ADS124S08: Analog input range.

Hi Namitha,

I'm not totally sure what you are asking for with 'hardware'.  Do you mean hardware setup within the ADS124S08?  Or are you asking about hardware as it relates to components related to the input filtering, connecting sensors, etc.?

The ADS124S08 has an absolute input range relative to the analog supply voltage and the sensor input must be within the supply range.  The measurable input range, or full-scale range, relates specifically to the value of the reference voltage and any PGA gain applied (+/- VREF/PGA).  The input range can be restricted further if the PGA is enabled and gains greater than 1 are used.  

The reference selection and PGA gain selection are examples of hardware configuration within the ADS124S08.  An externally supplied reference, and analog filters relate to hardware outside of the ADS124S08.  So there are configurable options, but the only specific requirement is that the input voltage is within the range I stated above.

I would suggest taking a look at how the example application circuits in the datasheet are configured to give you an idea of what needs to be done.

Best regards,

Bob B

Hi Namitha,

You state that you will be using an external 5V reference.  I will assume this also means you will be using 5V unipolar analog supply (AVDD = 5V and AVSS =AGND) and the full-scale range will be +/- 5V/PGA gain.  If you need to measure a sensor with an output from 0 to 5V, the you will need to set the AINP input on the ADC to the sensor output and AINN to AGND (and sensor GND will also need to connect to AGND).  This will allow a measurement of 1/2 of the full-scale range (the positive half).  The LSB value will be 596nV, but conversion noise will lower the overall resolution to a best case scenario which will depend on the desired data rate and the PGA settings.  But in general even with noise you should be able to measure 10mV with no issue.

However, the PGA cannot drive all the way to the supply rails, so if the measurement is 0 to 5V, then you will not be able to enable the PGA nor can you increase the gain beyond 1.  So the PGA must be disabled and bypassed so you can measure the input voltage to the supply rails. This measurement is called single-ended ( a measurement referred to a common point (GND)).

If the output of the sensor is differential, such as from a Wheatstone bridge, then the PGA can be enabled and gain can be used.  The gain is adjusted in the configuration register in 0x03 (the Gain register).  See the datasheet for details.

Best regards,

Bob B

Hi Namitha,

For some time TI has used ENOB and effective resolution interchangeably for many of the low speed data rate Delta-Sigma ADCs.  ENOB better represents AC input signals where a signal is injected and the outcome is relative to SINAD.  But ADCs such as the ADS124S08 are meant for very slow moving signals at or near DC.  These devices do not input a signal, but rather are measured by shorting the inputs and determining the noise which provides the best possible resolution of the ADC.  This measurement determines effective resolution.  The calculation for effective resolution is given in equation 1 of the ADS124S08 datasheet on page 24.

Tables 2, 4, 6, and 8 give the number of bits of effective resolution for the given operation at the top of each table.  However, you may really want to know the noise-free resolution (stable counts) instead of effective resolution (RMS value).  Also, depending on the reference voltage being used the resolution may vary.  I prefer to use the noise voltage and in particular peak-to-peak noise and back calculate that way to determine my actual measurement resolution.

There will be a large difference in potential resolution depending on if the measurement is single-ended (referenced to ground) and using a unipolar analog suppy (AVDD = 5V and AVDD = AGND).  I talked about this in an earlier post.  In this case you will be limited to a gain of 1.  The lower the data rate the lower the noise as can be seen in the noise tables.  At 20sps and a gain of 1 using the low-latency filter (Table 3) you can see that the p2p noise is 8uV.  Even though the converter resolution is 596nV, the 8uV of noise is the limiting factor. 8uV is your actual resolution size and allows 1250 (10mV/8uV) measurable counts for a maximum 10mV signal range.

If you can apply the input signal where you can also apply gain up to 128, then you can make significant improvement in resolution.  At 20sps, gain of 128, low-latency filter and 5V reference the p2p noise lowers to 510nV (Table 3).  The ADC resolution becomes 4.65nV, but because of noise the actual measurement resolution is 510nV.  The measurable counts improves as well to 19608 (10mV/510nV), so you can see that gain can greatly improve the measurable resolution.

Notice that I never used ENOB or effective resolution as the number is only viable for a 2.5V reference.  Instead I used worst case noise numbers for stable measurement values.  However, this does not include the effect of any external noise or noise of the source but rather just the ADC.

Best regards,

Bob B