We simulated a large scale spiking neural network characterized by an initial developmental phase featuring cell death driven by excessive firing rate, followed by the spike timing dependent synaptic plasticity (STDP) in presence of Poissonian noise. The analysis of the effective spike trains recorded throughout the whole network reveled appearance of precise firing sequences that suggests a self organization of initially randomly connected networks. The firing dynamic of initially identical network topologies was analyzed and compared in three cases: in presence of temporally structured stimuli, in absence of any stimulation and for coupled networks, where the output spike trains of selected (efferent) neurons of a neuronal network correspond to the the input to selected (afferent) neurons of another similar network. The analysis of recurrent precise firing sequences detected in these three cases showed differences in their structure and dynamics; in particular, the precise firing sequences detected on the downstream network are characterized by a more complex distribution of their internal time structure.