• State TI Thinks Resolved
  • Locked Locked
  • Replies 13 replies
  • Answers 3 answers
  • Subscribers 46 subscribers
  • Views 524 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.

DAC7760EVM: DAC7760 not outputting 10V - 1LSB when control register is set to 0 - 10V

Ola Oni
Prodigy 30 points
Part Number: DAC7760EVM
Other Parts Discussed in Thread: DAC7760,

Hello,

I am currently using the DAC7760EVM as a sanity check for our propriety board. I am trying to get the DAC7760 to output 0 to 10V. When I enter the maximum code I get 9.986V. I was expecting that the output should be approximately 1 LSB less of the 5V reference voltage. What have I missed here? I have the following register settings 

Many thanks in advance.

  • 0
    TI__Guru* 93975 points

    Ola,

    What you are seeing is likely some sort of gain error. If you go back to the original equations defining the output voltage of the DAC you have:

     VOUT = VREF * GAIN *CODE / 2^N

    As an example if the VREF is lower than expected then the output voltage will also be lower than expected. In a gain of one, if the DAC is ideal and the VREF is 4.99V instead of 5V, then the maximum output will still be 4.99V. Additionally, the DAC itself also has a certain amount of gain error that can give additional error. This error is shown as a percentage of the signal.

    In the datasheet, the gain error is listed as about 0.04%, and based on the numbers you've reported, the error that you see is about -0.14% which is certainly higher. There may be an additional offset error, but it is generally smaller than the gain error when seen at full scale.

    I would first check the value of the reference voltage to see how much error the reference is giving you, and then check the offset error (you can measure it when the DAC is set to 0V. 

    Joseph Wu

  • 0 in reply to Joseph Wu
    TI__Guru* 93975 points

    Ola,

    One other thing to consider is the multimeter that you are using. If you are using a handheld meter, the accuracy might be in the 0.1% to 0.2% range, which could be the source of the entire error. If you're using something like a Keysight/Agilent 34401A multimeter, then the basic DC accuracy is about 0.0015%.

    Joseph Wu

  • 0 in reply to Joseph Wu
    Prodigy 30 points

    Hi Joseph,

    Once again thanks for your reply. On reading your message I checked the VREF and measured 4.9961V instead of the 5V which I was assuming.

    I am measuring the voltage with an Owon XDM 1041 bench multimeter with a listed specification of +/-0.05% accuracy. 

    https://www.owon.com.hk/products_owon_4_1%7C2_digits_xdm1041_bench-type_digital_multimeter

    Regards

    Ola

  • 0 in reply to Ola Oni
    TI__Guru* 93975 points

    Ola,


    I would guess that the DAC output is correct, but it's a bit hard to tell without a more precise measurement. If the output is 9.986 with the full scale output code, and the reference is 4.9961, then the equivalent gain error would be 0.062%. With the DAC datasheet spec at 0.04% and the multimeter spec at 0.05%, this is still within some expected error.

    I would note that the Owon multimeter specification is 0.05%+5LSB of the range. The extra 5LSB of error could also change the gain error calculation by a large part of the measurement (especially with 9.986V measurement).

    I did run my own quick measurement of the EVM output voltage and reference and got a gain error of about 0.023%.


    Joseph Wu

  • 0 in reply to Joseph Wu
    Prodigy 30 points

    Joseph,

    When I turn off auto ranging on my Owon XDM 1041 and Tektronix DMM914  multimeter both read 4.995V on the VREF and for 0x0000 code I measure 0.000V on both meters and when at max 0xFFF0 I measure 9.988V on both multimeter.

    For our application I only need a 0.1% accuracy but I do want the output to be able to range between 0V to 10V and between 4mA to 20mA which I am yet to play with.

    I am thinking of using the Over Range bit in Control register to achieve it, what are your thoughts? I am planning to turn on this bit if I detect that the user code equals to 0xFFF0, I set the bit in the control register but then I need to figure out how to deduct the difference between the over voltage and desired 10V.

  • 0 in reply to Ola Oni
    TI__Guru* 93975 points

    Ona,

    I'm sorry I didn't get to this last week, but I'll still need a bit of time to get you an answer. I'll get back to you in another day.

