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.

ADC and DAC accuracy of AMC7812

Other Parts Discussed in Thread: AMC7812

Dear TI Team,

We are using AMC7812 in one of our design and I need to know the accuracy of ADC as well as DAC of the IC.

% accuracy values are not mentioned in the datasheet. Please share the document if you have any regarding this.

Our requirement is ADC 6mV and DAC +/- 3mV.

Awaiting your response.

With Regards,

Bhanuprakash Nayak

  • Howdy Bhanuprakash,

    Accuracy is dependent on the total amount of error within a system.  The datasheet does include all the parameters required to successfully calculate the total unadjusted error of a system.  Some datasheets also include this information, which I will get to very soon. 


    To recap, a statistical analysis technique referred to as RSS, root sum squared, is used to sum uncorrelated error sources for error analysis.  These error sources include linearity error (INL), offset, and gain error.  These error sources are added using a root sum squared method.  The end result is referred to as total unadjusted error, TUE.

    TUE = sqrt( inlError^2 + gainError^2 + offsetError^2)

    More information can be found in the following blog post:

    e2e.ti.com/.../dac-essentials-how-accurate-is-your-dac

    The TUE of the DAC within the AMC7812 device is listed as having a maximum TUE value of +/-10mV for a 5V Range.

    You can use the formula above to calculate the TUE of the ADC.

    To convert LSBs to Volts use the following equation:

    (LSB/2^bits)*(Vref)

    Again this information is also posted in the above blog post.


    Best Regards,

    Matt

  • Hi Matt,

    Thanks for your reply.

    Can you suggest any part with TUE within 6mV for ADC and +/- 3mV DAC .

    Please let me know Multichannel as well as individual IC's with the above requirement.

    With Regards,
    Bhanuprakash Nayak
  • Howdy Bhanuprakash,


    The lowest TUE in the AMC7XXX product line is the AMC7812 device with +/-10mV TUE for the DACs and +/- 9mV TUE for unipolar ADC inputs.  There may be other non-integrated devices that display better performance in terms of accuracy, and can be found by using the parametric search tool at:

    http://www.ti.com/lsds/ti/data-converters/precision-dac-less-10msps-products.page

    This will allow you to pick an individual DAC as required by your design guidelines.  This information includes desired parameters such as INL(Max), Gain Error, and offset.  Additional parameters include DAC range, resolution, settling time, etc.

    It is also important to note that DAC errors resulting from offset, and gain can be calibrated out of the system by updating your algorithm in code.  INL is the main error source if gain and offset error are calibrated out.  For example, if you calibrate gain/offset error out of the AMC7812 device that would leave you with only +/- 1LSB INL error for the DAC and ADC, which equates to 1.2mV of error for the DAC output and ADC input (AVDD=5V).

    If choosing individual devices, I would encourage you to make a separate post querying for a precision ADC device.  Again, make sure you specify the resolution, range, and accuracy wanted. 

    Best Regards,

    Matt

  • Dear Matt

    Thanks for your quick response. Please provide some details regarding calibration process and supporting documents if any.

    I have seen ADC/ DAC having zero and Gain registers to which we can write a value which will eliminate the zero error and gain errors.

    If I can eliminate both these errors then I can use this particular ADC/DAC for my application as the error is very less.

    With Regards,

    Bhanuprakash Nayak

  • Software calibration does involve some lengthy testing with the end DAC device, as it involves measuring the performance of each DAC channel and creating an equation that will compensate and nullify the effects of gain error and offset for each channel.

    As an example, high precision measuring equipment is used to obtain the complete DAC transfer function.  A good rule of thumb is for the measuring equipment to have 2 bits of greater resolution than the device under test.  A 2pt or best-fit approach can be used in obtaining the gain and offset term.


    Ideally, a DAC transfer function would follow a y=gain_ideal *code curve, but because of the offset and gain error term this produces a measured transfer function of y=gain_measured * code + offset.


    Once offset and gain error are determined from the measured response of the DAC channel the following equation is used to calibrate:

    Digital_Code(calibrated)= (code)*(gain_ideal/gain_measured) - offset[LSB]

    As an example, consider gain_ideal=1, gain_measured=1.1 and offset =-20mV for a 5V span, 12 bit DAC.

    Digital_Code = code*(1/1.1)-(-20mV*2^12/5) = code*(1/1.1)+16.384LSB

    As you mentioned, there are other devices that employ calibration registers to help reduce accuracy error.  The registers are set within code, reducing the amount of software for error correction.


    Best Regards,

    Matt

  • Thanks Matt,

    Can you provide the PCB guidelines for AMC7812, Please share the documents if you have any.

    I did not find AMC7812 part in ADC section for WEBBENCH simulation. Please suggest as I want to simulate my VREF voltage as we are using 1.35V.

    With Regards,

    Bhanuprakash

  • Howdy Bhanuprakash,

    There currently is no model for the ADC section of the AMC7812 device.  Regarding the PCB layout, you can visit the following link which is the user's guide for the AMC7812EVM.  Page 22-25 displays the EVM's schematic and PCB, which can be used for reference.

    EVM UG link: http://www.ti.com/lit/ug/sbau177d/sbau177d.pdf

    Best Regards,

    Matt