[SOLVED] Correct differential inputting to an AD converter

I'm using a TI LMP7721 3-femtoamp bias current op amp to measure an extremely small current in an ion cell. I have it in inverting mode with a guard around the input, driven to follow the non-inverting input voltage. This works extremely well and gives me an output between +/-5V.

After some discussion in an earlier post relating to choice of a suitable AD converter I have fixed upon an Analog Devices AD7171 which uses the sigma-delta routine; has 16 bit resolution and a conversion rate of 125 S/s. It also has a differential input of +/-5v (with Vdd=Vref=5v).

I am proposing to pass the -5v rail voltage to the negative AD input and the amplifier output to the positive. Now, relative to ground, this may mean that there will sometimes be a negative voltage applied to the positive input (unless I apply bias which would be a pity since it then cuts my dynamic range).

Looking at the spec and absolute limits however I see this:


Does this kill my idea? If so my fall-back will be to ground the negative input and ensure the signal is suitably biassed. Your thoughts would be appreciated.

Like with similar single supply differential input ADC, the input voltage range is limited to voltages between GND and VDD. To utilize the full 16-bit resolution, the differential input voltage ranges from -Vref to +Vref. Please review the datasheet paragraph "bipolar configuration" for details.

Hi,

I think your configuration doesn´t make sense.

* You say: "I am proposing to pass the -5v rail voltage to the negative AD input"
* and then you say: "Now, relative to ground, this may mean that there will sometimes be a negative voltage applied to the positive input (unless I apply bias which would be a pity since it then cuts my dynamic range)."

If you put -5V to the inverting input of the ADC..
* then your input is referenced to this -5V and not to GND (besides -5V to -IN is not allowed)
And since the differential input is +/-5V --> this means the decodeable input range at the IN+ is -10V...0V (still not allowed)

I assume your Ion-Cell generates currents in one direction only. If so, then choose an ADC with unipolar decoding.

Klaus

KlausST said:

Hi,

I think your configuration doesn´t make sense.

* You say: "I am proposing to pass the -5v rail voltage to the negative AD input"
* and then you say: "Now, relative to ground, this may mean that there will sometimes be a negative voltage applied to the positive input (unless I apply bias which would be a pity since it then cuts my dynamic range)."

If you put -5V to the inverting input of the ADC..
* then your input is referenced to this -5V and not to GND (besides -5V to -IN is not allowed)
And since the differential input is +/-5V --> this means the decodeable input range at the IN+ is -10V...0V (still not allowed)

I assume your Ion-Cell generates currents in one direction only. If so, then choose an ADC with unipolar decoding.

Klaus

Click to expand...

I think your configuration doesn´t make sense.

Click to expand...

And I think you're right. You can see that doing this stuff is not my usual line of work. The AD7171 has an analog differential input and here I was treating it as digital (or, rather, trying to). In any event as FvM points out, with the single positive supply you can't input a negative signal (relative to ground).

Thanks for pointing this out - we get a little smarter each day (supposedly).

Hi,

In any event as FvM points out, with the single positive supply you can't input a negative signal (relative to ground).

Click to expand...

True, but there are exceptions. Like the energy metering ADC MCP3911.

Therefore I always recommend to read the datasheet.
* Absolute input voltage range (each input referenced to GND)
* differential input voltage range (voltage difference between two inputs)
Any application needs to comply with both specifications.

Klaus