DSP56725 JTAG

[Nikko]

Good day,

I'm finishing the schematics of a board involving the DSP56725, and I've kept the best for the end.

I need now to figure out how to access the DSP on board via it's JTAG module.

I've seen a similar thread on this forum where posters were reporting some troubles to implement a parallel port/JTAG connection - which is what's on the DSPAUDIOEVM board.

My idea is a bit different because i'd pull out part of the Soundbite schematics, in particular the FT2232 chip, then attempt to debug the DSP56725 code in Symphony Studio with a USB connection.

Has anyone tried this or does anyone have reasons to believe that this would not work?

Thanks

Nick

[rocco]

Hi Nick,

Have you tried Symphony-Studio with USB? I have been told that it does not work, after filing a service request because Symphony-Studio would crash anytime I tried to connect over USB. But others seem to be using USB with no problems, mostly with the Soundbite board. Here is a snippet from my service request:

---

In reply to your message regarding Service Request SR 1-592924541: For the crash prblem problem, frankly speaking, I can do little to help you,it it the tools bug. That is why I told you to use suite56 tools. If you want to start your project, from hardware view, I suggest you use a PC with parallel port, and order a Wiggle device from the following website: http://www.freescale.com/webapp/sps/site/prod_summary.jsp?code=DSPCOMMPARALLEL

---

We later were told that Freescale had dropped support for Symphony-Studio, as mentioned here:

http://forums.freescale.com/t5/forums/searchpage/tab/message?location=forum-board%3AFORUMUSE&q=Symph...

But that was two years ago, and I believe they still offer it for download. I just couldn't get any support for it.

So we abandoned our use of USB, bought a workstation with a parrallel-port, and have been using Suit56 for our firmware development. It has its deficiencies, like all Freescale tools, but at least it runs.

But if the Soundbite is working with Symphony-Studio, and you base your JTAG interface on the Soundbite, there is a good chance it will work for you. But I would wire a "backup" 14-pin parallel-port JTAG connector between the FTDI chip and the '725, just in case. I don't think it would take any additional parts, and then the board could be stuffed either way.

I could send you schematics of my JTAG interface, if you like. The JTAG on the DSPAUDIOEVM board is a poor example, as it is bloated with components to support so many DSPs. For the '721, I needed just one pull-up resistor, which you probably already have, besides the connector.

[Nikko]

Hi Rocco,

Thanks a lot for your reply.

I indeed have the Soundbite board and it's working fine with Symphony Studio and a USB connection. I'm not experiencing any crash - rather sometimes the PC doesn't manage to connect with the board and then i need to close and restart the application and it's usually fine again.

When you tried to connect Symphony Studio to your board, was that a Soundbite or was it already your own design? In the latter case, did you simply copy - as i intend to do - the stuff around the FT2232 chip on the Soundbite or did you have your own design?

Your idea of having a backup for a parallel port on the JTAG is great - but there are a few things i don't quite understand. I've been a long time user of Suite56 some years ago and had in particular a design with the 56321.

The first thing is that the 56725 is not supported by Suite56. What i mean is that Suite56 let's you declare what DSP you're using, and that none of the dual-core are in the drop-down list . Also, even it's not critical to declare the proper DSP type, i can't imagine how Suite56 would handle the 2 cores?

The second thing is that the JTAG is not quite the same between the single core and dual-core. In particular, dual-core don't have the TRST pin anymore. How can you go around that with Suite56?

I have seen your implementation of the JTAG for your DSP56721 board, and it's slightly different from the EVM. The main difference is that there's nothing on pin 14 because there's no TRST signal. Also the 10k pullup on pin 13 is not present on your schematics. Can you tell what it is for?

Finally there's a 10k pullup on pin 9, which is connected i suppose to the DSP RESET pin.

Greetings

Nick

[rocco]

Hi Nick,

I will try to answer some of your questions.

---

Nikko wrote:

When you tried to connect Symphony Studio to your board, was that a Soundbite or was it already your own design?

---

I bought the Xverve Signalyzer (DT-USB-ST-K002), mentioned here and recommended by my Freescale rep. The board was my own, with the JTAG interface verified by Freescale tech-support, in another long service request. It seemed to work, as I could connect, but Symphony-Studio would crash when I tried to debug (typically, as soon as I pressed RUN, STEP, or anything else that attempted to run code).

---

. . . but there are a few things i don't quite understand.

The first thing is that the 56725 is not supported by Suite56. What i mean is that Suite56 let's you declare what DSP you're using, and that none of the dual-core are in the drop-down list . Also, even it's not critical to declare the proper DSP type, i can't imagine how Suite56 would handle the 2 cores?

---

Yes, Suite56 does not know about the 56725 or the 56721. But since each core is a regular 56300, the two cores appear as two separate 563xx DSPs in a JTAG chain. So you can convince Suite56 to access both cores with the following two lines in the debugger's startup file:

device dv0 cc0 tms0 pos0 56366
device dv1 cc0 tms0 pos1 56366

So I can switch between cores in Suite56 by selecting either dv0 of dv1 in the debugger. Note that I am telling the debugger that they are both 56366 DSPs.

One annoying shortcoming is that Suite56 has no concept of the shared-memory between the cores. The things we needed to do to get around that could be a topic of a whole-nuther thread.

---

The second thing is that the JTAG is not quite the same between the single core and dual-core. In particular, dual-core don't have the TRST pin anymore. How can you go around that with Suite56?

