I'm using multiple ADC channels on the Robovero expansion board to sample neural signals captured from human muscles. Because the neural signals are bipolar and the acceptable range of Robovero ADC is 0-3.3v. I built a circuit which added 1.5 v to the raw neural signals so that the ADC input is now between 0v and 3.3 v. However, I observed a lot of impulse noises in the sampled data, which were random impulses that ranged from 0 to 3.3v.
I also tried to use a function generator to generate a sin wave as the analog input. Same impulse noises were observed. As long as I connected some wires to the ADC pins, there would be impulse noises. I cannot figure out why.
I tried to use some other commercial ADC systems (one from Motion Lab, one from Measurement Computing, and one from Freescale MPC5566 evaluation board) to collect data from my voltage adding circuit and the neural signals. The results all looked perfect.
In my testing, I left the Vrefn and Vrefp floating. I believe these two pins are already tied to VDD and VSS. I only connected the GND of Robovero to the ground of my voltage adding circuit as well as the ground of the input sensor. Is this configuration correct? Should I connect Vrefn and Vrefp to some external voltages?