DM6467T crystal or XDS510USB problem

Oh, one more thing.  Should the Spectrum Digital XDS510 USB work with a 33MHz clock speed on the target?  That's another aspect of the changes to this newer board.  I couldn't find a limit in the user's manual.  Also, the darned ABM7 crystals are not amenable to getting off the board to switch speeds, because the pads are almost fully hidden beneath...

Now, on a second TMS320DM6467ZUTD1 board, in CCS I can launch the configuration, but when I try to connect to the target, I get the error:

Error connecting to the target:
Error 0x80000244/-2131
Fatal Error during: Register, Initialization, OCS,
Cannot access register at 0x00000000

Also, 1.2469V core supply is good on paper for both ZUT7 and ZUTD1, but I tried 1.3V anyway for the ZUTD1.  Didn't help (just got hotter, too).

Helmut,

Its better to go step-by-step to debug your board.

1> We never used CCS4 with DM6467/DM6467T. I recommend you to use CCS3.3 and make sure you are able to connect CCS.

2> You mentioned core voltage of 1.244V. Note that core voltage is 1.3V for DM6467T and 1.2V for DM6467.

3> Are you using the same crystal as on DM6467T EVM?

4> Do you have DM6467T or DM6467 EVM? What p-p voltage you see on the EVM?

5> Is your crystal meeting "Crystal Input requirements" mentioned in DM6467T data sheet (http://www.ti.com/lit/gpn/tms320dm6467t) Table 7-5 (Page 151)? Are you also meeting CL = (C1C2)/(C1+C2)?

Regards, Srirami.

Srirami,

Thanks very much for your advice.  I was off this project between when you replied and now, but now I'm back on.  

Please considered my correspondingly numbered responses...

1) Regarding CCS3.3 vs CCS4, I was successfully using CCS4 on a prior assembly of the same board (using ZUT7 chips).  Regarding changing to CCS3.3, the whole CCS software tool is so painful in setup, licensing, and operation, that I really, really dread having to open that bag of worms.  Furthermore, I had to send off my "golden" ZUT7-carrying board, so I have lost my good working baseline.  Changing CCS at this point seems unwise.  Again, CCS4 was working.

2) The 1.244V core voltage satisfies the spec sheets for BOTH the TMS320DM6467ZUT7 and TMS320DM6467TZUTD1.  Nevertheless, I did modify one of my separate power supply boards (FFC connected) to produce right at 1.3V.  This didn't help.

3) ((Note later you'll find the crystal I'm using doesn't matter.  See point #6))  The DM6467*T* EVM uses CS10-33.000MABJ-UT but it's not in stock at Digikey, so I'm using ABM7-33.000MHZ-D-2-Y-T.  Note that previously I was using ABM7-27.000MHZ-D-2-Y-T successfully; the same for me but at a different frequency, and still not equal to the EVM BOM.

4) ((Note later you'll find the crystal I'm using doesn't matter.  See point #6))  I have a DM6467T EVM, aka Spectrum Digital HD1080P EVM.  On the EVM for the 33MHz crystal, I read 1.64V p-p.  

5) ((Note later you'll find the crystal I'm using doesn't matter.  See point #6))  

NEW 6) I have subsequently REMOVED the 33MHz crystal from my board, and shorted the jumper inherited from the EVM reference design, feeding in the 27MHz clock coming from the CDCE949.  This signal measures 3.2V p-p on DEV_MXI/DEV_CLKIN and then 2.64V p-p on DEV_MX0.  I think the 2.64V on DEV_MX0 matches the EVM, because that corresponds to the drive power of the circuit inside the processor.  The 3.2V p-p on DEV_MXI/DEV_CLKIN should be plenty strong enough.  So this clock circuit should be running fine.

NEW 7) Meanwhile, the AUX clock at 24MHz I have found on some other board isn't needed to be working at all for the JTAG connection.  Nevertheless, it's working at 2.72V p-p.

So, getting back to my original problem, I can't "Connect to Target" from CCS4.  I have two primary UNKNOWNS to consider:

1. Is my design and intention correct for the TMS320DM6467TZUTD1, it having worked for the TMS320DM6467ZUT7?
2. Is there an assembly error on my board, specifically when it comes to difficulties my board house may have with BGAs?

I'm trying to eliminate #1 in order to be able to assert #2.  So, regarding "design and intention", is the following a complete list of considerations for using the same board and otherwise same components for these two processors:

• Core power supply increased from 1.2V to 1.3V (which comes from a separate power supply board)
• Crystal circuit tentatively replaced with CDCE949 source at 27MHz; TZUTD1 should run just fine at this slower external clock

Remember, I had three of these core boards assembled using the ZUT7.  One worked great.  The other two didn't, and I believe due to BGA assy problems.  Then I had two more boards assembled, using the TZUTD1 instead, and neither works.  THEY GET PRETTY HOT, up to 150degF on the surface, as well.  I don't know if the TZUTD1 is supposed to get this hot when it's not programmed.  The ZUT7 did not.

Thanks very much,

Helmut

I guess there is a third unknown, and that is if my tools have broken.  Not having my golden board, I can't insure that they are.  I just know that they were working.

Helmut,

Sorry for my late resonse. Any progress on this? What is the latest update?

Regards, Srirami.

Srirami,

Thanks for checking on me.  A couple things...

1) It turned out I did have some XDS510USB operational reliability issues with my VMWare virtual WinXP machine.  Installing CCS V3.3 got me nowhere, but installing CCS V4.0 on the host got me very reliable XDS510USB communication.

2) With reliable XDS510USB communication, I found I could connect to one of my boards carrying a ...TZUTD1 processor.  However, I couldn't program the nand.  It also seemed like things got worse after a couple of minutes powered on.  This was suspiciously similar to a problem a prior ...ZUT7 board had, where it would run overnight with a fan on it but quit after a couple hours with no fan.  This reinforced the question of bad assembly.  The folks at www.process-sciences.com have done quality x-rays and analysis and see problems that my in-house fab x-ray folks didn't see.  So they are next going to rework a few BGA's for me.  I'm hoping to get at least one working board back.

So, reconsidering those unknowns, I'm left with:

- I had definite tool problems but I think I'm beyond that now.

- I have definite assembly problems and am hoping to get past that.

- This leaves design problems relative to TZUTD1 vs ZUT7.  I'm hoping I've done my homework right and this question dissolves.  I'm getting one of each board reworked.  If I end up with a working TZUTD1, fabulous.  If neither, yuks.  If a working ZUT7 and failing TZUTD1, then it makes we worried about design.

-Helmut

Helmut,

Thanks for the update and keep me up to date with your findings.

Regards, Srirami.