We show that old isolated neutron stars in groups and clusters of galaxies experiencing a Quark-Nova phase (QN: an explosive transition to a quark star) may be the sources of FRBs. Each fragment ("chunk") of the ultra-relativistic QN ejecta provides a collisionless plasma for which the ambient medium (galactic/halo, the intra-group/intra-cluster medium) acts as a relativistic plasma beam. Plasma instabilities (the Buneman and the Buneman-induced thermal Weibel instabilities, successively) are induced by the beam in the chunk. These generate particle bunching and observed coherent emission at GHz frequency with a corresponding fluence in the Jy ms range. The duration (from micro-seconds to hundreds of milli-seconds), repeats (on timescales of minutes to months), frequency drift and the high occurrence rate of FRBs (a few per thousand years per galaxy) in our model are in good agreement with observed properties of non-repeating and repeating FRBs. All FRBs intrinsically repeat in our model and non-repetition (i.e. the non detection of the fainter QN chunks) is detector-dependent and an artifact of the bandwidth and of the fluence sensitivity threshold. Key properties of FRB 121102 (its years of activity) and of FRB 180916.J0158+65 (its ∼ 16 day period) are recovered in our model. We demonstrate that FRBs can be caused by a cataclysmic event (the QN) and that the observed FRB rate can be explained without the need for FRB sources to repeat over their lifetimes. We give specific predictions, notably: (i) because of the viewing angle (Doppler) effect, sub-GHz detectors (CHIME) will be associated with dimmer and longer duration FRBs than GHz detectors (e.g. Parkes and ASKAP); (ii) CHIME should detect on average ∼ 5 times more FRBs from a given QN than ASKAP and Parkes; (iii) super FRBs (i.e. tens of thousands of Jy ms fluence) should be associated with intra-cluster medium QNe; (iii) monster FRBs (i.e. millions of Jy ms fluence) associated with inter-galactic medium QNe might plausibly occur with frequencies at the lower limit of the LOFAR's low-band antenna.