• State
  • Locked Locked
  • Replies 4 replies
  • Subscribers 83 subscribers
  • Views 1270 views
  • Users 0 members are here
Support feedback
Options
Options
Related

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.

TMS320F28335 shorted, destroyed, during programming

Steve Anderson2
Prodigy 40 points

Other Parts Discussed in Thread: TMS320F28335

I just destroyed a TMS320F28335 for the third time. The 3.3V rail becomes shorted to ground through the DSP. The heat builds up very quickly, concentrated in the center of the IC package. I am using an eZdsp from Spectrum. For the first two times, it happened when I inserted the eZdsp board into the expansion board that I designed, so I thought my expansion board was the problem, either a short or overvoltage.

This third time I destroyed a DSP, it was not inserted into the expansion board or anything else. Here's what I noticed: when using CCS 6 and clicking the debug button to flash the code, etc, all of the pins go high. I was probing around with my meter, measuring voltage, current, resistance. Most of the pins remained at 3.3V even though I had shorted the pin to ground through the ammeter. I touched one of the pins in the GPIO71 to GPIO79 range and then the DSP instantly got destroyed. In the expansion board, those pins are outputs connected to bases of a darlington array, ULN2003. There is current limiting resistance inside the ULN2003. The darlingtons are relay drivers. Thus, I noticed that the relay clicked on whenever I entered debug mode in CCS 6 (I only have 1 relay installed on my board at the moment). This is bad because I'm designing a power application and I don't want relays unexpectedly turning on. Most of the pins seemed to be high-Z, but obviously not all of the pins because the relay clicked on.

I am almost 100% sure that there isn't a short to ground on any of the DSP's pins (on my expansion board).

I really don't want to remove the eZdsp from my expansion board every time I enter debug mode to flash new code. I also really don't want to buy a bunch more $40 DSP's when I still don't know exactly why they're getting destroyed. Can someone please shed some light on this? Thanks very much, Steve

[Steps Needed to Recreate Problem: 

get an eZdsp 28335, enter debug mode in CCS 6, probe the expansion header pins to a meter in voltage, current and resistance mode.  I don't remember which mode broke it.
  • 0
    TI__Guru 60865 points

    Hi Steve,

    "Most of the pins remained at 3.3V even though I had shorted the pin to ground through the ammeter."

    I suspect that this is causing the issue.  It is probably better to measure the current between the expansion header and the input to the darlington array, and not between the expansion header and ground, since if the pin is not high-z (which is what you suspect is happening) the meter will create a short to ground.

     

  • 0 in reply to Devin Cottier
    Prodigy 40 points

    Thank you for the reply.

    of course everyone's thinking duh, you shorted the pin to ground and the pins are 4mA max.  but does someone know what actually happened? 

    first of all I haven't found anywhere in any datasheet a warning saying that all the pins go high when in debug mode before you press the green arrow to start executing code.  this fact that the pins go high necessitates a very specific requirement for external circuitry which I haven't seen anywhere.  can someone elaborate on this?

    what if I'm using a pin as a GPIO input and I tie the pin to the output of a logic gate or some circuit that is normally outputting zero volts?  this creates the same condition as me shorting the pin with my ammeter.

    what's the deal, are all the pins high-Z in debug mode (before clicking on the green arrow) or not?  is debug mode known to destroy chips?


    Thanks sincerely,

    Steve

  • 0 in reply to Steve Anderson2
    Prodigy 40 points

    also, by the way, when I had inserted the eZdsp board into my expansion board and destroyed the first 2 DSP's, I didn't short the pin.  so I was wondering maybe the damage was caused by some overvoltage noise spike due to the relay switching.  however, I didn't see any overvoltage on my scope.

    I will keep testing and try to find more specific information...

    question:  this fault condition of the shorted 3.3V rail; can it be caused by overvoltage or short circuit, both? or anything else?

    question:  which document describes the condition where all of the pins go high in CCS 6 debug mode before I click the green button?

    question:  does it say in any document that you cannot short any pins to ground?  I did read in multiple places that at reset, the DSP makes all of the pins GPIO inputs.  so they should be fine being dead shorted to ground.

    Thanks very much,

    Steve

  • 0 in reply to Steve Anderson2
    TI__Expert 5950 points

    Steve,

    Sorry for the delay in reply.  I think you've got two questions.

    1) Debug if GPIO pins are going high when the debugger connection is made.  This should not be happening, so needs some follow up for sure.

    2) Is it okay to short a GPIO output directly to ground.  This should not be done and can damage the device.  The output buffers are only rated to 20mA maximum current before damage can occur on the pins.  If you want to pull these low and measure current for debug reasons you will need a series resistor to limit the current, say 1kOhm.  For the case of a GPIO being active output and driving against another device on the board this will also potentially limit the current since the other device will have some finite output impedance.  It may not be the best thing to do, but it would be better than a direct 0-Ohm short to VSS.

    Best regards,

    Jason