CCS6 install no longer allows debugging of MSP430G2553 over MSP-FET430UIF

Additional Data:

A virgin CSS project with the following main.c was tested:

int main(void) {
WDTCTL = WDTPW | WDTHOLD; // Stop watchdog timer
return 0;
}

I could get this simple code to debug in CSS a few times, but never twice in a row:  I always had to unplug the USB and re-plug it in, and even that was no guarantee.  The most common error messages were:

Load Program Error

File ..... .out: Load failed"

AND ...

Error connecting to the target:

Unknown device

As mentioned, all of this worked fairly well under my prior CCSv5 install.  

Hello Katie,

Yes, I fear your guess:  Maybe my PCB configuration was only marginally working with CCSv5 and some innocuous change has now tipped it over the edge.  That said, I will answer your questions as best I can below, and then will provide some more data

1) See attached diagram.  You might wonder about the extra wires going off screen attached to TDO and TDI:  These are connected to high-impedance transistors that are used in another portion of the circuit.  That might be the problem, but it has never given a problem before.  

2) I can configure the board to be powered by the FET or by its own power supply.  Both modes have always worked with the debugger.  By changing the connector cable between my circuit and the FET, I can direct the PROBE_PWROUT line to either the JTAG Vcc(Tool) pin (for powering by FET) or JTAG Vcc(sense) pin (for powering by my circuit).  Both modes have always worked.  For example, for my MSP430GANG programmer, I let the gang programmer power everything.  Works without fail.

3) Yes, only one MSP processor on this particular target.

4) Tested that today with a MSP430G2231.  No problems.  Also tested my MSP-FET430UIF on an old MSP430F2618 Dev board, and then on an MSP430F5529.  All of them debug, so the MSP-FET430UIF is (thankfully) not the problem.

5) Excellent idea, and I was worried about it.  My custom cord was indeed over 8 inches, so I gave it a trim, and re-wired.  Sad to say, this did not fix the problem.

6) Version: 6.1.2.00015 for CCS.  The text for "MSP430 Emulators" is "CI 6.2.1.0 including MSPDebugStack 3.5.1.001".  The "MSP430 Compiler Tools" has version 4.4.6.  Unfortunately, in order to avoid driver incompatibilities, I removed CCSv5 from my computer.  (Also had something to do with running out of disk space on my C: drive).

* * *

Now, here's another possible idea:  Could there be something with the new CCSv6 install that is now mistaking JTAG 4 wire with SBW ?  Apparently, there is some auto-magical way in which CCS figures out which protocol to use.  Perhaps it is getting confused in my case?  

One other piece of data:  Elprotronic software seems to be able to communicate with my PCB and I can burn hex images with it ... albeit with an occasional burp.

Thank you for helping me look into this.  Again, I suspect my PCB board might be the culprit, but I still want to figure out why everything went south with CCSv6. Perhaps some driver changed, or maybe the MSP430UIF firmware is different.  Or maybe the baud rate used has been boosted so that marginally cheesy links will get broken?  Couldn't say.

Thad.

Yes, I did some testing this morning:  Broke connections of TDO and TDI  to other parts of circuit in order to isolate connects with the JTAG.  I also replaced that 22K with a 47K.  No dice.  Still does not work.   As nice as it would be to start using SBW instead of JTAG, I really want to try and figure out why this won't work.

My theory originally was that it had something to do with changes in the FET firmware.  To test that theory , I downgraded the firmware on the FET from V3 to V2. ( There is a batch script in the ccs_base\DebugServer\drivers directory that does this).

After doing that I was able to use the Elprotronic debugger without burps.  This then isolates the problem to an incompatibility between my PCB board and the new FET firmware.  As you say, probably timing.

So short of doing a complete re-design of my PCB boards, I'm now looking for a work around for this legacy project. What I would like to get CCSv6 to let me use this older firmware on the FET.  To do this, I replaced msp430_emu.drv and msp430.DLL with the files from the old CCSv5 directory.

Problem is, CSS still wants to upgrade the firmware on the FET before allowing me to debug.  I thought reverting the driver would solve this, but it did not.  Is there any way that CSS will allow me to use my old driver/DLL/firmware combo that works? 

Summary Details:

1) Trying to use firmware version 2.04.09.001

2) Trying to use MSP430.dll dated 9/20/2012

3) Trying to use msp430_emu.drv dated 11/20/2012

Thad.

Hey Katie,

Honestly, I probably should not have brought up this new result since my primary concern is still with the MSP-FET430UIF.  I will make a second posting if I continue to have problems with the MSP-EXP430G2 and CCSv6.

But to answer your questions:  Yes, I'm using my original (unmodified) CCSv6 installation for this result.  I can undo any of the changes mentioned above with the DLLs and drivers.

Update:  I want to follow up on this so the thread is complete.  Most of the problems here were in fact due to the CCSv6 installation.  Since I followed the steps for installation, I'm a bit worried about how things got messed up.  Please see this post for details about installation issues.

That said, it turns out that the cord length did have something to do with things. I have since made a very short connector (2") to my board, and that appears to be working.  Unanswered is exactly what changed between versions 3 and versions 2 of the FET firmware to make this an issue.  I am noting that right now for the sake of anyone else that might encounter problems on their own PCB design.