loop filter design in CT SDM question

[surZ90]

I want to built Sigma Delta Modulator and I have some question about loop filter in SDM.
SDM comprises a quantizer and loop filter (H) before quantizer.
then subtraction output signal (Y) and input signal (R) to be input for loop filter.

so, the filtered error (E) will be:
E = H * (R - Y)
then it can rewrite to be noise-shaping equation:
Y = R + E * H^(-1)

what is consideration design of filter H?
does it just need to H and H^(-1) to be stable?
how if I use Butterworth or Chebyshev or inverse Chebyshev filter?

thanks.

[surZ90]

I use this SDM for my class D Amp project.
Can anybody help me?