Hello,
I am trying the ADS1243 converter in an application that requires to read the output of a capacitive half-bridge (ratiometric) sensor.
The sensor excitation is provided by square wave signals with 180 degree phase reversal produced by the 'Q' and '-Q' outputs of a 74HC74 in a divide-by-2 configuration. The excitation ranges from roughly 300 Hz to 1kHz.
The output signal is demodulated through a syncronous rectifier driven by the same squarewave excitation that alternatively toggles the positive and negative ADC inputs between the sensor output and a reference 1:2 resistive divider of the excitation signal. One RC 22k / 10uF common mode filtering + 1uF differential mode filtering averages the positive and negative ADC input signals.
Also the ADS1243 full scale reference voltage is obtained with an additional syncronous rectificator that detects the amplitude of the squarewave AC excitation. One RC 22k / 10uF averages the detector output that is then buffered by a zero drift precision buffer and then fed to the ADS1243 +Vref pin that has 22uF+100nF local decoupling.
I used great care in adding RC filter on the clock of the excitation generator driving the 7474 and capacitors on the 'Q' and '-Q' excitation signal outputs in order to suppress noisy ringings (overshoots and undeshoots) at the edges of the squarewave excitation. At present the excitation waveform output from the 7474 is basically free of such ringings being a trapezoid waveform with rising time of 150ns and 200ns falling time.
Despite all my cares, 150-200mV glitches are still observed on the scope both on the ADC inputs and full scale reference. Glitches are syncronous with the primary clock of the 74HC74 (they appear with a frequency twice of the actual sensor excitation squarewave that is exactly the original driving clock-, and I suppose they are related to some sort of ground coupling. The ground connection of the sensor conditioner board (a separate PCB featuring the excitation generator, the sync rectifiers and the RC filters) is tied close to the ADS 1243 ground. I tried to lower the impedance of the ground connection using a wider cable but with no change (as from the suggestion of the TI appnote SBAA052) .
At present I am not able to squeeze out more than 15 bits of ENOB from the ADS1243 that become 14 or even less over longer period (10min or so), and I believe this is due to the glitch noise since the same ADS1243 circuit reads up to 20 bits of ENOB from a DC excited full bridge sensor.
I am wondering if there can be any other trick I can try to have a less noisy reading. Thanks in advance.
