I don't know your Buck Converter model. However, in this example, the filter divisor N is 751. Generally, the value for N is high. Since you asked about the value for N, then I guess you designed yours using the ideal form:
If your theoretical ideal 
gives a good performance on paper, then you should make this term 
in the denominator of the derivative filter relatively small, 
because
gives a good performance on paper, then you should make this term in the denominator of the derivative filter relatively small, because
% Buck Converter Transfer Function
V = 100;
C = 1e-6;
L = 2.2e-3;
R = 500;
Gp = tf(V/(C*L), [1 1/(R*C) 1/(C*L)])
margin(Gp) % Stable but Phase margin is very small, oscillatory response is expected
step(Gp) % as confirmed by the step response
% standard-form PID with 1st-order derivative filter
Kp = 0.03; % proportional gain
Ti = 3.6e-6; % integral time
Td = 0.0006; % derivative time
N = 751; % filter divisor
Gc = pidstd(Kp, Ti, Td, N)
margin(Gc*Gp) % Phase margin is improved to approximately 60 deg
% closed-loop system and check if no RHP poles
Gcl = zpk(feedback(Gc*Gp, 1))
tau = 4e-5;
[u, t] = gensig('square', tau, 2*tau, tau/2^10);
lsim(Gcl, u, t), ylim([-1 2]), grid on
S = stepinfo(Gcl)