---

I'm not sure how it works, but it does. My understanding from tech-support is that there is a reset command through the JTAG port that the wiggler uses as well as the TRST pin, so the TRST is no longer necessary, especially since there is also a DSP reset on pin 9. I've not looked into it, because it has always worked.

---

I have seen your implementation of the JTAG for your DSP56721 board, and it's slightly different from the EVM. The main difference is that there's nothing on pin 14 because there's no TRST signal. Also the 10k pullup on pin 13 is not present on your schematics. Can you tell what it is for?

Finally there's a 10k pullup on pin 9, which is connected i suppose to the DSP RESET pin.

---

Pin 13 is the /DEZ signal, and I don't think it has been used in years. I had to refer back to my service-request for JTAG issues, because I just don't remember. Here are two of the responses to that service request:

---

In reply to your message regarding Service Request SR 1-481028911: You are correct that the documentation for the command converters was written a long time ago, when the DSPs included all of the JTAG/OnCE pins. However, the DE and TRST pins have not been included in any of the Symphony DSP563xx devices (DSP56367, DSP56374, etc.). The Freescale debuggers do not require the DE and TRST lines to be connected for proper operation. The operation of these signals is duplicated by JTAG/OnCE software. So all of the DSP tools (debuggers and command converters) that work with the DSP563xx Symphony devices will work with the DSP567xx devices as well. The only difference is that the DSP567xx devices have two cores with the JTAG ports daisy chained.

---

---

 . . . So let's just get to the heart of the issue and answer your questions:

Does CC_RESET/ (pin 9 on JTAG) need to drive RESET/ on the DSP56721?
YES

Does TRST/ (pin 14 on JTAG) need to drive RESET/ on the DSP?
NO

Can we leave CC_RESET/ (pin 9 on JTAG) unconnected, with just a pull-up resistor?
NO

If CC_RESET/ (pin 9 on JTAG) is unconnected, does it even need a pull-up resistor?
You cannot leave CC_RESET unconnected

Can we leave TRST/ (pin 14 on JTAG) unconnected, with just a pull-up resistor?
YES  

If TRST/ (pin 14 on JTAG) is unconnected, does it even need a pull-up resistor?
NO

Can we leave DEZ/ (pin 13 on JTAG) unconnected, with just a pull-up resistor?
YES

If DEZ/ (pin 13 on JTAG) is unconnected, does it even need a pull-up resistor?
NO

The normal JTAG signals TMS, TCK and TDI have internal pull-up resistors. Do they also need external pull-up resistors?
NO

---

Hope that helps.

[Nikko]

Thanks a lot for taking the time to recollect your info on that topic, i think that thanks to your post i'm now ready to put a JTAG port together.

My understanding is that the JTAG port on the Symphony dsps is still compatible with the legacy tools. I still have a couple of home-made parallel-port 'wigglers' from my previous project with the 56321. It's basically the 321 EVM part of the debugging port and consists in a bit 74 logic. It works well but it's picky with the patallel port (typically it will work ok with a desktop computer but won't work with a laptop).

So I'll try to have these working with the new project. I don't mind using Suite56 - it was doing the job with the single core dsps.

Greetings

Nick

[marcelbelmans]

Old topic, I know, but I wanted to add also a succes story here with Symphony Studio and debugging via USB.

I first worked 2 years with the Soundbite, without problems (apart from the fact it doesn't seem to make contact sometimes, but then it works again on a second or sometimes third try). Now I moved to an own design PCB, with the DSP56725 and it works fine. To debug the DSP56725 i just copied the debug-part around the FT2232 from the schematic of the Soundbite.

Here is how I did it:

- carefully copy the debug electronics around FT2232 in your own pcb design. You can omit the I²C and ISP parts and only keep the JTAG part, as I did. Still, I kept the 74244 in there (which was probably not necessary).

- assemble your pcb :smileywink:

- attach the usb-cable. If you used the soundbite before, all the necessary drivers should be installed already. Otherwise you can download them on FTDI website

- use MProg (FTDI website) to program the eeprom 93LC46. You can read the contents of this chip from the soundbite and copy it to the eeprom on your own board. If you have no soundbite I can send you this file (deUNDERSCOREselAThotmail).

- setup the openOCD settings in "external tools" (choose dsp56700 and dongle 'soundbite') and launch openOCD. If it gives errors, there is probably an error in your electronics (I switched 2 pins on the 74244 for example, it cost me a week of total despair before I found this error :smileyangry:).

- to debug, create and launch 2 debug settings (1 for every core) with gdb56300. Carefully follow every step in the 'symphony studio eclipse for symphony DSPs users guide' part 2.2.3 and 2.1.4. There seems to be a special clue in there that I missed the first time. First I tried to get my code running and it didn't work. After carefully following every step in the manual it did. Don't know what I missed the first time.

Success for everyone trying to get his dsp56 alive!

[clangen]

Anyway if someone needs to have a working PCB available that can be used following the steps described by Marcel Belmans I can offer my help. I made a batch of these USB/JTAG interfaces (using the 14-pin Motorola JTAG pinout). It works pretty well for all DSP563xx as well as the dual-core DSP567xx devices. I can offer these JTAG devices for price of 50$ including cables and a Hammond housing (not necessary and unmachined yet). On demand please email to clangen@gmx.de.