Ok to make its slightly cleare. three are severral tings we have been discussing in this thread. first is the frequency seperation of the channel centers in the table of channel assignements. This is also known as channel step. ie 155.01, 155.025 155.04. This channel step affects only part 90 users (for this discussion purposes) it does not affect part 97 users (exception: voluntary band plans). Most public safety "channels" are 20Khz wide, which is different than channel step. a 20Khz channel has and emmision designator of 20K0F3E for FM modulated analog voice (most common) There are two portions of the emmission designator 20K0 and F3E. 20K0 refers to the bandwith occupied of the emmission. 20 (whole number) K (multiplier) this case 1000, 0 (decimal). F3E denotes the type of transmission F (Frequency Modulated) 3 (single analog channel) E (Telephony, Voice). You will notice that the wideband channels will overlap in some bands.
Under the narrowband plan channel step is 12.5 Khz emissions designators are 11K2 or 11k3 or 11K0 typically. This actually does leave some guard space between channels.
The audio deviation for a wideband channel is roughly 4.5Khz for a narrowband is 2.5 kHz.
So what happens when you are programming the radio? First you can enter any frequncy requardless of channel step. So some of the frequencies i mentioned in my first post, that do not exist, could be entered into the software. the software also has a dropdown to select channel width. If you select the 12.5Khz designation, the radio will use the narrow front end filters and have a lower transmit audio deviation. If you were to mix modes what you would notice is lower audio volume from the narrowband signal into the wide band recieve (assuming no AGC), the narrow recieve of a wideband singal may have some clipping, the amount of clipping may be insignificant or very noticeable depending on the praticular speach characteristics.
Hope this helps to clear up some confusion, and i hope i didnt confuse anyone more.