• State
  • Locked Locked
  • Replies 11 replies
  • Subscribers 82 subscribers
  • Views 2001 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.

CCS/TMS320F28335: TMS320F28335 ADC Result problem

jungun lee
Prodigy 40 points

Part Number: TMS320F28335

Tool/software: Code Composer Studio

I use TMS320F28335PTPS CA-3CAEPGW G4.
I have some problem with this item.
It worked good from test to delivery.
After my customer did test some times, It makes some unstable ADC result.
ADC result is moving about 20 at peak to peak when ustable.
ADC result is moving about 1 at peak to peak when normal.
Input is temperature sensor value.

This problem do not occur all times.
But very repetition at power on time after power off more than 20 minutes.
So I tested ADC input pin.
But input signal was always stable at ADC input pin.
I could find some information that is ADC inaccuracy at low frequencies from silicon errata(rev k).

But I set up that ADCCLK operate at 1.25MHz.
It is not same to errata.

So, I have some question.
1. Is it possible the ADC inaccuracy at low frequencies occur sometimes or by aging?

2. Is my problem same problem of ADC inaccuracy at low frequencies?

3. If yes, is it possible to solve this problem by rise of ADCCLK frequencies?
    Because I already use more than 1Mhz.

4. I didn't see this problem at errata at design period.
   When did you add ADC inaccuracy at low frequencies to errata?

5. Could  you provide a errata when was just added  ADC inaccuracy at low frequencies?

Following is my setting.
XCLKIN = 30MHz
HSPCP = 1
HSPCLK = 75 MHz

void Init_ADC(void)
{
 AdcRegs.ADCREFSEL.bit.REF_SEL = 0U;
 AdcRegs.ADCTRL3.all = 0x00E0U;
 AdcRegs.ADCTRL3.bit.ADCBGRFDN = 0x03U;
 AdcRegs.ADCTRL3.bit.ADCPWDN = 1U;
 AdcRegs.ADCTRL3.bit.ADCCLKPS = 15U;
 AdcRegs.ADCTRL1.bit.CPS = 1U;  // ADCCLK = 1.25MHz
 AdcRegs.ADCTRL1.bit.ACQ_PS = 15U;
 AdcRegs.ADCTRL2.bit.EPWM_SOCA_SEQ1 = 1U;
 AdcRegs.ADCTRL2.bit.INT_ENA_SEQ1 = 1U;
 AdcRegs.ADCTRL3.bit.SMODE_SEL = 0U;
 AdcRegs.ADCTRL1.bit.SEQ_CASC = 1U;
        AdcRegs.ADCMAXCONV.bit.MAX_CONV1 = 0x5U;
       AdcRegs.ADCCHSELSEQ1.bit.CONV00 = 0U;
       AdcRegs.ADCCHSELSEQ1.bit.CONV01 = 1U;
       AdcRegs.ADCCHSELSEQ1.bit.CONV02 = 2U;
       AdcRegs.ADCCHSELSEQ1.bit.CONV03 = 8U;
       AdcRegs.ADCCHSELSEQ2.bit.CONV04 = 9U;
       AdcRegs.ADCCHSELSEQ2.bit.CONV05 = 10U;
}

  • 0
    TI__Guru 62975 points
    Jungun,

    The ADC Inaccuracy at Low Frequencies advisory was introduced with Rev H in February 2015.

    Is the customer ADC clock at 1.25MHz? That should be fast enough to avoid the Low Frequencies advisory.

    I would expect the low frequency issue to affect all ADC channels so if you only see fluctuations on one channel, the ADC is most likely working fine.

    Are the readings only unstable when the temperature sensor is recently powered on? Is there a power-on time requirement for the temperature sensor?

    Is the temperature sensor signal buffered? Is there an input capacitor on the pin?

