I need to test a differential ADC (with AIN and /AIN).

After some tests, I don't understand the behavior.

Let's use the AD6644 (taken from this Analog devices document):

As shown in the functional block diagram, the AD6644 has complementary analog input pins, AIN and /AIN. Each analog input is centered at 2.4 V and swings ±0.55 V around this reference (Figure 21). Because AIN and AIN are 180° out of phase, the differential analog input signal is 2.2 V peak-to-peak.

Does this mean I cannot feed AIN or /AIN with a voltage higher than 2.95v (2.4+0.55) or lower than 1.85v (2.4-0.55). Then the input range is 1.85-2.95v. What is the behavior if I put /AIN at ground (0V) and the AIN input signal at 2.4v +-0.55v ?


Maybe my misunderstanding comes from the board I am using. This board doesn't contain any differential output opamp and contains only two different signals for feeding AIN and /AIN. With the possibility to set one or both inputs at 2.5V DC. It uses the ADC with no AC-coupled inputs... just as a "differential" (AIN - /AIN) ADC...

Obviously, the board uses this diff ADC as a single-ended ADC.

Does the input offset of the signal matter or not ?

A) /AIN=0V AIN=SINE 0v offset, 2.2V p-p
B) /AIN=2.4V AIN=SINE 2.4v offset, 2.2V p-p

A and B are supposed to give the same result. Aren't they ?

Or A doesn't work or is not efficient because doesn't respect the 1.85-2.95v input range.

And AIN and /AIN AC-coupling at 2.2v p-p with 0V offset AND 2.4V offset (or any other offset) are behaving same ? Only the 2.2v p-p does matter or the 1.85-2.95v input range too ?


2 Answers 2


Yes, the AD6644 has a differential input stage. Citing its datasheet, the maximum damage free input voltage is your analog supply voltage. Which doesn't mean that's the maximum usable voltage, i.e. a point that isn't already clipping. That would be, as per the datasheet, p.3 a 2.2V betweeen AIN and /AIN.

To stress this further (emphasis by me):

The AD6644 input voltage range is offset from ground by 2.4 V. Each analog input connects through a 500 Ω resistor to a 2.4 V bias voltage and to the input of a differential buffer (Figure 21). The resistor network on the input properly biases the followers for maximum linearity and range. Therefore, the analog source driving the AD6644 should be ac-coupled to the input pins.

In other words: you don't use a 65MS/s ADC to measure a DC voltage. You use it with capacitors in series to its inputs, and the ADC itself establishes the 2.4V bias. It doesn't really matter what absolute voltage offset your analog signal has – that offset doesn't pass through the capacitors – it's only important that the peak-to-peak voltage doesn't exceed 2.2V.

  • \$\begingroup\$ Means I can feed the ADC with two sinewaves at 0V offset 2.2v pk-pk and 180° shifted ? The signal will be a sinus from -1.1v to 1.1v. That's the things difficult to understand (the internally 2.4V bias). Maybe this is just the board I am using which makes me misunderstand. (There is no Diff output opamps) Thank you for you reply. \$\endgroup\$
    – None
    May 28, 2016 at 23:16

You are correct on the maximum input range for valid conversions.

Because the internal ADC is differential, if you put the \$\overline A_{in}\$ at 0V, I would expect the Overrange bit to assert.

AD6644 Pin functions

If you wanted to operate this in a single ended input manner, then I would expect to need to set the \$\overline A_{in}\$ at 2.4V which is internally generated by the ADC, so you could just leave it floating (albeit with potential noies issues).

  • \$\begingroup\$ Behind Marcus Muller said the DC offset doesn't matter, only the 2.2v p-p between AIN and /AIN does matter. With this assumption, I could set /AIN at the ground and AIN at a sinewave without offset and 2.2v p-p. \$\endgroup\$
    – None
    May 28, 2016 at 23:21
  • \$\begingroup\$ I said the DC offset didn't matter because you should be AC-coupling the inputs! Not in general that DC offset doesn't matter! \$\endgroup\$ May 28, 2016 at 23:27
  • \$\begingroup\$ I edited my post in order to be more precise. \$\endgroup\$
    – None
    May 28, 2016 at 23:32

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.