Well, ...where to start! Firstly, given the posts on this thread by you and several others, I rather think I am the one who will be "corrected and steered right" the most frequently! But if I have an occasional nugget of info to contribute that is needed, relevant and accurate I'm happy!
Pure P25 Phase 1 systems should never have any analog FM voice channels as far as I know. If the systems are labeled as "P25 Phase 1" then they should be P25 FDMA CAI voice channels with a 9600 baud control channel, period. Having said that, C4FM is, technically, just FM modulated by four digital dibit symbols represented by four fixed FM deviations. So, from a technical standpoint, it is "just FM". I recall many years ago where a coworker mentioned to me about a friend of his that worked on the then new digital LMR gear who kept saying that it was "just FM"! I'm sure he was referring to the C4FM modulation used. It made sense to use C4FM as it was the easiest to implement and deal with in terms of using the well understood FM-friendly design techniques for both subscriber gear and infrastructure. It's a nice clean constant envelope modulation that behaves well under the conditions we're used to that work well with analog FM. Just like analog FM you don't need linear PA's, friendly and familiar, from a FM LMR standpoint, (and efficient) class C's work just peachy so no problem there. From a transmitter designer's standpoint, it's almost a walk in the park to dip the toe into the digital modulation waters using C4FM. And, as we have discussed, in the receiver side of things, using common FM demodulator techniques works fine too (so long as we're avoiding things like LSM, of course). Also, according to the literature, C4FM can be relatively easily accommodated alongside CQPSK which, as I understand it, was the early on determined upgrade path to the future Phase 2 two slot TDMA format. I'm a little unclear on that as my experience with QPSK was with the pi/4DQPSK TDMA US DAMPS cellular system and it needed a linear transmitter and an I/Q demodulator in the receiver. I'm thinking that either the lesser complex form of CQPSK uses a form of shaping filter that does not "modulate the amplitude" so much that you need to recover that at the receiver end OR that they were just talking about the fact that a CQPSK receiver could easily handle a C4FM transmission - which is true, of course, and makes more sense - so they were talking only about newer gear HEARING the older only and not the other way around. This, too, might bring us to the next subject.
As far as I know, LSM (in P25 Phase 1 simulcast trunking systems) uses CQPSK modulation but only in the outbound path from the control site. So, if you use that for pure FDMA Phase 1 systems, then the subscriber units do not need to make their transmitters fully linear in the PA's - old cheap standard Class C's are still acceptable. Their receivers do need I/Q demodulators but that should not impact the unit's total power consumption near as much as adding a less efficient linear or even linearized PA to the transmitter. So you get a "half and half" scenario. When we move to full Phase II systems, however, you will, as far as I know, need to have CQPSK coming from the transmitter of the subscriber units as well which means that they will need linear or linearized PA's too. But, in the last 20 years, relatively efficient linearized PA's with sufficient linearity to handle the complex modulations now used in the LMR and Cellular markets has come a very long way - it was a big problem for us back in the early 90's but I don't think that is the case anymore. I don't know if the new highly efficient linear/linearized PA designs come down to a Class C's power draw level and up to a Class C's total efficiency but I believe they are far closer now then they used to be, 15-20 years ago. Apparently, it's close enough, though, so that, given the improvements in other areas of the transceiver's design, the total power draw still comes within acceptable levels for a LMR portable unit - and then, too, battery technology has also improved. Anyway, that means that the Phase 1 simulcast (LSM) systems probably use LSM for the outbound (control to subscriber) control channels and for the voice channels (site to subscriber) using CQPSK modulation (or some variation thereof) but receive the Phase 1 subscriber units using standard C4FM modulation. This really makes sense as the subscribers certainly wouldn't "simulcast" from multiple transmitters (we're not talking 802.11n advanced MIMO techniques here, just yet). In Phase 2 systems, however, all Phase 2 TDMA users, subscribers and control sites, will use CQPSK (or whatever variation it is) and Phase 1 subscribers can still be easily accommodated within the same system as long as their receivers can easily handle CQPSK - which the newer ones obviously do. Another way to put it might be - if you can effectively handle CQPSK you can effectively handle LSM and, of course, C4FM. You can kinda see the logic of the upgrade path here:
1) First, implementing mixed mode analog FM trunking combining digital C4FM and analog FM subscribers together using standard Class C PA's and standard receiver FM demodulators.
2) Second, adding effective simulcast technology using LSM when subscriber units have all moved up to using fully I/Q demodulator based receivers which they need to do anyway for future upgrading to TDMA Phase 2 CQPSK systems (albeit in FDMA mode unless the transmitter PA is upgraded also).
3) Third, phasing out all analog FM only subscribers and subsequent implementation of pure digital P25 Phase 1 FDMA systems using receiver I/Q demodulators in subscriber units thereby allowing LSM when needed as well as a path to Phase 2 inclusion as above.
