StarFox itself uses the glop top Mario Chip in earlier versions, later versions report it to use GSU-1. StarFox doesn't have a clock-gen on it (usually), which means it gets its clock from the SNES itself. 21Mhz according to most things I've read.
Actually, the clock pins super easy to identify with the GSU-1 models, and maybe with the gloptop. The GSU-1 accepts a master 21Mhz signal from the SNES, and a second 10/21/whatever Mhz clock for it's internal purposes. Overclocking StarFox with a GSU-1 would take a matter of 5 minutes. Just create a simple astable oscillator at whatever frequency you'd like to try and connect it to the internal clock pin.
I just built the StarFox2 cart from SNESdev (pics at http://www.liquidninjas.com/bbs/showthread.php?t=13388, sorry if cross site linking is not allowed). It works great, but as my posts suggest, slowly
There are probably people who know a lot more than me about this SFX chip situation. The emulator crowd. Information is very sparse tho in public circles. On Stunt Race FX, the internal clock is given from that 21Mhz crystal, and I am almost convinced that the clock generating circuit is splitting that in half to 10.7mhz. There are two Inverters in use on hex inverter chip on the clock gen sections of these carts. Each pulse from the xtal at 21mhz is flip-flopping the inverters, so im sort of thinking that the pulse outputted is 10.7.
Whether the SuperFX has a clock splitter built in I cant confirm. The patents never mention it, but most emu docs/people suggest it. Why generate a 10.7Mhz clock on the cart then?
When I get my new ceramic xtals in, Im going to try a 40-42Mhz xtal on the clock gen of Stunt Race FX which I suspect to be v1.0 (SFX chip 9409). I beleive my other Stunt Race FX cart which I broke is v1.1 (SFX chip 9416). Im assuming that if the clock generator, or the SFX itself, splits the clock, then a 40mhz pulse should put me at 21Mhz, which may work on the superFX chip if the specs are right.
Or, SuperFX just isnt stable at 21mhz. I dont really know, its all assumptions and I really need to test.
