What is a "true differential input" ADC really for?

[opampsmoker]

Hi,
The ADS8867 is a "true differentail input" ADC


ADS8867 datasheet

https://www.ti.com/lit/ds/symlink/ads8867.pdf?ts=1603976950110&ref_url=https%253A%252F%252Fwww.ti.com%252Fproduct%252FADS8867

Presumably the circuitry used to supply the reference, and the diff amps shown supplying the AINP and AINM inputs are only there to provide very high accuracy?

I mean, if a signal is not quickly changing, and great accuracy was not required, then would it be ok to just feed the unipolar signal into AINP and AINM via a simple opamp buffer?...and for the reference, just put a divided down voltage into the REF pin with a big capacitor on the REF pin?

...i mean, when one uses the ADC in a simple PIC micro, you dont bother with all that circuitry shown (in datasheet) feeding the reference, and certainly dont bother with the shown diff amp input to the ADC input.

 

[KlausST]

Hi,

Yes, accuracy and precision.

Single ended ADC need to refer to GND. But GND may be noisy, it may bounce from one point of a PCB to the other. With single ended wiring there is no chance to ovoid these noise and bounce errors. With difference signaling those errors don't exist.

Klaus

 

[danadakk]

Differential mode also rejects CM signals that affect the measurement of a differential
voltage riding on top of a CM signal/bias. As Klaus points out that could be noise, or
DC offsets or a bias used in the circuit.

Some A/Ds can handle signals, albeit of a few mV, outside their rails, like when
sensing a current thru a shunt in a power supply lead. Read data sheet carefully for
this capability.

One example of a processors DelSig capability where CM measurement capability
can do 100 mV outside its rails.

Regards, Dana.

 

[dick_freebird]

Mainly, you want a true differential ADC when
you are dealing with an otherwise-true-differential
signal chain / source. As many low voltage analog
signal chains are, after about the 2.5V / 250nm
node.

 

[danadakk]

True or Fully Differential, what it means -

https://pdfserv.maximintegrated.com/en/an/AN1108.pdf

Regards, Dana.

 

[opampsmoker]

Thanks, so the attached is also a possible way to use ADS8867 ADC, but more noisy and less accurate?
However, if the signal trace distances are short, and the sampling rate low, then the attached is fine, as long as great accuracy is not required?

ADS8867 ADC

https://www.ti.com/lit/ds/symlink/ads8867.pdf?ts=1604765648463&ref_url=https%253A%252F%252Fwww.ti.com%252Fproduct%252FADS8867%253Futm_source%253Dgoogle%2526utm_medium%253Dcpc%2526utm_campaign%253Dasc-null-null-GPN_EN-cpc-pf-google-eu%2526utm_content%253DADS8867%2526ds_k%253DADS8867%2526DCM%253Dyes%2526gclid%253DCjwKCAiAqJn9BRB0EiwAJ1SzteMR6r5p9ZGJQkewWa5lpFfT5VX3WJGgWgAArV3WeOnoJGD7u56gVBoCK8gQAvD_BwE%2526gclsrc%253Daw.ds

LM7321 opamp

https://www.ti.com/lit/ds/symlink/lm7321.pdf?ts=1604749264897&ref_url=https%253A%252F%252Fwww.ti.com%252Fproduct%252FLM7321

 

[KlausST]

Hi,

if great accuracy is not required you may do anything... ;-)

REF-Driver: be sure it can handle 2uF directly connected to it´s output. Maybe a 100R (mabe less) from output to C could be helpful.
Be sure the REF source is stable and free of noise. An RC low pass in front of the buffer should be a good idea.

What´s the use of the OPAMP unity gain buffer at the signal input? It depends on signal source impedance, signal bandwidth, cable length if it´s a benefit.

If the signals are of low frequency .. an RC shouldn`t be too much effort.

******
From the accuracy and precision view:
* use the same REF for ADC and signal source - if possible
* use separate GND_sense wiring from signal source back to ADC AIN-. With identical source impedance than the signal line itself.
* Treat signal line and GND_sense most equally.
* Use at least a simple anti aliasing filters in front of the ADC.

Klaus