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.

TMS320F28069: ADC Noise

Part Number: TMS320F28069

We have boards that need to read a bunch of analog signals. In our system, we need the signal reading to be as noiseless as possible. We achieved lower than 20 ADC counts peak to peak for 1.65V signal. But somehow 6 out of 32 boards, have noise more than 30 counts. We have tried to check everything in our system, signal integrity, supply quality that may affect the reading. But there is nothing wrong. In the end, after changing TMS chip, the noise drop to an acceptable value. So the problem lays in a couple of TMS chip ADC. There is no noise in the signal. But after TMS read it, the resulting reading have noise. So is this something like obsolete chip? or are there anything that i need to account for or not knowing? or is it still withing TMS ADC % error? For your information, we usually buy it from mouser.

above is the signal with noise (1.65 V but with offset to 0). All value below 10 count is the signal.

  • Angga Wardana,

    I would not expect to see noise spikes like that.  Can you describe the nature of the 1.65V input signal and its conditioning?  For example, is it an on-board power supply?  Is it buffered or filtered?

    Have you tried different ACQPS values?

    Can you review this Errata document to make sure that all known ADC issues are handled in your application?

    -Tommy

  • Hi Tom,

    The ADC is used to read collision sensor which used photo transistor. When doing functional testing, we switch the sensor with voltage divider circuit using 2 10K resistors from 3.3V. So the output is 1.65V. 

    The signal is buffered and filtered as shown above.

    I am not in charge of the programming. But we are in beta version already.  I do not think there is a problem with our board design and firmware. ACQPS values should not be the problem because most of the board with the exact same firmware works fine. I have tried using the chip with noise spike to the board which chip had no noise spike. The noise spike chip still give bad reading. So i am pretty sure the problem is with some of the chips. We still seldom encountered it. But i am hoping to find solution for this before RTM. 

    Thank you

  • Hi Tom,

    The ADC is used to read collision sensor which used photo transistor. When doing functional testing, we switch the sensor with voltage divider circuit using 2 10K resistors from 3.3V. So the output is 1.65V. 

    The signal is buffered and filtered as shown above.

    I am not in charge of the programming. But we are in beta version already.  I do not think there is a problem with our board design and firmware. ACQPS values should not be the problem because most of the board with the exact same firmware works fine. I have tried using the chip with noise spike to the board which chip had no noise spike. The noise spike chip still give bad reading. So i am pretty sure the problem is with some of the chips. We still seldom encountered it. But i am hoping to find solution for this before RTM. 

    Thank you

  • Angga Wardana,

    I would advise caution with concluding that the existing system is optimized based on a small sample set of boards. Manufacturing tolerances in the fab are expected to cause normal device behavior to drift between production runs, and these variations can hide or amplify some of the known issues described in the Errata. All devices sold by TI can be expected to meet the datasheet specifications, but there will be differences between units.

    The buffering looks reasonable to me, but the op-amp bandwidth may be marginal if using an ACQPS register setting of 6. My first instinct is to confirm that the first-sample issue is handled in the firmware. My second instinct would be to experiment with the ACQPS values and/or the C74 value to see if the noise improves. For a buffered static signal like this, I would expect the standard deviation of the codespread to be well under 2LSBs.

    -Tommy

  • Angga Wardana,

    It has been a while since your last update.  I assume that you were able to resolve your issue.

    If not, please reply to this thread.  If the thread has locked due to timeout, please create a new thread describing the current status of your issue.

    -Tommy