There are many rules governing how the DSPs are allocated. One rule applicable here is that conf and voice termination (the FXO and T1/E1 ports) cannot share DSPs, conferencing on the PVDM2s require a dedicated DSP (this is not true for the PVDN3s where they can share). Further, the number of services/channels per DSP is a mathematical calculation of how much "resources" each DSP has and how much resources each call type (voice termination, xcoding, conf of a particular size and codec, encrypted calls etc.) requires.
In your particular case, assuming all G.711, you voice termination channels add up to 32 (8 analog and 24, assuming T1, digital). This by itself requires 2 DSPs, which is a PVDM2-32 with no spare capacity. For your 2 conferences you now need another DSP, so you need to go up to a 3 DSP PVDM, the first of which is the PVDM2-48. You could add more confs to this (up to 8) and the DSP result will remain the same, as the "conf DSP" has spare capacity.
However, adding more voice termination will change the result. If you add 4 more analog ports, you now have 36 voice termination channels, which by itself, requires 3 DSPs, i.e. a PVDM2-48, with spare capacity. Adding conf'ing to this (even a single conf), requires a 4th DSP (as conf cannot share the spare capacity on the partially used 3rd DSP allocated to voice termination), which is a PVDM2-64. On this config you have spare voice termination and conf capacity so adding up to 28 more G.711 voice channels, and/or up to 8 total G.711 confs, will remain PVDM2-64 before the next jump occurs.
CH