4) Fourth, implementation of full P25 Phase 2 TDMA systems with a mixed subscriber base of TDMA and FDMA units (when done as an upgrade from an older Phase 1 system) all using receiver I/Q demodulators and all or most subscriber units using linearized CQPSK capable PA's (newest units having TDMA capability while older units lacking such PA's will either be slowly phased out or upgraded in hardware when possible).
5) Finally, phasing out all FDMA only Phase 1 subscribers and subsequent implementation of pure CQPSK 2 slot TDMA Phase 2 systems with Phase 1 and analog FM capability in subscribers retained for direct off network simplex and mutual aid use when needed. LSM implemented as needed - all hardware easily supports it.
The above is simplified and is my best guess only. Also, there is much left out that I am aware of (as I said - to keep it simple). One thing that should be pointed out is that there were many different "half-way paths" in and between the steps noted above. As the standards were awaiting finalization, manufacturers, especially Motorola, made partial steps using proprietary technology that was supposed to be upgradeable to the final standard when it was finalized. They did this with their proprietary Astro C4FM modulation prior to P25 Phase 1 FDMA C4FM digital voice modulation finalization and again with their X2 TDMA systems while awaiting Phase 2 finalization. As I indicated before, the 3600 baud Control Channel Motorola mixed mode analog FM plus digital P25 Phase 1 CAI subscribers systems are proprietary systems in preparation for full P25 Phase 1 as well, providing an upgrade path for large systems already in place and having many analog FM subscribers. Hence, you can see why there is much confusion among hobbyists and even many in the industry. Such confusion can be summed up in what I call "Total Hooey Distortion" which is a result of the following: what I call "Specialization Isolation Distortion (or "Specialization Isolation Disease" or Specialization Isolation Distortion Disease" take your pick)" mixed with "Acronym Overload" mixed with what I call "Gotta Know It Now, Error Correct Later Syndrome" and finally, mixed with an all too hardy dose of "Massive Marketspeak Mangling" (aka, the infamous, "M Cubed Factor"). And you know all that mixing is a nonlinear process, right?!;-)).
You can see where the consumer scanner development got left behind in the steps 1) and 2) above, at least in terms of full handling at the receiver level. Perhaps the cellular arena, as it has done for other aspects of the transceiver chain, plus other high volume RF consumer devices, will result in cheap I/Q demodulators that the scanner folks can use effectively. Or else - well, "software I/Q" after direct A/D of the IF? I'm dumb, here, no clue, is that what's done in the biggie SDR stuff?
Regarding talkaround capability - that is a very good point, actually! Apparently, when the LAPD went to full P25 conventional (non-trunked) usage they ran into this issue. One would think that the C4FM signals would behave similarly to standard analog FM in terms of the capture effect inherent in FM signals. Apparently, this did not appear to be the case. Initially, the LAPD attempted to use unit-to-unit simplex channels in the same way they always had been doing when using analog FM - that is, they used the base/repeater output frequency as the simplex talk around channel. This had worked fine with analog FM but, apparently it played holy havoc with the C4FM stuff making it unworkable. I'm not sure I can explain why - if it were CQPSK or some other complex digital mode I could see problems but I am not sure why the C4FM had issues. Maybe something to do with the instantaneous transitions of frequencies and phases versus the more gradual and linear changes with analog voice. Anyway, they finally abandoned that practice and changed to completely separate simplex usage frequencies (interestingly, those new channels were used in both repeater and simplex mode but, in LAPD lingo, they stuck with calling them all "simplex" with "direct and repeat" option available). So you may have something there concerning why talk around is still preferred to be in analog FM mode. FM is also simple to use in all systems now given it is easily generated and received by all of the hardware as we have discussed and is a great fall back if everything else goes haywire and/or for the final "everybody's got it" mutual aid choice.
Everything you wrote about the I/Q demodulator used for FM makes sense and jives with what I know. Pretty sure you're right about the zero crossings and needing only one I or Q channel.
And I know all about the "little green men"! I went from a large corporate telecom doing digital cellular, DSMR, and mixed mode sat phone subscriber gear to a small startup company focusing on developing large scale RF integrated circuits. In my old company "systems engineering" referred to the infrastructure equipment development and implementation while in the new company everything outside the IC was "Systems Engineering"! So a subscriber unit to them would have been a "system" while it was just a basic user device to me. The folks there initially were brilliant in terms of IC design all of which was new magic to me but they lacked RF system knowledge and that I had (system in terms of proper layout on a pcb and impedance paths, coupling, decoupling on power leads, etc., etc., and simply knowing what was what and where when building a transmitter and receiver using the IC's they were designing). Learned a lot about the IC magic there - quite an eye opener for this "systems guy"!
Man this post got long - hope it was helpful and meaningful!
-Mike
