SBAA497B May 2021 – April 2022 PCM3120-Q1 , PCM5120-Q1 , PCM6120-Q1 , TLV320ADC3120 , TLV320ADC5120 , TLV320ADC6120
PurePath, Burr-Brown, are trademarks of Texas Instruments.
MATLAB is a registered trademark of The Mathworks Inc.
All trademarks are the property of their respective owners.
Each channel of the TLV320ADCx120 and PCMx120-Q1 devices follows the signal chain shown in Figure 1-1. In this signal chain, each channel supports an analog differential or single-ended signal or a digital pulse density modulation (PDM) digital microphone. In TLV320ADCx120 and PCMx120-Q1 device families, the analog input signal is amplified by a Programmable Gain Amplifier (PGA) and then converted by a high-performance ADC into a digital signal. The PGA gains the input signal to match the full scale of the ADC. The digital signal has a programmable phase calibration to adjust the phase delay of each channel in steps of one modulator clock cycle. This allows the system to match the phase across different channels. The phase-calibrated digital signal is then decimated through a set of linear phase filters or low-latency filters. DC offset is removed from the decimated signal through a Digital High-Pass Filter (HPF) with three preset cutoff frequencies or a fully programmable cutoff frequency. Note that DC shifts are caused by mismatches in common-mode voltages. The output of the HPF is gain calibrated with 0.1-dB steps and summed with other channels. The gain calibration matches the gain across different channels, particularly if the channels have microphones with varying gain values. The output is then filtered by the Digital Biquad Filters and gained by the volume control.
This application note concentrates on how to configure the Digital Biquad Filters. The Digital Biquad Filters are digitally implemented as a set of IIR filters.
Equation 1 specifies the transfer function of infinite impulse response filters (IIR).
When the coefficients of this transfer function are quantized for fixed point implementations, the resulting errors due to quantization and the recursive nature of the filter can significantly alter the desired filter characteristics and lead to instability. Partitioning this transfer function into a set of cascaded lower-order filters reduces the sensitivity to coefficient quantization. Cascaded Biquad IIR filter implementations have been proven to be effective in minimizing these effects.