• State TI Thinks Resolved
  • Locked Locked
  • Replies 5 replies
  • Answers 2 answers
  • Subscribers 46 subscribers
  • Views 466 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.

ADS1148-Q1: gain calibration

Tamas Fenyvesi
Prodigy 180 points

Part Number: ADS1148-Q1
Other Parts Discussed in Thread: ADS1148, ADS1120-Q1, ADS1248

Hi,

ADS1148 / ADS1148-Q1 datasheets state offset error of +/-1LSB is after calibration.

How about gain error 0.5%? Is it without calibration or after calibration?

Thanks,

Tamas

  • 0
    TI__Guru 56892 points

    Hello Tamas,

    This would be without gain calibration.

    Generally for the ADS1148, I wouldn't use the gain calibration command. The gain is factory trimmed for each gain, and you can get very good gain error results without any gain calibration. I would only short the inputs together at a supply midpoint and run a self offset calibration.

    If you do want to run a self gain calibration, you would put in an input that would give you a full scale ADC output and the calibration would scale that to get 7FFFFFh. In PGA=1, this would be putting in the VREF as an input. In a different gain, you would need to put in what you think is the full scale voltage. Even in PGA=1, the reference voltage needs to be in the input range of the PGA, which means that the reference voltage's VREFCOM cannot be at AVSS (GND for a unipolar supply).

    Regardless, my advice would be to not run a gain calibration, and only do a self offset calibration.

  • 0 in reply to Alexander Smith
    Prodigy 180 points

    Hi Alex

    Thank you. What does then ADS1148-Q1 datasheet SBAS674A / part 8.4.5.3.2 and MUX setting part 8.6.2.3 / MUXCAL[2:0] = "010" refer to?

    It's easy not to run gain calibration. But what you mean "very good gain error results", if datasheet parameter is 0.5%, i.e. ENOB=7.7 for a 16 bit converter (with ideal reference) ?

    Best regards,

    Tamas

  • 0 in reply to Tamas Fenyvesi
    TI__Guru 56892 points

    Hi Tamas,

    I don't understand your first question. The device has a gain calibration register that can be written to, however I do not recommend that you use it as each gain is trimmed at the factory. 8.4.5.3.2 is in regard to the system gain calibration and 8.6.2.3 / MUXCAL[2:0] = "010":

    101: REF0 monitor. The analog inputs are disconnected and AINP and AINN are internally connected to (V(REFP0) – V(REFN0)) / 4

    Can you please elaborate on what you would like clarified? 

    I would consider 0.5% max to be very good gain error for a 16 bit converter. ENOB is more of a way to express the noise of the converter as opposed to the gain error. Please see table 2 in the datasheet for details.  

  • 0 in reply to Alexander Smith
    Prodigy 180 points

    Hi,

    It is a nice product for e.g. ECG, where resolution is important, but absolute accuracy is not.

    0.5% might seem a good gain error for a sigma-delta converter. But, regarding to absolute accuracy, even a common 10 bit SAR is usually better than 0.5%.

    Related ADS1120-Q1 is 0.1%, so it's not clear to me if 0.5% is a typo or ADS1148-Q1 is so much worse.

    Best regards,

    Tamas

  • 0 in reply to Tamas Fenyvesi
    TI__Guru 56892 points

    HI Tamas,

    Fair enough. ADS1148-Q1 is the 16bit version of the 24 bit ADS1248, whose gain error is 0.02% max. From this respect the ADS1120-Q1 has higher accuracy.