The HOLMES experiment aims to directly measure the ν mass studying the ¹⁶³Ho electron capture decay spectrum developing arrays of TES-based microcalorimeters implanted with O(300 Bq/detector) Ho atoms. The embedding of the source inside detectors is a crucial step of the experiment. Because the ¹⁶³Ho production process (neutron irradiation of a ¹⁶²Er sample) is not perfectly free from impurities, Ho source must be separated from a lot of contaminants. A chemical processing removes every species other than Ho, but it is not sufficient to remove all isotope-related background sources: Indeed, ^166mHo beta decay can produce fake signal in the region of interest. For this reason, a dedicated implantation system was set up. It is designed to achieve the separation power better than 5σ at 163/166 a.m.u. allowing an efficient Ho ions implantation inside microcalorimeter absorbers. Its main components are a 50 kV sputter-based ion source, a magnetic dipole and a target chamber. A specially designed co-evaporation system was designed to “grow” the gold microcalorimeter absorber during the implantation process, increasing the maximum achievable activity which can be implanted. The machine performances were evaluated by means of calibration runs using ⁶³Cu/⁶⁵Cu and Mo beams. A special care was given to the study of the more effective way to populate source plasma with Ho ions obtained from different Ho compounds by sputtering process. In this work, the machine development and commissioning are described.