• State
  • Locked Locked
  • Replies 3 replies
  • Subscribers 45 subscribers
  • Views 432 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.

ADS1224 noise

Steven Burr
Prodigy 110 points
Other Parts Discussed in Thread: ADS1224, ADS1248, ADS1220

What is the noise of the ADS1224? It says in the datasheet on page 3 that the noise is 0.8e-6 of the FSR. If I figure the FSR is a 5V span, this is 6uV-rms of noise, which would be 21 counts of noise or just over 4 bits of noise. I figure these calculations must be wrong because if there was 4 bits of noise there would be little reason to use this ADC over a 16-bit ADC with very low noise (I suppose decimation would give you a bit but not much) .

  • 0
    TI__Guru** 112715 points

    Hi Steven,

    Many of our Delta-Sigma converters have variable decimation/data rates and noise tables are developed and placed in the datasheets.  For this device it was done differently. Generally what is seen is the noise get worse as you approach full-scale.  If you were to plot the noise you would see sort of a smiley face affect. With this device it is not uniform from side to side.  This is shown in Figures 11 and 12 of the datasheet.

    If noise tables had been created for this device data would have been collected for a shorted input case.  The inputs are shorted and biased to mid-supply.  This is what is normally done for noise measurements.  The ENOB is calculated as ln(Full-Scale Range/noise RMS)/ln(2), which is around 20 bits as you stated.

    Decimation does help. What is happening with the oversampling converter is quantization noise is pushed into the higher frequency bands, and the digital filter is low pass stripping out the higher band noise. You can probably get another bit or so of improvement with additional averaging.

    20-bit performance still gives a greater degree of resolution than 16-bit.  If your sensor input is a small signal and you use an ADC with an integrated PGA, the input referred noise will reduce with increased gain.  You can get a better idea of this by looking at the the noise tables for the ADS1248 family of devices.

    Basically you have to weigh the benefit of a few more bits versus increased cost.  For some situations it may not matter, as the sensor output may be so noisy that you may not be able to resolve better than 16-bit anyway.  In other situations the ability to resolve 1 or 2 more bits may be of great benefit.

    Best regards,

    Bob B

  • 0 in reply to Bob Benjamin
    Prodigy 110 points

    20 bits does not sound too bad. I am specifically trying to avoid the ADS1248. While it is a great ADC, its complexity makes it difficult and costly to write hardware\firmware code.

  • 0 in reply to Steven Burr
    TI__Guru** 112715 points

    Hi Steven,

    We will be releasing a new part, the ADS1220, in a couple of weeks that may be of interest.  It is less complex in terms of communication and register set compared to the ADS1248, but retains many of the same features.

    http://www.ti.com/product/ads1220

    Best regards,

    Bob B