• State Resolved
  • Locked Locked
  • Replies 4 replies
  • Answers 1 answer
  • Subscribers 45 subscribers
  • Views 577 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.

ADS127L11: SNR vs. Input range (1xVREF, 2xVREF)

TonyMCC
Intellectual 770 points
Part Number: ADS127L11
Other Parts Discussed in Thread: THS4551

Please explain why changing the analog input range from 1xVREF to 2xVREF only increases SNR by 1db ? I would have expected that increasing the input range by a factor of 2x would increase the SNR by 6db.

I have tried testing this using the ADS127L11 eval. board and I do see ~5db increase in SNR when changing from 1xVREF to 2xVREF input range.

Thanks,

Tony

  • 0
    TI__Mastermind 47830 points

    Hello Tony,

    Please see below comments from the ADS127L11 systems engineer:

    +1 dB increase in dynamic range for the 2x input range is correct. I think the root of confusion could be that the eval board represents noise in codes, not volts. When noise is represented in codes, the noise code values are nearly the same between the ranges, so it appears that DR should increase 5-6dB for the 2X range because the input range is now twice as large.

    But the volts-per-code weight changes with the input range.

    In 1x range with 2.5V ref, the volt per code is 2.5V/2^23 = 0.3uV/code

    In 2x range with 2.5V ref, the volt per code is 5V/2^23 = 0.6uV/code

    I measured 36 noise codes for 1x range and 33 noise codes noise for 2x range on the eval board. Converting noise codes to uV noise results in

    1x range =  10.8uV

    2x range = 19.8uV

    Converting to dynamic range for each range:

    DR(1x) = 20log(5V/(2*rt(2)*10.8uV = 104.3dB

    DR(2x) = 20log(10V/(2*rt(2)*19.8uV = 105.0dB

    These are within the range of expected DR results.

    ****

    The ADS127L11 changes it’s input range to 2x by switching in a smaller input sampling capacitor. The input sampling capacitor is reduced approximately 50% for 2x range, therefore twice the amount of voltage is need to accumulate the equivalent amount of charge compared to 1x range. The input charge is internally summed with the charge of the reference-voltage sampling capacitor within the modulator to establish the ADC FS code level. When the input sampling capacitor is decreased by 1/2, and everything else being the same, the dominant ADC thermal noise component - “kT/C” noise - also nearly doubles, but not quite matching the 2x increase of input range, therefore the +1 dB increase in dynamic range.

    Another way to think of this is that the input sampling capacitor network in 2x mode creates a capacitive voltage divider to extend the input range.  This does not fundamentally change the internal noise of the ADC, and when the noise is referred to the ADC inputs, before the 1/2 attenuation, the over dynamic range changes very little.

    Regards,
    Keith Nicholas
    Precision ADC Applications

  • 0 in reply to Keith Nicholas
    Intellectual 770 points

    Thanks Keith for the detailed explanation, that very helpful.

    The reason I ask is because I am trying to determine why my ADS127L11 evaluation board test results are not close to the datasheet performance with regard to SNR. My results indicate about 95.5db of SNR and the SNR spec is listed at 106d typ. so about a 10db difference.

    The evaluation board document only has one plot in section 6.5 but this is not the exact same test conditions as what I am using.

    One other thought...is it possible that the ADC drive amplifier (THS4551) adds the additional noise ? FYI for my testing I am using DS360 signal source.

    Here is a copy of my test results...

  • +1 in reply to TonyMCC
    TI__Mastermind 47830 points

    Hello Tony,

    The EVM is capable of around 106dB.  The THS4551 amplifier does degrade the SNR, but the reduction is much less than 0.5dB.

    I actually used the DS360 signal source for some testing.  This generator has excellent frequency stability, but its broadband noise is the limiting factor.  You would need to use an external multi-order passive bandpass filter to reduce the noise to an adequate level.

    For most of the EVM testing, I used an Audio Precision 2700 Series signal generator.  This generator still limits the maximum SNR that can be measured, but you should be able to get around 105dB.  If you add a simple RC filter on the output with a cutoff around 2kHz when using a 1kHz source frequency, you can then get close to 106dB measurement with this instrument.

    Regards,
    Keith 

  • +1 in reply to Keith Nicholas
    Intellectual 770 points

    Thanks Keith,

    I will add the filter as you suggested and let you know how I make out.

    Regards,

    Tony