# How to choose a suitable ADC ?

Status
Not open for further replies.

#### glias

##### Full Member level 2
hello all,
I'm wondering what is the best way to use ADC adaptation stage.
I have a sensor that covers the voltage range from -12 to +12V. and I wondering if I should better use a bipolar ADC or unipolar ? what is the risk to adapt and reduce the voltage range ? Does it have an influence in noise and performances in the analog chain ?

regards

#### karesz

Hi,
Most moderner ADCs have an input range of +/-2,5V, +/-5V or +/-10V (or only unipolar all these= 0..+2,5/5&10V), also you must so & so use some attenuator/correction...
I would select the biggest possible input range type & if your signal is +/-12V, you can atteuate it to +/- 10V as relative less (20%)loss on dynamic range...
Better is to change the output amplitude of your source to i.e. +/-10V_if you ca do it.
K.

Pete.York

Points: 2

### Pete.York

Points: 2

#### glias

##### Full Member level 2

Hello karesz
and thanks for your reply. I'm not sure to understand why is it better. Take an example :

First example :
10V output sensor range => divider by 2 stage => ADC (5V dynamic range)
imagine that the noise of the sensor is 1mV.
the quantum of ADC is : 5V/2^14=305µV
so the noise that the ADC will see is about 1,64bits lost for the dynamic

2sd example :
10V output sensor range => ADC (10V dynamic range)
1mV (noise sensor)
noise that the ADC will see is 1mV/600µV is the same : 1,66bits

The only thing that the 2sd example is better is for the offset (which is constant).

What do you think ?

Pete.York

### Pete.York

Points: 2

#### karesz

, Hi!
I think, that I dont know how big is your ADCs resolution.! Is it a true 14-Bit system pls? Why did you calculate the LSB value with 2^14? OK, for theoretical limits I can understand, but for real situations use pls the real ADC data!
Be care pls, calculate every time with the concrete ADCs true "Bit-Resolution", they arnt equivalent from all types & brands, and specification is often some unclear...
A 0..10V range at true 10 Bit is practically 10mV (10V/2^10) for a Bit Quant/LSB value. Or do you have a 16-Bit ADC (with true 14-Bit resolution)?
If you have a +/- system(bipolar) => your resolution will be double of the unipolar, so its better to the same noise floor with factor two version...
I did wish to say: if sensor output has a range of +/- 10V: its better than +/- 12V, than you dont need a (depending of ADCs input circuit_maybe need a plus buffering) plus attenuating in the ADC circuit...
K.

Pete.York

Points: 2