• State TI Thinks Resolved
  • Locked Locked
  • Replies 5 replies
  • Answers 2 answers
  • Subscribers 98 subscribers
  • Views 481 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.

TMS320VC5509A HPI can't run stable on -20 degree low temperature

jianjun deng
Prodigy 30 points
Other Parts Discussed in Thread: TMS320VC5509A

Application: TMS320VC5509A works as a slave proccessor accessed by HPI from a ARM host in none-multiplexed mode.

The DSP use a 20MHz outside cristal,work on 200MHz freq. by inner PLL.

Problem:The system can work stablely on normal room temperature and on high temperature about 60 degree.

On low temperature(about -20 degree),the system can start normally,BUT can't run stable.some board crash in minutes,some crash in hours,there's only one board never crash.

I have a test to stop the HPI access from the ARM host,let the slave DSP  run independly,the result is: Host and DSP both can run stable in -25 degree for a long time (about 16 hours test).

 

My design used a 1.6V LDO power chip to supply the DSP core according to the datasheet. I have a try to low down the core supply voltage to 1.5V,the system stability improved a lot,some board become never crash,some board can run longer time(from minutes to hours).

At last,I low down the core supply voltage to about 1.45V,and test all board for about 20 hours on -20 degree temperature,and all boards seem run OK.

I don't understand the theory and afraid of any hidden problem.

I am waiting for a answer urgently.

Thanks.

  • 0
    TI__Mastermind 23785 points

    Well, really there is not enough data to determine the cause of the problem. However, let's start for here.

    First, is this a brand new design? Is this design in production? What stage is this design?

    Based on the provided data, assuming your system all runs on the same core voltage and I/O voltage, plus all components are graded to operate for that temperature, thus, when the temperature getting lower, the ICs run slower. Then when core voltage downs, the I/O interface timing will be slow down.

    I would suggest scoping the interface signals to confirm no timing violations.

    Regards.

  • 0 in reply to Steve Tsang
    Prodigy 30 points

    Steve,

    Thanks for your rapid response.

    The bug is located! I stopped use the DSP inner OSC and replaced the 20M cristal with a external 20M OSC chip feed to X2 pin. All things go right with 1.6V core power supply.

    The inner OSC circuit is not stable in low temperature. I scoped the core power,maybe the ripple is too high,I guess  the inner OSC is powered by  the core voltage,sensible to it.

    I want to know which power pin the inner OSC and main DPLL is powered by,and then try to lower the ripple and test again. 

    Would you help?

    Regards.

  • 0 in reply to jianjun deng
    TI__Mastermind 23785 points

    Great to be able to help locating the cause.

    What stage is this design? Is this design in production or a brand new design?

    Regards.

  • 0 in reply to Steve Tsang
    Prodigy 30 points

    Steve,

    The design is in production and previous customers all used in room environment without any problem reported.

    Recently,a new customer needs this device to work in ultra low temperature,and this issue rised.

    By the way,I replaced all components in the board to industrial grade and double checked it,and there's no any help. 

    Regards.

  • 0 in reply to jianjun deng
    TI__Mastermind 23785 points

    Jianjun,

    This design has been in production for many years. I cannot find the designer anymore.

    According to the data manual, Both DPLL and OSC should be on CVDD domain.

    Regards.