The bizarre Casio FX-840P

[Jeff_Birt]

I picked up a Casio FX-840P recently and noticed it has the same PCB as the FX-850P and FX-880P. The FX-840P and FX-850P come with 8K internal RAM and the FX-880p 32K. For the FX-850P you can swap the 4464 SRAM chip for a 62256 and change one solder jumper. Oddly this does not work on the FX-840P! The FX-840P will see and RP-8 and RP-33 option RAM packs though.

I can POKE and PEEK into the full 64K address range with a 32K RAM installed internally and an RP-33 Redux installed. It would seem then that the initialization function in the ROM does not even look for the extra internal RAM. The size of the ROM in the FX-840P is smaller than the FX-850P/FX-880P so perhaps they left that out to save a few bytes?

I suspect if someone understood where the RAM pointers lived in the Casio memory map it would be possible to POKE those in by hand and let BASIC see the entire amount of RAM available.

[BruceH]

(06-22-2022 01:01 AM)Jeff_Birt Wrote:  I suspect if someone understood where the RAM pointers lived in the Casio memory map it would be possible to POKE those in by hand and let BASIC see the entire amount of RAM available.

If the calc doesn't bother to check for more RAM at power-up, why would there be memory allocated for RAM pointers? The values could just as easily be held in ROM.

[Jeff_Birt]

(06-22-2022 10:14 PM)BruceH Wrote:  If the calc doesn't bother to check for more RAM at power-up, why would there be memory allocated for RAM pointers? The values could just as easily be held in ROM.

It does check for the presence of the external RP-33, 32K RAM module. When installed it finds this and the amount of free RAM displayed is adjusted accordingly. It just does not seem to discover the additional 24K of RAM internally.

I played with it more last need PEEKing the values at $1000 intervals, increasing by 1, and POKEing back then PEEKing again to make sure the values changed. The FX-850P page on Wikipedia says it will reflect $0000-$1FFF to $A000-$BFFF, $C000-$DFFF, $E000-$FFFF without a full compliment of RAM, but I see no evidence of that on the FX-840P with the internal 32K and external 32K installed. I'll have to pull the RP-33 module and try the tests again.

[Jeff_Birt]

I found a bit more information on the FX-840P just now. It seems it is a sister to the FX-860P which did come with 32K internal RAM! These seem to share the same PCB as the FX-850P/FX-880P and an earlier revision of the HD62002 processor. The FX-860P does seem to have a different ROM, i.e. more built in functions so maybe it all comes down to the ROM?