• State TI Thinks Resolved
  • Locked Locked
  • Replies 16 replies
  • Answers 1 answer
  • Subscribers 83 subscribers
  • Views 516 views
  • Users 0 members are here
Support feedback
Options
Options
Similar topics

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.

TMS320F28379D: TMS320F28379D

vadim piskun
Expert 2230 points
Part Number: TMS320F28379D

Hi,

I am using 16 bit diff ADC  as in Launchpad. If nothing is connected to the input of difference amplifier I have this ADC reading

The noise fluctuates 40*3/2^16 = 1.8 mV. From many application notes the thermal  noise is is much smaller than this range. From where is this 1.8 mV coming from?

thanks vadim

  • 0
    TI__Prodigy 470 points

    Hi,

    The expert is out of the office because of Holiday. Please expect a reply by Tuesday.

    Thanks.

  • 0
    TI__Mastermind 30185 points

    Hello Vadim,

    Have you seen this noise when applying an actual voltage or signal? We do not spec our ADCs for when they are not connected to an input as outlined in the datasheet. The noise that you're seeing can be caused for any number of reasons such as environmental noise.

    Best regards,

    Omer Amir

  • 0 in reply to Omer Amir
    Expert 2230 points

    Hi Omer,

    Sorry I missed your response. This noise is on the Launchpad. ADCIND is set for differential 16 bit mode. And the output from the difference amplifier is connected to ADC, so ADC is biased. If you have this board you can check it.

    https://www.ti.com/tool/LAUNCHXL-F28379D#tech-docs

    thanks vadim

  • 0 in reply to vadim piskun
    TI__Mastermind 30185 points

    Hello Vadim,

    Have you tried applying an actual known voltage and reading that to see any noise? We don't spec what the ADC does when no voltage is applied to it. Even if I did check this on my side, I might not the issue you're getting depending on your setup, so this is something you need to do on your own setup to verify that the data you get is within the limits specified in the datasheet.

    Best regards,

    Omer Amir

  • 0 in reply to Omer Amir
    Expert 2230 points

    Hi Omer,

    ADCIND4 and ADCIN5 are both around 1.52V. It is well set bias voltage from the difference amplifier. Inputs of this amplifier are terminated by 50 Ohms , so there is no inputs hanging in the air.

    thanks Vadim

  • 0 in reply to vadim piskun
    TI__Mastermind 30185 points

    Hello Vadim,

    I will look into our specification on the ADC regarding DC voltages and get back to you tomorrow.

    Best regards,

    Omer Amir

  • 0 in reply to vadim piskun
    TI__Mastermind 30185 points

    Hello Vadim,

    If you are inputting two signals of the same voltage into the both inputs for an ADC configured in differential mode, then the results you are getting will not be following the specifications we have listed in the F2837xD datasheet. Looking at section 8.10.1.1.1 (Signal Mode, page 101) of the device datasheet, you can see that the positive and negative inputs for the differential ADC must be the same distance from the common mode voltage to be properly calculated.

    This common mode voltage should be VREFHI/2, with a 50 mV margin of error. As such, if you are trying to measure 1.52 V using the differential mode of the ADC, you need the voltage supplied to the positive input and negative input to be 0.76 V above/below VREFHI/2, where the positive and negative inputs are opposite each other along this line. You can look at the datasheet for more details.

    Best regards,

    Omer Amir

  • 0 in reply to Omer Amir
    Expert 2230 points

    Hi Omer,

    Does TDK USA has an account manager or dedicated application resources?

    Thanks vadim

  • 0 in reply to vadim piskun
    TI__Mastermind 30185 points

    Hello Vadim,

    I will send you a friend request for this info, so as to not post it publicly.

    Best regards,

    Omer Amir

  • 0 in reply to vadim piskun
    TI__Mastermind 30185 points

    Hello Vadim,

    Do you still have a question or concern here, or have you been able to solve the problem you observed?

    Best regards,

    Omer Amir

  • 0 in reply to Omer Amir
    Expert 2230 points

    Hi Omer,

    i am working with dedicated apps team.

    thanks vadim

  • 0 in reply to vadim piskun
    TI__Mastermind 30185 points

    Yes, and they have contacted me since I'm the expert for this particular topic. I'm not sure what the issue is, if you need to share confidential info let me know so we can handle it.

  • 0 in reply to Omer Amir
    Expert 2230 points

    Hi Omer,

    Could you please load the firmware DiffADC_noise, I sent to your colleagues, onto launchpad? It would be much more efficient to move forward if you see the problem and measure it. Should not take much time.

    thanks vadim

  • 0 in reply to vadim piskun
    TI__Mastermind 30185 points

    Hello Vadim,

    I will email you to get the info.

    Best regards,

    Omer Amir

  • 0 in reply to Omer Amir
    TI__Mastermind 30185 points

    This issue was handled offline. The 16-bit differential ADC was tested on a F2937xD LaunchPAD with driven and undriven inputs for channels D4 and D5 on the LaunchPAD, where the noise was being observed (these two channels go through a difference amplifier which also acts to buffer the inputs; this is ONLY on channels D4 and D5 for the LaunchPAD, other ADC modules/channels have no such buffer). Testing on the LaunchPAD, the DC code spread for the differential ADC was about 21 LSB. Although this is higher than some of the testing we have done before, this is not done with the most optimum setup (i.e. the device is on a board designed for analog measurements, following the guidelines laid out in <SPRACZ9A>). In the case where there is still noise, the settling time for the ADC can be increased to allow more stable readings (although the sampling rate will need to be lower).

    It is also notable that because the differential 16-bit ADC is higher resolution than the single-ended 12-bit ADC mode, it will be more sensitive to noise since there is a larger digital range representing the same voltage range on the input channel. The 16-bit mode allows a 16x increase in precision than the 12-bit mode, and as such the DC code spread will see more experience from noise. For a 3V reference, 1 code on the 12-bit ADC represents about 730 µV, whereas 1 code on the 16-bit ADC represents 45 µV. Thus, 700uV of noise may not be observed using a 12-bit ADC with 3V VREF but the same noise could impact the DC code spread of a 16-bit ADC by 16 LSBs

    Generally, the best way to reduce noise on ADC channels is to isolate the different types of signals (analog, digital, power, etc.), buffer the inputs to the ADC, use an accurate VREF (such as a high precision external VFREF), and having a clean power supply to the device. The LaunchPAD is limited in many of these aspects because it is an out-of-the box device that is used for quick development.