Added note on derivative-on-measurement.

If you want to use derivative-on-measurement, you will need to negate the 'D'-gain you would typically use in a 'standard derivative-on-error' PID controller - I did not mention that in the video. Since the 'error signal' effectively going into the differentiator does not depend on the setpoint: e[n] = 0 - measurement, and therefore (e[n] - e[n - 1]) = (0 - measurement) - (0 - prevMeasurement) = -Kd * (measurement - prevMeasurement).
So, for example, if you require a controller with a D-gain of 10, you would set pid->Kd = -10. This is a slight quirk of the derivative-on-measurement form - however, the code can of course be adapted to include the minus sign.

PID.h

@@ -6,7 +6,7 @@ typedef struct {
 	/* Controller gains */
 	float Kp;
 	float Ki;
-	float Kd;
+	float Kd; /* Note: since using derivative-on-measurement, Kd needs to be negative (in contrast to conventional 'derivative-on'error') */
 
 	/* Derivative low-pass filter time constant */
 	float tau;

PID_Test.c

@@ -24,7 +24,7 @@ float TestSystem_Update(float inp);
 int main()
 {
     /* Initialise PID controller */
-    PIDController pid = { PID_KP, PID_KI, PID_KD,
+    PIDController pid = { PID_KP, PID_KI, -PID_KD,
                           PID_TAU,
                           PID_LIM_MIN, PID_LIM_MAX,
                           SAMPLE_TIME_S };