    -Tommy
  • 0 in reply to tlee
    Prodigy 40 points
    I Checked another pin. Therer are same problem.
    Customer ADC clock is 1.25MHz. They don't change it.
    Temperature sensor has no power-on time requirement.
    Temperature sensor signal is buffered and no capacitor on the pin.
  • 0 in reply to jungun lee
    TI__Guru 62975 points

    Jungun,

    Can you check a few more things:

    1. You mentioned that the errant conversions occur soon after power up.  How long does this errant period typically last?  Is the ADC allowed 5ms to power up before the first conversion?
    2. Is this behavior seen on multiple devices?
    3. Would it be possible to try a higher frequency to see if the condition improves?  25MHz would be ideal, but I think 5MHz or 10MHz would be enough.
    4. Do you have the device symbolization with the manufacturing codes?

    -Tommy

  • 0 in reply to tlee
    TI__Guru 62975 points
    Jungun,

    Would the customer be able to try a longer ADC power up time? Maybe 20ms instead of 5ms?

    -Tommy
  • 0 in reply to tlee
    Prodigy 40 points
    1.You mentioned that the errant conversions occur soon after power up. How long does this errant period typically last? Is the ADC allowed 5ms to power up before the first conversion?
    -> If it occur, it is continued until power off.
    2.Is this behavior seen on multiple devices?
    -> No. Only one device is until today. We delivered more than 80 set. One set has one device.
    3.Would it be possible to try a higher frequency to see if the condition improves? 25MHz would be ideal, but I think 5MHz or 10MHz would be enough.
    -> I tried 6MHz once. At that time it is good but I can't belive this result, because this device was good more than six month.
    So, I asked No 1. question at first post.
    And I can not test enough times, because my team is very busy for another project.
    4.Do you have the device symbolization with the manufacturing codes?
    -> If you mean device marking, I mentioned it first line at first post

    Would the customer be able to try a longer ADC power up time? Maybe 20ms instead of 5ms?
    -> My source code don't has time delay setting for power up.
    -> ADC immediately operates after device initializing.
  • 0 in reply to jungun lee
    TI__Guru 62975 points

    Jungun,

    I appreciate the additional background information.  I have a better understanding of the situation now.

    I expect some inaccuracy in ADC conversions while the circuitry is powering up, but I would also expect it to stabilize after the recommended delay time so this may not be the cause.  However, I definitely recommend adding the ADC power up delay as a best practice.

    With respect to aging, the ADC should not degrade like this.  It is expected to have a power-on operating life of multiple years.

    It looks like the device was manufactured before the ADC low frequency issue was identified so it would not have been screened for 1MHz ADC operation in the factory.  Also, with the problem occurring on 1/80 devices, I cannot rule out the possibility of a random manufacturing defect.

    Do you have access to a local TI sales office?  They should be able to help determine if this unit should be sent back to TI for further evaluation.

    -Tommy

  • 0 in reply to tlee
    Prodigy 40 points
    I tested with 20ms time delay.
    This case does not effect.
    I tested about 10 times with 6.25Mhz ADCCLK clock.
    This case has effect.
    In my opinion, I can fix this problem by changing the ADCCLK.
    And this problem may be same problem the ADC Inaccuracy at Low Frequencies in errata.
    But I can not find the root cause of this problem and cause of happening with aging.
    You mentioned TI sales office.
    Is it TI Korea or the company which is the seller this chip to us?
  • 0 in reply to jungun lee
    TI__Guru 62975 points

    Jungun,

    Thanks for trying the experiments.  It does sound like it could be a frequency related defect.

    Both TI Korea and authorized distributors should be able to assist you with returns.  

    -Tommy

  • 0 in reply to tlee
    Prodigy 10 points
    Hello, My name is Sang Hyun, and I am a co-worker with Jungun. We contacted TI Korea and authorized distributors, but we were both told that they can not support us. So, we will contact you if we can send the product directly. If possible, I would like to know the address and contact person.
  • 0 in reply to SANG HYUN LEE
    TI__Guru 62975 points
    Sang Hyun,

    I am sorry to hear that you were not helped. Please allow us a few days to contact our co-workers in Korea on your behalf.

    -Tommy
  • 0 in reply to tlee
    TI__Guru 62975 points
    Sang Hyun,

    I sent you a friend request with local contact information.

    -Tommy