I hope you guys don't mind me jumping in here but I have been working with the land mobile delay stuff for almost 20 years in both celllular and PS. I'll start with a few tidbits on cellular:
Equalizers were first developed for GSM in the early 1990's to combat geographic multipath of the same signal; i.e., a direct path signal and a delayed version of that signal bouncing off of the Alps, for example. With a 5+ usec symbol period, it only tok a 1/2 mile or less of relative delay to cause issues. The GSM equalizers would combat up to 16 usec (3 symbol periods) of multipath delay in the lab and maybe 12 usec in field conditions.
TDMA in the US followed and needed equalizers for the same reason. The tolerable overlap delay was around 40 usec as I recall, or a bit over one symbol period.
No cellular system technology has ever simulcasted the same signal on the same frequency from multiple sites. In the cell world for AMPS, TDMA, and G1 GSM, the control channels and voice channels were all on different frequencies. So no cellular system ever needed or used equalizers to fight simulcast delays from different sites.
And all cell sites overlap with other cell sites so some overlap was designed in; without it you dropped the call between sites. So the idea that this is not a cell issue due to limited overlap is not true.
(And we can talk about CDMA and UMTS and why they DON'T need equalizers but I'll not start that now.)
And, yes G2 GSM went to frequency hopping as stated. This was done because all the fixed frequency plans per site were becoming somewhere between he** and impossilbe to manage and implement when the site densities got high in the 2000 time frame, and with fast hopping, the occasional time slot collisions between different calls are corrected by link error correction and the vocoders. TDMA was phased out long before they ever got that far.
Now for PS simulcast. For older analog systems the simulcast voice channel growl referred to is caused by small carrier frequency tolerance mismatches between overlapping site radios. This is due to how FM demodulators work and gets louder in the lower frequency ranges, in proportion to 1/f. Think of it as a beat note.
For the analog FM simulcast control channels, there were never any equalizers in the equipment. This problem was managed by making sure that the overlap delay between sites never exceeded 35-40% of the modulation symbol time. The older EDACS for example had a symbol time of apporx 100 usec, and the system site designs and time adjustment between sites was limied to 35 usec and maybe 40 usec of relative overlap delay between adjacent sites in a pinch. With this limited overlap delay, >60% of the symbol time was undistorted and the detection/decision process was reliable.
And PS simulcast sites have been HIGHLY time synchronized from the get-go. In the old days, there were precise time references in each site and landline links between sites to keep the times locked together. GPS has replaced that, but the precise time sync between all sites in a system has always been there in PS simulcast. It is the very essence of making simulcast control channels work, and the setup of the sites was partly to set this time sync to minimize the overlap delay between all site adjacencies.
BTW, This works great in places like FL, IL, and KS< but has some issues in the low mountain areas like VA, as the signals often go where they are not expected; i.e., the simulcast zones overlap s get really messy in hilly areas, and you can't control all overlap zones to the time limit as described above.
With phase P25 1 control channels, the same basic thing is still being applied as far as I know: since the symbol times are around 200 usec on the control channels, an equalizer is not what is used to fight this. You just keep the overlap delays limited, so that the simullcast symbols never overlap enough to casue issues. And for Phase 2, all I can find says that the same tactic is being used; the symbol times are around 160 usec on both control channels and voice channels. I am in "intense P25 delay learn mode" these days, and have yet to run across a qualification test for P25 equipment that requires time overlap delay measurements on base or subscriber units that would indicate any actual equalizer is being used like in GSM or cell TDMA. But I am going to keep digging.
The key to using equalizers or not is how long are the modulation symbol times in the system. GSM is around 5 usec; cell TDMA symbol time was 20 usec if I recall right. The physics of the world mean you DO need equalizers just to manage multipath versions of the same non-simulcast signal bouncing off of mountains with short symbol times. With symbol times of 100-200 usec in PS systems, you don't need to worry about this, you only worry about simulcast-created 'multipath' of 2 separate signals with the same info. In that case, managing the relative overlap delay between simulcast signals to a maximum of 35-40% of symbol time is the trick.
Hope this helps.
Mark B.