This theoretical study describes a new phenomenon which occurs in mechanical interactions between a macroscopic stmcture (the plate) and microscopic ones (the nanopillars). Let us consider a square a1rny of nanopillars comprising a huge number of them : 108, 1010, 1012, or more. Their bases lie on a plate (a substrate) such as a silicon wafer (fig. 1), their tops are free to vibrate. Nanopillars are tiny resonators, and their under-consideration vibration is a bulk wave one, directed along their axis, called "length mode". The time histo1y of tis vibration looks like a sine wave of slowly vaiying amplitude and phase [ 4]. In this anay, the nanopillars vibrate at the same frequency, but with random phases. There are many ways to set in phase those pillai·s, and it is assumed that at time t1 the whole aiTay does vibrate in phase. But later, at any time t2, this phase coherence will be lost, because of the varying phase of Brownian motion. Now, the question is to maintain one phase throughout the anay. For this purpose, the plate is taken to be a resonator, with a thickness compression mode, a haimonic of which has the same frequency as the pillar length mode. Then, as soon as the pillai·s vibrate in phase, they will induce a strong resonance of the plate. In its tum, the plate, by its vibrating big mass, will impose to the small pillars one single phase. For this resonance to happen, there ai·e two main requirements : a high accuracy, first in the tuning of the pillar frequencies, and secondly in the ratio pillai· height / plate thickness, so that the two vibration modes fit together perfectly.