• State TI Thinks Resolved
  • Locked Locked
  • Replies 9 replies
  • Answers 3 answers
  • Subscribers 48 subscribers
  • Views 988 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.

DAC8750: DAC8750 CALIBRATION

Bob Shaikhvand
Intellectual 420 points

Part Number: DAC8750

Hi,

What is the easiest way to calibrate DAC8750? Is there an example that we can use to quickly

calibrate this part?

Thanks,

Bob

  • 0
    TI__Expert 5820 points
    Hello Bob,

    Section 8.3.11 in the datasheet describes the calibration process. Basically you will multiply the written code in the DAC data register by the DAC gain calibration register to correct for gain error with 0x8000 (midscale) = 1. The DAC zero calibration register will be added to this to correct for offset.

    Let me know if you have additional questions on this as you have likely already seen this description in the datasheet.

    Thanks,
    Garrett
  • 0 in reply to Garrett Satterfield
    Intellectual 420 points
    Hi Garrett,

    We did the calibration as described in the datasheet but we are running into 2 issues.

    Issue 1 : The maximum output we can get out of the DAC is 19.9mA instead of 20mA regardless of the gain. It looks like the DAC saturates at 19.9mA.

    Issue 2: The gain is not linear across 4mA - 20mA range. It looks like it's accurate at lower range ( 4mA - 10mA). But at middle and higher range the output is less accurate. For example, instead of 15mA we get 14.98mA.

    Please let me know if there is anything else we can do to improve output accuracy.

    Thanks,
    Bob
  • 0 in reply to Bob Shaikhvand
    TI__Mastermind 46290 points
    Bob,

    Can you elaborate on how you have arrived at the codes you're applying to the calibration registers?

    I will be taking over support of this thread. I will also be on business travel this week, so there may be some delays or you may find someone else from my team assigned to reply to this if I am unable to.
  • 0 in reply to Kevin Duke
    Intellectual 420 points
    Hi Kevin,

    We do the following to get the calibration gain and zero registers:

    1) Chose RANGE 4mA – 20 mA, so , when CODE = 0x0000, we should get 4mA, and when CODE = 0xFFFF, the IOUT = 20mA
    2) First, we enable CALEN, and set ZERO Reg = 0, and GAIN reg = 0x8000, and send CODE = 0, this give us the IOUT = 3.97 ~3.98
    3) With this value, we need to add a compensation of about 0.02~0.03 mA to the Zero, the corresponding CODE for 0.03mA for RANT 4-20mA in
    0 – 0xFFFF is 0.03/16.0 * 0xFFFF = 122, or it's derived by: (3.97-4.0)/16 * 65536 = -122, so the Zero Reg is set to 0x7A.

    Bob
  • 0 in reply to Bob Shaikhvand
    TI__Mastermind 46290 points
    Bob,

    Two things to comment on as I seem to have missed one of your original comments.

    The calibration feature implemented in the device is a purely digital feature. Meaning that any limitations in the analog domain from things like input offset voltage of the internal amplifiers or limitations of gain due to internal resistor mismatches or output voltage swing to rail implications cannot be overcome. So in the case of a negative gain error causing the full-scale voltage at 0xFFFF producing an output of ~19.9mA cannot be overcome by the internal digital calibration. Most likely this error is coming from reference error as this goes beyond the signal chain specifications.

    Concerning your calibration methodology, I did not see how you arrived at your gain calibration coefficient to explain the nonlinearity you mentioned. However, I can still try to provide some level of guidance.

    Typically calibration is done from a two point line of best fit - that line of best fit is used for both offset and gain calibration coefficients. Those two points are chosen within the linear region of operation of the system at something like 10% and 90% full-scale range to avoid non-linear behaviors of amplifiers near their rails.
  • 0 in reply to Kevin Duke
    Intellectual 420 points

    Kevin,

    We are using an internal reference. I measure 4.99V on the REFIN voltage. We use a 0.1uF decoupling capacitor near REFIN input.  

    Attached is the schematic of DAC8750. Please let me know what could prevent output to go to 20mA as it's currently max. out around 19.9mA.

    Thanks,

    Bob

  • 0 in reply to Bob Shaikhvand
    TI__Mastermind 46290 points
    Bob,

    The internal reference is specified at room temperature with minimum and maximum tolerances of 4.995V and 5.005V respectively - or 0.1% error. There is a maximum temperature coefficient of 10ppm/C. I'm not sure what you are making your measurements with (i.e. hand-held DMM versus 8.5 digit benchtop DMM), but 4.990V would suggest something is off.

    If that value is correct, then the ideal full-scale output of the DAC in 4-20mA mode would be ~19.96mA. If we also take into consideration the worst case offset and gain errors at room temperature this could be as low as ~19.93mA at full-scale. This does not take into consideration any potential leakage from D29.

    Since the reference appears to display more error than is specified in the datasheet, can you give some more details concerning your measurement methodology for both the voltage reference and the current output?
  • 0 in reply to Kevin Duke
    Intellectual 420 points
    Kevin,

    I used hand held Fluke DMM when I measured 4.99V internal reference voltage. By using a benchtop DMM, I am now measuring 4.9982V for the internal ref. voltage and the full scale current output is 19.98 mA with a 878 Ohm load.

    Thank you for your help.

    Bob
  • 0 in reply to Bob Shaikhvand
    TI__Mastermind 46290 points
    Bob,

    Thanks for the update.

    Those results are much better aligned with the datasheet.

    If you have any further questions, please let me know. This thread will automatically lock after 2 weeks but if you create a new thread myself or someone on my team will follow up with you.