During the last years it has become clear that global O(N) defects and U(1) cosmic strings do not lead to the pronounced first acoustic peak in the power spectrum of anisotropies of the cosmic microwave background which has recently been observed to high accuracy. Inflationary models cannot easily accommodate the low second peak indicated by the data. Here we construct causal scaling seed models which reproduce the first and second peak. Future, more precise CMB anisotropy and polarization experiments will however be able to distinguish them from the ordinary adiabatic models.