ADS114S08: ADC input range with 5V ref, AGND with gain.

Ok lets say I have 5 inputs connected (single ended referenced to AGND) to AIN0-4 that can range from 0-5V. I also have 5 inputs connected to AIN5-9 that range from 0-125mV (single ended referenced to AGND) I read channels 0-4 with gain of 1 and then read channels 5-9 with gain of 32. Should this work?

Ok I think I got it. For 5-9 I have to mux in 2.5V as Vinn and that will center the 125mv between AVSS and AVDD and let me use a gain of 32. Yes?

Hi Greg,

AINN and AINP have similar restrictions. You could make a measurement relative to 2.5V at AINN, but if the sensor output is single-ended relative to AVSS, then AINP will only traverse from 0 to 125mV which will be outside the input range for AINP even at a gain of 1 with the PGA enabled. You would still need an op amp or INA to drive the + output of the sensor to be 0 to 125mV relative to the AINN (2.5V) input.

Best regards,
Bob B

So do you have any ADCs that can read 0V- 0.125mV with gain internal gain of 32 when configured to do so by SW AND read 0-5V with no gain when configured to do so? All single ended.

Need a totla of 16 channels that can do that. Of course in the fastest most cost effective way possible.

Hi Greg,

our ADS1120 (SPI), ADS112C04 (I2C) and ADS112U04 (UART) could measure single-ended signals using an external 5V VREF while providing gains of 1, 2 and 4. The devices can only support four single-ended measurements, but you could easily extend that using an additional MUX.

Another option could be ADS1158. This device supports 16 single-ended inputs, but will require one external gain stage.

One general misconception that many customers have is that they think they always need to utilize the complete full-scale range of an ADC. But in the end it is only important that you can resolve the input signals to the level that is required by your application.

Instead of using a gain of 32 in a 16-bit ADC, you could for example simply use the 24-bit options of the above mentioned devices (ADS1220, ADS122C04, ADS122U04, ADS1258, ADS124S08) at a lower gain to get the same performance even though you only utilize a fraction of the full-scale range.
If you get one more bit of resolution, this is nothing else than having an additional gain of 2.

To give you an example:
ADS1120 at 20SPS and Gain=4 yields an input referred noise of 15.6uVrms. If you could use Gain=32 in a single-ended configuration (not possible with ADS1120) you would achieve 1.95uVrms.
ADS1220 at 20SPS and Gain=4 yields an input referred noise of 1.2uVrms.
So you can see that ADS1220 at Gain=4 would yield better resolution than ADS1120 at Gain=32.

ADS124S08 at 20SPS and Gain=1 yields an input referred noise of 0.8uVrms.

Regards,

If I am willing to accept 12 bits of resolution then using 1/16th the range is just as accurate measuring 0-250mV as 0-4V using the ADS114S08. Correct. 

Or to state it another way of I bypass the gain I can measure either a 0-5V input or a 0-250mV input with as much accuracy I just limit my resolution to ~12 bit by not using a gain of 16 on the 0-250mV signal. Correct?

Hi Greg,

that is correct.

However one thing I should note is that the ADS114S08 - and most of our other ADCs - use a bipolar coding scheme. Means the 16 bits are distributed over the differential input voltage range from -FS = -5V to +FS = +5V in your case. As you are implementing a single-ended measurement you are only utilizing the positive half of the code range, thus effectively losing 1 bit of resolution.

The ADS114S08 provides a noise free resolution of 16 bits at Gain = 1 at almost all data rates or in other words an input referred noise of (2 x 5V / 2^16) = 153uVpp.

That means you can resolve a 250mV signal with log(250mV/153uVpp) / log(2) = 10.7 bits.
The 5V signal is resolved with 15 bits.

If you would select the internal 2.5V VREF for the 250mV signal measurement you would get one more bit of resolution. Means you could resolve the 250mV signal with 11.7bits.

If this does not meet your needs, it might be better to look at the 24-bit option, ADS124S08, or to use ADS1120 at gain = 4, PGA bypassed.
Both options would be able to resolve the 250mV signal with more than 12 bits.

Regards,

Thank you. Are the ADS124S08 and ADS114S08 pin for pin compatible in the same package?

Hi Greg,

yes, they are.

Regards,