Résumé
Models which address both the Hubble and $S_8$ tensions with the same mechanism generically cause a pre-recombination suppression of the small scale matter power spectrum. Here we focus on two such models. Both models introduce a self-interacting dark radiation fluid scattering with dark matter, which has a step in its abundance around some transition redshift. In one model, the interaction is weak and with all of the dark matter whereas in the other it is strong but with only a fraction of the dark matter. The weakly interacting case is able to address both tensions simultaneously and provide a good fit to a the Planck measurements of the cosmic microwave background (CMB), the Pantheon Type Ia supernovae, and a combination of low and high redshift baryon acoustic oscillation data, whereas the strongly interacting model cannot significantly ease both tensions simultaneously. The addition of high-resolution cosmic microwave background (CMB) measurements (ACT DR4 and SPT-3G) slightly limits both model's ability to address the Hubble tension. The use of the effective field theory of large-scale structures analysis of BOSS DR12 LRG and eBOSS DR16 QSO data additionally limits their ability to address the $S_8$ tension. We explore how these models respond to these data sets in detail in order to draw general conclusions about what is required for a mechanism to address both tensions. We find that in order to fit the CMB data the time dependence of the suppression of the matter power spectrum plays a central role.