The Museum of HP Calculators

HP Forum Archive 14

[ Return to Index | Top of Index ]

Spice series extra RAM
Message #1 Posted by Eric Smith on 8 Apr 2004, 10:03 p.m.

Based on microcode simulations, it appears that some of the Spice series calculators have much more RAM implemented in the hardware than is actually made available to the user. It's not uncommon for a few extra registers to be used for various internal purposes. But the HP-33C appears to have 32 hardware registers, of which 8 are available as user registers and 7 provide the 49 program steps. One is probably Last X, but even if there are a few used for internal status, there are still quite a few unaccounted for.

The register count of 32 was determined by observing the cold-reset memory initialization sequence, which appears to check values in registers 08 and 14 (hex), then zeros all registers from 1F down to 00, then writes initialization constants to registers 08 and 14.

The Spice series, like later Woodstock and Topcat models, uses chips that combine ROM and RAM. Thus the amount of RAM in a Spice is to some extent determined by the ROM size. Possibly the amount of user memory made available was restricted for product positioning purposes.

Simulations of the HP-32E and HP-37E seem to indicate that these models have 32 and 48 registers, respectively. However, this is less conclusive because the initialization in these models is performed by the use of a "clear data registers" instruction that clears an entire 16-register chip in one cycle.

The HP-32E at startup selects and clears the RAM chips at address 10 hex then 00 hex. The HP-37E selects and clears 10, 20, then 00.

      
Re: Spice series extra RAM
Message #2 Posted by Wayne Stephens on 9 Apr 2004, 1:51 p.m.,
in response to message #1 by Eric Smith

Is there any way to make it usable? Also, does this apply to the 38E as well?

Take care.

Wayne.

            
Re: Spice series extra RAM
Message #3 Posted by Eric Smith on 9 Apr 2004, 2:58 p.m.,
in response to message #2 by Wayne Stephens

Quote:
Is there any way to make it usable?

Only by hacking the microcode. This would take a pretty substantial amount of word, both in making the microcode changes, and in building and installing a suitable replacement for the ROM/RAM chip.

Quote:
Also, does this apply to the 38E as well?

I'm not yet certain. The 34C, 38E, and 38C have more ROM than the processor can directly address (7K, 5K, and 5K, respectively). I need to implement ROM bank switching in the simulator before I'll be able to log what memory locations the microcode initializes.

      
Re: Spice series extra RAM
Message #4 Posted by Raymond Del Tondo on 9 Apr 2004, 3:37 p.m.,
in response to message #1 by Eric Smith

Hi,

did you count in the RPN stack (X,Y,Z,T) ?

Regards,

Raymond

            
Re: Spice series extra RAM
Message #5 Posted by Eric Smith on 9 Apr 2004, 4:51 p.m.,
in response to message #4 by Raymond Del Tondo

Quote:
did you count in the RPN stack (X,Y,Z,T) ?

No, because those are part of the processor. Last X is normally stored in RAM, though.

The Nut processor, as used in the HP-41C and Voyager families, was the first processor HP designed for handhelds calculators that did not have a hardware RPN stack.

                  
Re: Spice series extra RAM
Message #6 Posted by Wlodek Mier-Jedrzejowicz on 9 Apr 2004, 10:30 p.m.,
in response to message #5 by Eric Smith

Eric, since you have already mentioned the HP-41 family in this context, maybe I should add a reminder that the HP-41C microcode had instructions to clear all addressable RAM memory (if it assumed memory was corrupt because the cold-start-constant was incorrect). Yet only an HP-41 with a full complement of Extended Memory really had all that RAM. (Apart from Jim de Arras' HP-41C to which he added memory himself...) Maybe the Spice calculators played the same trick of clearing more memory registers than actually existed :-) Regards, Wlodek


[ Return to Index | Top of Index ]

Go back to the main exhibit hall