As far as I know, the 500/600 does not have a narowband crystal filter in the IF. They just compensate for lower audio recovery when you switch to NFM. I'd be glad to pay for a narrower bandwidth and gain a better signal to noise ratio and have some good sounding NFM audio.
It's been about 40 years since the Regency TMR-8H scanner came out and you could buy a narrower Murata IF filter for those. Replace the filter and the 2 chokes that came with the kit and you had a much better scanner, much less prone to interference. You could install the same kit into a Regency TMR-8U, but the cheap channel crystals drifted so much that even with stock AFC couldn't keep it close enough to frequency so I switched back to the stock Murata IF filter.
Incorrect, I have verified to my satisfaction (I have a background in RF electronics testing and design) that the GRE 500/600 series and the newer 310/410 series DO switch in a narrower IF filter when the NFM mode is selected. They, however, DO NOT compensate for the audio change when +/-2.5KHz deviation signals are detected. You must use the "audio boost" feature to manually compensate for the lower "loudness" of the narrower FM deviation. This is completely the opposite of what Uniden does with their BCT15 scanner and, I think, the rest of their current series (I only have the BCT15 to test myself). In Uniden's case, they do compensate for the narrower +/-2.5KHz deviation audio when in "narrow" mode but DO NOT switch in a narrower IF filter. Here, where I am, the CDF uses the newer 7.5KHz channels in the VHF-High band quite extensively. I can hear at least three different districts where I am and when multiple fire in multiple districts are being handled the Uniden is pretty much useless due to extensive IF adjacent channel interference (IF "bleedover") while the GRE units handle it much better when the NFM mode is selected.
Each brand, Uniden and GRE, seem to handle half the solution to the newer "narrowband" FM signals reception. Uniden deals with the audio compensation (either through adjustment of the FM discriminator circuitry or through post processing adjustment) but completely ignores the IF filter (just uses the same IF filter as the "normal" narrowband FM filter) while GRE does the opposite. Ideally, both sides should be addressed, of course, but, in my experience, where I am, if I am forced to choose one or the other, I'll take the tighter GRE NFM IF filter as it at least makes the reception usable.
I do still prefer the Uniden's front end design, however. Seems to be significantly better front end filtering and/or dynamic range than GRE's very poor design. Frankly, I seriously wish I could "Frankenstien" the Uniden front end with the GRE's back end IF section and add the Uniden's discriminator audio approach and make a decent final model in terms of the RF signal handling.
-Mike