I challenge you prove that you cannot mix narrow with wide band radios. We are doing it right now, there was no way to reprogram over 2000 in one shot, so after considerable testing we found that all of our analog radios WILL work mixed with narrow and wide transmissions. They are a mixture of GE Orions and P7100's, Motorola Spectra's and Mitreks, CDM 1250, HT-1250 and Vertex 9000's.
Figure this, the average bandwidth of a wide band receiver is about 7kc, anything narrower and you would get chopping on a loud 5kc transmission. When the radio is programmed narrow, the average bandwidth is almost 4kc, so unless the radio operator is screaming at the top of his lungs, there isn't a problem. Digital radios are another animal, they convert the incoming signal to digital, then decide which path to take to decode. Havn't had much luck with getting them to work consistently in mixed mode, but you can program them wide RX and lower the TX deviation to work.
BTW we do have voting, so unless the voters are set up wrong they should vote on the quietest signal, not the loudest transmission. There are very few simulcast systems out there below 800mhz, and they are not mandated to go narrow.
No one said it was going to be easy, but throwing more down for unbuilt and untested radio systems is foolish. I will wait and let the problems be solved by someone else with more money. I am not convinced that 700 mhz is the cure all that everybody hopes it is. Maybe in 20 years, but not now.
By the way you speak I would suspect that you are either a salesman for a radio manufacturer or from an agency that wants the latest and greatest and spare no expense. To hell with education, I want my MTV.
For the sake of disclosure, I worked for two manufacturers - one 20 years ago, the other 12 years ago - never in sales! Both have changed radically and I have no allegiances to either of the "new" companies (maybe the opposite). I recently retired from doing communications engineering for a public safety agency. Latest and greatest includes Midland Syntec mobiles with some new Astro Spectras, XTS-3000, XTS-2500 radios, simulcast Quantars, and some stand-alone MASTR-III bases, with still a bunch of MSR, MSF, and Micor stations. I like radio better as a hobby. I've also got 3 decades as a "responder," now watching my kids volunteering while the geezer rolls over and goes back to sleep.
MTV? Bleh. WNEW-FM, WDHA-FM, or WMMR. Preferably 1979. With headphones, no less.
Anyway, back to narrowbanding. Glad yours is working out, but my experiences were different. The passbands were a lot tighter for narrowband (~3.5 kHz before distortion). At best case, there will be volume differences. At worst case, your wideband radios will pop out of the receiver passband of the narrowband radios. Narrowband signaling deviation might not consistently decode on wideband radios. Some systems may have unique audio characteristics going across remoted lines and alert paging may be inconsistent. I'm not saying it SHOULDN'T be done, but if you do that, you may run into snags. Those snags may be safety issues. Especially when screaming (but at least it's not wind noise in P25).
I did mine in two steps. Program narrowbands into one bank, keep widebands in another. Change less-used channels' infrastructure, switch over, then change most common channels. Then take out the widebands (couldn't do that because all of the mutual aid channels were still WB).
Be careful with compandoring. Each manufacturer has a different scheme. Using a Kenwood and Motorola both with compandoring enabled will result in distortion. Compandoring also changes audio characteristics, so going through a comparator might yield soft, mushy audio below compression and then booming audio above, with inconsistent transitions. My 'spatchers only stopped complaining about funky sound after turning off any compandoring.
I ran the Astro Spectras, XTS, and Quantars (and one EFJ mobile) in mixed mode with the P25 mode strapped to encryption. I also had several same-frequency repeaters with different input NACs and PL/DPL codes, but the same output (they wouldn't budget me for an AstroTAC, master oscillators, or more connectivity). What are you trying to do? PM me if if you want to exchange notes.