Hi!
We are working in the design of an EEG acquisition system with the ADS1299. We are facing some problems when evaluating the CMRR of our front end.
The design was made following the recommendations of the datasheet. The device is used in referential mode, with unipolar supplies of 5V and 3.3V. The configuration registers where set as follow:
CONFIG1= 0x95 // sample rate of 500sps
CONFIG2=0xC2
CONFIG3=0xE0
CHnSET=0x60 // Gain=24
For the input RC filter we used 100k ohm resistors and 200pF capacitors (differential), obtaining the cutoff frequency near 4kHz.
To measure the CMRR we connected all the channels including the reference channel to a 10Hz, 2Vpp sine signal with a DC offset of 2.5V. Then we calculated de CMRR as follows:
Acm= Vo/Vi
CMRR=20log(24/Acm)
Where Acm is the common mode gain, Vo is the peak to peak voltage obtained at the output and Vi is the peak to peak voltage at the input. Vo was measured after applying a LPF to the output signal.
When we measure the CMRR of the complete front end we obtained values around 60dB, far from what we were expecting. The results improve significantly when bypassing the RC filter, reaching over 110 dB (closer to the datasheet values). The following images show the output obtained in those conditions:
The first image shows the output signal with the RC filter and the second one shows the output without it.
We try using a greater capacitance (1uF) in the RC filter of one of the channels but this didn't improve the results.
-
Any suggestion on which could be the reasons for these results?
Additionally, when we do the fft of the data acquired in the CMRR measurements, we obtain appreciable harmonics of the original signa, especially when we bypass the RC filter:
-
Is this normal?
We are developing a second version of the circuit so any further recommendation for the design would be highly appreciated.
