more urgent！TMS320F28335: a urgent 28335 ADC problem

Hi Devin,

Very appreciate  for helping me all the time.

Today I have do some test with customer engineer, and found a strange phenomenon ,

We add some code(nothing to do with the ADC Init) in the main function, for example : the yellow back ground is the code we add,

We found that ,when adding the code(not limit to this “if” code, any code is ok), the ADC worked normally, when remove the code, the ADC result are wrong!

It is unbelievable! Unable to explain this ! The only possibility I can think of is  that: when add the code , the location of the InitAdc() function stored in the flash may changed.

Another thing is that, the customer’s V1.02 SW worked normally, the V1.03 SW worked abnormal. These two SW used the same  initialization process and ADC Init code, and the difference between the two version SW is the V1.03 is bigger than the V1.02 .

Can you give me some ideas about this? Thank you!                            

void main(void)

{

Int a = 1;

  InitSysCtrl();                                        

  MemCopy(&RamfuncsLoadStart, &RamfuncsLoadEnd, &RamfuncsRunStart);

  InitFlash();

  DINT;

  InitPieCtrl();

  IER = 0x0000;

  IFR = 0x0000;

  InitPieVectTable();

  EALLOW;

  PieVectTable.ADCINT1        = &Current_Isr;            

  EDIS;

  PieCtrlRegs.PIECTRL.bit.ENPIE = 1;              // Enable the PIE block

  PieCtrlRegs.PIEIER1.bit.INTx1 = 1;              // Enable PIE Group 1,

  InitGpio();      

  InitAdc();                            

If(a==0)

{

}

……..

}

 Hi Devin,

Thanks for your help , we do a lot of test today, maybe we have found the cause of this problem : after the InitADC(), we read the value of the ADC Register ,found that the value of  “ADCTRL1 ADCTRL2  ADCTRL3” are all zero! The InitADC() failed to configure the ADC register!Below is the code they used:

What are the possible reasons that cause the register configure failed?

And another test is as you have mentioned: “If you had completely powered everything up, then shut just the ADC down (not the BG and references) then you can power the ADC up in 20us. ”

We use above code to power the ADC, found that this way can reduce the frequency of the problem. Have you ever encountered similar problems? Do you have any experience or basis for this proposal?

Hi Shuai,

For the power-up time, the suggestion was directly taken from the module documentation:

But based on the datasheet, you should still probably wait at least 1ms in this case and not 20us, as 20us is the best case power-up time.

As far as why your registers aren't getting correctly initialized depending on some adding/removing unrelated code, I'll ask someone else to comment. 

When executing code from flash, the location of the code may have an effect on the performance. There is a prefetching mechanism on the flash. If code is moved around and or branches are added to the code then the performance may vary because the prefectching may behave differently. This may cause some timing issues between the different builds.

Please use the Clock feature in CCS to profile this. Run->Clock->Enable.

sal