This design provides current-to-voltage conversion from a current-output audio DAC into a voltage-input, class-D amplifier. The order of design priorities are as follows:
- Select feedback-resistor values based on the gain required from the current-output stage to the voltage-input stage. For this design, the full-scale, peak-to-peak output current of the PCM1795 ( IOUTL/R+ – IOUTL/R–) is ±4 mA. A gain of 1k gives a wide voltage swing of ±4 V, allowing for high SNR without exceeding the input voltage limit of the TPA3251.
- After the gain is fixed, select the output common-mode voltage. The output common-mode voltage determines the input common-mode voltage in this configuration. To set the nominal output voltage of the PCM1795 to 0 V (which corresponds to the input common mode voltage of the OPA1637), shift the output negatively from the desired common-mode input voltage by the gain multiplied by the dc center current value of the PCM1795 (3.5 mA). In this case, –3.5 V satisfies the design goal.
- A bypass capacitor from the VOCM pin to ground must be selected to filter noise from the voltage divider. The capacitor selection is determined by balancing the startup time of the system with the output common-mode noise. A higher capacitance gives a lower frequency filter cutoff on the VOCM pin, thus giving lower noise performance, but also slows down the initial startup time of the circuit as a result of the RC delay from the resistor divider in combination with the filter capacitor.
- Select CF so that the desired bandwidth of the active filter is achieved. The 3-dB frequency is determined by the reciprocal of the product of RF and CF.
- Use a passive filter on the output to increase noise filtering beyond the desired bandwidth. The passive filter formed by RD1,2 and CDF adds an additional real pole to the filter response. If the pole is designed at the same frequency as the active filter pole, the overall 3-dB frequency shifts to a lower frequency value, and the step response is overdamped. A trade-off must be made to give optimal transient response versus increased filter attenuation at higher frequencies. For this design, the second pole is set to 106 kHz.