    Joseph Wu

  • 0 in reply to Joseph Wu
    Prodigy 30 points

    Hi Joseph,

    Thanks in advance.

    Regards

    Ola

  • 0 in reply to Ola Oni
    TI__Guru* 93975 points

    Ola,


    I think the Over Range bit may work for you it still may be difficult to use depending on how you use it. Here's how I tested it.

    I turned on the DAC and set the output value to the maximum setting of 0xFFFF (0xFFF0 is sufficient because of the 12-bit DAC). This gave me an output of 9.9903V. After I turned on the over range bit, I got an output of 10.9921V. You should note that this is a bit higher than the expected 10%, or you would have gotten 10.9893V. Regardless, if you do use the over range bit in this way, you would need to calibrate to the over range value.

    After that, you can adjust the gain calibration to adjust for this 10% over-range. I was able to set the gain calibration code to 0x68E0, which makes an equivalent gain of about 0.9097 to reduce the over range. This gives me an output of 9.9990V and this gives me an equivalent DAC gain error of about 0.01%.

    I would note that setting the gain calibration code might be more limited than stated in the datasheet. It looks like the gain calibration is limited to 12 bits. I did try to set the value to 0x68E8, but none of the 4 LSB were retained in the device. I need to check to see if this is really a 12-bit register or if this is a problem with the GUI software. Regardless, I would count on the fact that the gain calibration code is limited to corrections of about 0.027%, so you could theoretically be within half of that value for any given calibration.


    Joseph Wu

  • 0 in reply to Joseph Wu
    Prodigy 30 points

    Joseph,

    Thanks once again for your reply.

    Bizarrely I have been having issues with the Owon XDM 1041 bench multimeter as it keeps bouncing around with its measurement and sadly I only bought this about 2 weeks ago. So I had to fall back on my old trusty Tektronix hand held multimeter that I have used for 20 years. I have also requested the company to supply me a calibrated meter for more accurate readings

    I tried to mimic the result that you showed in your message so when I set output to maximum value 0xFFF0 without over range turned on. I get an output value of 9.988V so this is with gain calibration register set to its default value of 0x0000 which is 0.5%. My VREF is measuring 4.995V.

    When I enable the over range bit, I get 10.990V and when I enable calibration and set the gain calibration register to 0x68E0 I get 9.996V.

    Is there a deterministic way in which you were able to arrived at the 0x68E0 or was it trial and error as I would need to find an algorithm that would allow us to do auto calibration for each board on the production line.

    Regards

    Ola

  • 0 in reply to Ola Oni
    TI__Guru* 93975 points

    Ola,


    You can certainly deterministically figure out the value of the gain calibration code, however, you must be able to measure the output of the DAC at full scale, when the over range bit is set high. 

    Using my measurements as an example. I got an output of 10.9921V from the output with the full scale setting and with the over range bit high. Using this value, I can set output maximum to be 10.000 V using the calibration code.

    Going back to page 34 of the datasheet, the output value is set by (click on the image to make it clearer):

    Ignoring the zero register, need the output of 10.9921V to go to 10.000V. This means the multiplier must be 0.9097. Calculating back from 0.9097, the User_GAIN value must be 26850h. In 16-bit hex, this comes to a value of 0x6E82. However, if the Gain Calibration code is only 12 bits, then this rounds to 0x6E80.


    Joseph Wu

  • 0 in reply to Joseph Wu
    Prodigy 30 points

    Joseph,

    Thanks and sorry for the late acknowledgment. I have been working on another part of the project while I wait for a calibrated meter to be delivered by Thursday. As soon as I get this I would feedback to you. Once again many thanks for your time.

    Out of curiosity what multimeter are you using?

    Regards

    Ola

  • 0 in reply to Ola Oni
    TI__Guru* 93975 points

    Ola,

    I have an Agilent 34410A meter and two other older 34401As.

    Joseph Wu

  • 0 in reply to Joseph Wu
    Prodigy 30 points

    Joseph,

    Nice. I should have just saved up and bought  something like the Agilent or even buy one second hand rather than purchasing the Owon. Not sure if I can still return it.

    Regards