my design reference.  My previous board carried a TMS320DM6467ZUT7 (aka DM6467 without the "T").  This all worked fine.  I was able to connect to the target, load the program "nand_flash_writer.out", and flash my NAND memory.  In detail, note that after loading the program, the ARM thread status is "Suspended".  Then selecting Target / Run will run the nand_flash_write program and do its thing correctly.

Now, I have a PCB from the same manufacturing run, but it's stuffed with a TMS320DM6467TZUTD1.  I've changed the outboard power supply from 1.2V to 1.3V as I should.  I've also removed the 33MHz crystal and connected the same jumper as appears on the EVM, to route the CDCE949 clock's 27MHz signal into the core instead.

I can connect to the target.  However, when I load the program, upon completion of the load the ARM thread status is "Running", not "Suspended".  I'm stuck right at this point.

Looking closer, I wonder if the target configuration is the correct one.  I notice that the "EVMDM6467" does not end with a "T".  I can not find ANY option in the General Setup for target Configuration that ends with a "T.  Is this my problem, or is my problem something else?

Please note another strange thing.  Using the prior board with the DM6467, my system power supply is providing 160mA at 12V before I connect.  After connecting both the ARM (first device of four) and the DSP (second device of four), my system current increases to 240mA.  After loading and running the nand_flash_writer program, but while the program is idle and waiting for the first user input, the system current remains 240mA.  (Remember, this is with a 1.2V core supply.  System current includes all the voltages and converter inefficiencies.)

On the other hand, using the board with the DM6467T (and a 1.3V core supply), my system power supply is providing 330mA at 12V before I connect.  Is this due strictly to the increased core voltage?  As I write this email, it has climbed from 320mA to 340mA...  now 350mA.  The top of the processor is very warm.  Now 360mA.  Once I connect to the ARM (first device of four), this current increases to 470mA.  Connect to the DSP (second device of four), and it's still 470mA.  Both devices "Suspended".  Now 480mA.  Now I load the "nand_flash_writer.out", and current remains 480mA.  Now 490mA.  But ARM (first device of four) status is "Running" when it should have been "Suspended".  Now 500mA.  Chip surface reads 179 degF with an infrared thermometer gun.

Is this higher current and temperature situation simply a fact of life for the DM6467T vs the DM6467, or is this also indicative of a problem?

I've been struggling with this for several weeks now, and am at a total development standstill because of it.  Your detailed assistance is GREATLY appreciated.

-Helmut

 

  • Helmut,

    Helmut Forren said:

    Now, I have a PCB from the same manufacturing run, but it's stuffed with a TMS320DM6467TZUTD1.  I've changed the outboard power supply from 1.2V to 1.3V as I should.  I've also removed the 33MHz crystal and connected the same jumper as appears on the EVM, to route the CDCE949 clock's 27MHz signal into the core instead.

    I am pretty sure the change in devices will impact any initialization routine. Either if you are doing this programatically or via GEL files, I strongly suggest you checking the GEL files for both DM6467 EVMs at Spectrum Digital support webpage:

    http://support.spectrumdigital.com/boards/evmdm6467

    http://support.spectrumdigital.com/boards/evmdm6467t

    Helmut Forren said:

    I can connect to the target.  However, when I load the program, upon completion of the load the ARM thread status is "Running", not "Suspended".  I'm stuck right at this point.

    That is an indication that the device initialization may not be happening correctly. To further investigate this, can you disable the autorun to main (check here).

    Helmut Forren said:

    Looking closer, I wonder if the target configuration is the correct one.  I notice that the "EVMDM6467" does not end with a "T".  I can not find ANY option in the General Setup for target Configuration that ends with a "T.  Is this my problem, or is my problem something else?

    From an emulation/debugger perspective the devices are the same, therefore the target configuration is the same.  

    Helmut Forren said:

    Is this higher current and temperature situation simply a fact of life for the DM6467T vs the DM6467, or is this also indicative of a problem?

    The two devices share the same silicon process technology (90nm / Cu), therefore a clock scale is followed by an increase in temperature and current. I would, however, send the details to the device forum so you can get more knowledgeable guesses there. 

    Hope this helps,

    Rafael

  • Rafael,

    Thanks very much for the suggestions.  I'll research the links you provided.  

    Before that, for right now however, it seems your advice is contradictory.  You said "From an emulation/debugger perspective the devices are the same, therefore the target configuration is the same".  Yet you suggest "checking the GEL files for both DM6467 EVMs".  Am I missing something? 

    Maybe what I'm missing is a nuance between target configuration and gel files.  I'm slow on gel files.  I'm only using what came with.  I though the gel files came with the configuration.  Are you saying the configs may be the same but the gel files different?  (The gel file stuff is very OBTUSE!)

    Thanks again,

    Helmut

  • I'm using CCSV4.  I can't figure out how to see or do ANYTHING directly with GEL files.  The word GEL does not exist in any of the menus.  The help documentation has bad links for GEL details.  It's NOT at Tools / GEL...

    Please tell me or remind me where to access GEL files, load them, select them, or maintain them, from within CCSV4.  A link to some valid doc would be very nice as well!

    Thanks very much,

    Helmut

  •  

    Hi Helmut,

    GEL Documentation can be found in the included Help files ('Help->Help Contents'):

    Helmut Forren said:
    Please tell me or remind me where to access GEL files, load them, select them, or maintain them, from within CCSV4.  A link to some valid doc would be very nice as well!

    See this wiki topic: http://processors.wiki.ti.com/index.php/GEL

    Thanks

    ki

  • Another good GEL doc on how they are used to auto-initialize a target when starting the debugger.

    http://focus.ti.com/lit/an/spraa74a/spraa74a.pdf

     

  • Thanks.  Neither search nor index revealed this help location, and contents navigation didn't show the word GEL at the end of the long line in a narrower window!  And I didn't notice the spelled out General Extension Language.

  • FYI, the "GEL Files" option does not appear on the Tools menu at all, not even grayed out, unless your target configuration is currently launched.

    I got to first base once, but not again...

    My target config was already launched.  I loaded the GEL file, which came from Spectrum Digital as a newer version than the one I already had on disk.  I connected to ARM and DSP targets.  With the ARM selected, I Tools / Load Program the "nand_flash_writer.out".  This time, lo and behold, after the load the status was "Suspended" as it should be.  I then ran the program and I believe successfully wrote UBL to NAND.

    So then I tried right after that to write U-Boot to NAND, but had no luck.  Note I never literally "RAN" the GEL file.  I just loaded it and later connected and loaded program.  Does CCSv4 know to use this GEL I loaded rather than one that came with the config?

    Anyway, I exited CCSv4, unplugged the XDS510USB, and powered off my target, to retry.  Loaded CCSv4. Plugged in the XDS510USB usb cable.  Launched my target configuration.  Tools / GEL files then loaded my GEL file.  Finally powered on target last, to avoid overheating.  Connected to ARM.  Connected to DSP.  On ARM, loaded program.  "Running" status.  Still not working.

    Am I having hardware trouble still, cold or hot?  Or am I not getting the GEL file in properly, except for the first time?

    Thanks very much,

    Helmut

  • Filename snafu, but still doesn't work.  Figured out GEL file view changes as you point to different devices in debug window.  But still having errors, even with correct GEL file.