I recently read an article by Stephen Woodward discussing different implementations of 2nd order low pass filters here: https://www.edn.com/cancel-pwm-dac-ripple-with-analog-subtraction-revisited/
In these designs an inverted PWM signal (eg. A microcontroller’s complementary output) is used to cancel some ripple
Top left is a typical single pole RC low-pass filter:
-high ripple
+short settling time
Bottom left is a double pole RC low-pass filter:
+low ripple
-longer settling time
Top right is a double pole RC low-pass filter (Woodward’s design):
*same effect as the typical double pole
+low ripple
-longer settling time
Bottom right is another double pole RC low-pass filter (Woodward’s design again):
*inbetween results of these other filters
+low ripple (slightly worse than double pole)
+short settling time
