• State Resolved
  • Locked Locked
  • Replies 9 replies
  • Answers 1 answer
  • Subscribers 62 subscribers
  • Views 315 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.

LM5122: LM5122 options for calibration

gilles mourier
Prodigy 230 points
Part Number: LM5122


I am using an LM5122 in 29.5V 5A boost converter with an input voltage in the range [17.5V - 27.2V] and I would like to know what would be the best option to calibrate the output voltage.

The tolerance on the output voltage returned by Webench is ~+/-3.5% and I would like to get something better than 1%.

At the moment, I have 2 options in mind:

1) adding a digital potentiometer to adjust the bottom resistor on the feedback divider.

2) sinking or sourcing a constant current to the feedback node to add a constant offset using for example a DS4303.

At the moment, the 2nd option is my preferred option as it is easiest to implement on the existing board and the adjustment resolution remains the same over the full adjustment range.

My problem is I don't know if I am right adding a calibration like this as this assumes Vref of the LM5122 is a fixed value in the range [1.188V - 1.212V] which never changes .

Any suggestions are welcome.

  • 0
    TI__Expert 7450 points

    Hi Gilles,

    Thank you for choosing LM5122.

    First of all, if you have your fixed target output and just want to make it more accurate, you can use some resistor combinations for the feedback. Note that the internal FB reference accuracy is 1% so you won't get anything better than that.

    If you do want to use external source, it is eligible but please pay attention to the resistance you select. The FB is a input of the error amplifier internally. So applying the current source should work but note that there is ~5nA bias input current on the FB of the amplifier.

    Please let me know if there is any question.

    Thanks,

    Yinsong

  • 0 in reply to Yinsong Cai1
    Prodigy 230 points

    Hi Yinsong,

    Thanks for your reply, below is a short scenario to illustrate the problem I have. 

    The FB reference voltage of the LM5122 is 1.2V +/-1% [1.188V - 1.212V].


    On a given LM5122 with a reference voltage of 1.195V determined during the calibration process.
    Can I consider this voltage is going to remain to its initial value or is it still subject to move anywhere in the range [1.188V - 1.212V]?

    1) If the reference voltage remains stable around its initial value or the tolerance around the initial value is less than 1% then I think adding an offset to the feedback node could probably help me to get a better accuracy on the output voltage.
    If the tolerance around the initial value is better than 1% how to quantify this error? 

    2) If the reference voltage is anywhere in the range [1.188V - 1.212V] whichever the initial value measured on the chip, then I cannot expect to get better than 1% and the simplest option is certainly to use resistors with a tolerance good enough so the final error comes mainly from the tolerance affecting the LM5122 FB reference. 

    Regards

    Gilles

  • 0 in reply to gilles mourier
    TI__Expert 7450 points

    Hi Gilles,

    There could be some additional change against temperature but usually less than 0.2%. I need to confirm this with our team.

    Thanks,

    Yinsong

  • 0 in reply to gilles mourier
    Prodigy 230 points

    Hi Yinsong,

    Did you get confirmation from your team about the 0.2% tolerance in your last message?

    Regards

    Gilles

  • 0 in reply to gilles mourier
    TI__Expert 7450 points

    Hi Gilles,

    Sorry for the late response. I have confirmed that the 1% is actually with the -40 to 125 degree C junction temperature and no additional tolerance. It should be good.

    Thanks,

    Yinsong

  • 0 in reply to Yinsong Cai1
    Prodigy 230 points

    Hi Yinsong,

    Thanks for your reply, but I was not expecting you confirmed the 1% [-40 125C] from the datasheet but the 0.2% from your previous post.

    If I cannot expect that for a given chip Vref remains to its initial value with a tolerance better than 1% (0.2% was fine for me) then, I think there is no benefit calibrating the output, the easiest option being probably using very accurate resistors as there would be no way to get anything better than the 1% affecting Vref.

    Did I miss something?

    Regards

    Gilles

  • +1 in reply to gilles mourier
    TI__Expert 7450 points

    Hi Gilles,

    Sorry I didn't make it clear. The 0.2% I mentioned was just a general guess and now I confirm that this is wrong. There is no 0.2% or any additional tolerance in LM5122. And your understanding is correct.

    Thanks,

    Yinsong