Home Page Forums MAME MAME TMS-09xx/1xxx thread (was New Dumps)

I played around with the S2152 some more, and I was not able to get it to change banks while in "verify internal ROM contents" mode. So it looks like "override bank 0 with external program" mode will be needed to get those extra 512 bytes. I haven't found any schematics showing an external ROM connected to an S2xxx, so I'm not sure if an address latch is needed.

I got a 1978 Waddington's 2001 The Game Machine. It has a 3870, so I should be able to dump the ROM: MOSTEK 3870 MK14154N Malaysia 2001 7909

That is kind of surprising... Why would anyone care now? And, I have to admit, I'm more surprised that he actually still _has_ any of it... Was that Denker that you contacted?

I know George Klose personally built and programmed the Auto Race prototype (the first one, he was the one that essentially invented electronic handheld gaming with the creation of this. And he had a hard time convincing Mattel to go forward with it!) After Auto Race was successful, they wanted to do a Football game, and apparently Klose had less to do with that one (thus maybe it was someone else that build and programmed that one, although obviously Klose worked on it as his hand writing is on the paper)... Klose died many years ago, so I know you weren't talking to him.

I think Mark Lesser still has a hand-made prototype of Football... I have a picture of it somewhere, built closely to the size and button layout of the final game, but using huge buttons like they took them out of a computer keyboard from the day... I think this one was more after the programming and layout had been determined though, so this code in the patent was likely in an earlier period of the development...

Hmm... I know a couple of them. I didn't think they had access to any of the handheld stuff since it wasn't Intellivision related. I may have to have a chat with them about that.

A couple of the early guys were hired for handhelds and moved on to Intellivision.

Rik- yeah, it was John Denker. I wish I had asked him more questions when I first emailed him, but he did say that he designed the game and designed, built and programmed the prototype hardware. The code in the patent wrapper was compiled on his account at Caltech and has vestiges of Auto Race in it, so I assumed he also wrote that. He mentioned copyright in his reply to me; I'm not sure if he was really worried about that, or just didn't want to dig up the old files!

I started looking at the National Semiconductor MM5799 again. That's the 4-bitter used in Mattel's Basketball, Hockey and Soccer games, as well as National's own Quiz Kid Racer and Quiz Kid Speller. It's part of the COPS I family; the 1977 databook on BitSavers has a lot of info, but leaves some questions unanswered. The COP4 series was released soon after and took over.

I started with Quiz Kid Racer since the source and object code are in patent US4051605. The die is bonded directly to the PCB, covered with a protective platic shield. I took pics of the ROM array and mapped the bits to the ROM dump in the patent; it matches exactly. Interestingly, the patent is specifically about competing with linked calculators, but mine doesn't have a connector to plug the cable into. The code allows solo play if a cable isn't inserted, so the same code works fine. Rik's site has a pic of a Ward's catalog page that mentions buying 2 to compete, and sending 2 proofs of purchase to National to get a free cable, and I saw a pic on ebay that looked like there was a bump on the right side. I guess they removed the functionality in later versions.

I first wrote a disassembler to match the assembler output in the patent, then started a simple emulator. QKR does a RAM test before flashing the symbols +, -, X and / as a menu. I guess they were worried that a RAM error could potentially teach children incorrect math. My emulator runs through the RAM test and flashes the symbols, waiting for a keypress, but I haven't implemented inputs yet. The MM5799 doesn't have enough output lines, so they used a shift register LED driver (DS8874) to multiplex the digits and strobe the keyboard lines.

I picked up a second unit that looks identical, but it flashes the symbols in a different order: X, /, +, -. I opened it up and it has a COP420 instead. The COP420 has plenty of I/O, so it doesn't need the LED driver. The bottom of the case has a spot for a piezo sounder, but none is installed.

There is a "space theme" version that lacks the Amateur, Pro and Complex buttons. I'm not sure if they removed that functionality, or if it's implemented differently.

http://www.seanriddle.com/mm5799.html
http://www.seanriddle.com/cop400.html#qkr

Could you make a photo showing the 7segs inside?

I posted a bunch of pics of the LEDs. Let me know if you need anything else. I also uploaded pics and ROM dumps of Quiz Kid Speller and Mattel Basketball, but haven't documented the hardware for those yet. And I "decoded" the instruction decoder. www.seanriddle.com/mm5799.html

63 7-bit words are decoded in the upper center part of the die; this handles all opcodes except CALL and GO. We might need to trace some of these to see how options are set for AXO, EXC+ and LB (*,0).

Also, there are only 96 nibbles of RAM, but the RAM address register is 7 bits- 3-bits for the "register" (Br) and 4-bits for the "digit" (Bd). RAM can be organized as 8 registers of 12 digits or 6 registers of 16 digits. Some instructions increment or decrement Bd, and others exclusive-OR Br with a constant. I'm not sure if those instructions respect the RAM organization: if RAM is set up as 8x12 and Bd is 11 and gets incremented, does it roll over correctly to 0, or go to 12, which is invalid? I'm guessing it goes to 12 and that index is a mirror of some other one. It would be up to the programmer to handle that.

I'm not sure how the 8x12 addressing maps to the 96 bits. I was using B as the index into RAM, but in 8x12 mode, if Br=7 then B=$70, which is beyond the end of RAM. I fudged it for now by setting the RAM size to 128 nibbles, but that won't handle mirroring. Hopefully no code relies on that anyway. It might be moot- so far as I'm emulating the 3 games, none of them have accessed beyond register 5, digit 15.

There's a similar question about ROM pages 8-15; the docs say they "cannot be used", but I'm guessing they just mirror other pages.

Successive LBs are skipped, but what about LBLs?

I compared the Quiz Kid Racer and Speller dies to Basketball and couldn't see any differences other than the ROM and output PLAs.

I added the COP420 version to MAME, but unfortunately (like half the games in hh_cop400) it's not working.

Other than the display, it seems to be working. RAM[0F:08] holds the displayed digits, with 0A=blank, 0B=E, 0C=+, 0D=-, OE=X and 0F=/. Digit 6 cycles through X / + - until you hit one of those keys, then a problem is shown using that operator. If you type in a wrong answer it shows E in the rightmost digit and gives you another chance. If you get it wrong again, it shows the correct answer in the last 2 digits. After 10 problems, it shows how many you got right, then goes back to cycling X / + -.

I put a breakpoint on LQID and D and G and Q all look OK. It looks like the segment data for each digit is being or'ed with the segment data for the previous digit, and the segment data for the rightmost digit is being or'ed with the segment data for the leftmost digit.

Other than whatever's not working with the display, the 3rd digit from the left should be the "operator" digit instead of a normal 7-segment digit, and the equals sign at the 3rd from the right should be masked with 0x41 since the physical display only has segments A and G connected, but they are illuminating it with the data for "3", since there wasn't any more room in the output PLA. And the green "right" LED doesn't light.