Numerically, I have found that the following formula seems to be independent of $N$ for any choice of $a$ and $b$ at large $N$: $$\sum_{k=1}^N \frac{\sin^2\phi_k}{N(a^2+b^2-2ab\cos\phi_k)^2}$$ with $\phi_k = \frac{2\pi k}{N}$. This seems somehow related to the cosine and sine formulas (at least for positive $a$ and $b$); however, I have not managed to find an analytical proof so far for this claim. I know this is true for extreme cases (such as $a\ll b$), but I would like a proof for the general case.
Could anyone help me solve this issue, or at least point me in the good direction?