Vignéras, in her Arithmetics of quaternion algebras, begin chapter II.4 recalling that we know the number of integer ideals of fixed norm of a quaternion algebra $H$ over a local field $K$, ramified or not (lemma 4.1). What is of interest here is that there are $1+q+\cdots +q^n$ such ideal of fixed norm $q^n$ if $H \cong M(2,K)$. That make possible an expression of $\zeta_H$ in terms of $\zeta_K$.
Then, it seems that Vignéras assume $n$ is always even, writing :
$$\zeta_H(s) = \sum_n \sum_{d \leqslant n} q^{d-2ns} = \sum_{d,d'} q^{d-2(d+d')s} = \zeta_K(2s)\zeta_K(2s-1)$$
And here I am totally lost : why do not write
$$\zeta_H(s) = \sum_n \sum_{d \leqslant n} q^{d-ns} = \sum_{d,d'} q^{d-(d+d')s} = \zeta_K(s)\zeta_K(s-1) \quad ?$$
Hence, is $n$ even ? Even if it is, what about the range of $d$, going to $n$ and not to $2n$ ?
Thanks for your help !