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

I got the ML102 dump to look better, but BX301 still has lots of duplicate pages. One of the page select bits must not be getting set correctly; I'm sure it has to do with the timing of the opcodes that I'm forcing in.

One interesting thing is that ML102 has the text "YOSHIDA".

www.seanriddle.com/bx301.bin

www.seanriddle.com/ml102.bin

Programcounter is linear in these dumps, right?
Anyway, even after converting to LFSR, ML102 locks up pretty quickly, something's still not right I bet.

Yeah, that's what Igor said. It's executing NOPs while dumping, so PL is incrementing in LFSR order and ROM is coming out in execution order.

I'm sure there are some pages that I'm still not getting, and I'm not 100% sure that all the pages are starting at the right place. I haven't found a "start of page" indicator, so all I can do is compare the dump right after ACL to the dump of page 55 and adjust the starting offset of each page based on that. After my latest tweaks, BX301 pages still start 8 bytes in, but ML102 pages start right at the beginning.

Just send "Page select", then "Jump #00", and immediatly start receiving. You always get new page from address #00.

Hey Igor! I was using the TL opcode to load the page and program counter, but BX301 page 55 didn't start at the same byte as right after ACL. I just changed the code to do a TL and then a T 0, and that seems to have fixed it. I also tweaked the timing a little and now I'm getting 64 unique pages: www.seanriddle.com/bx301.bin

Is it possible to dump the last byte of the page? On the die images it's $80, in your dumps it's $00, but in theory it could be any value.

In ROM dump mode the chip only outputs 63 bytes per page- I just inserted 00 to fill the gap. I guess TI must have had a patent on modifying an LFSR to reach all values.

Likewise, in the ML102 dump I inserted 00s for the 20 "missing" pages. I should be able to make a new dump of ML102 today with the new code.

It's possible to read the 64th byte. Just execute command "Goto #63" LFSR will load #63 and stuck on it. All bytes read from processor will be 64th in ROM. (in my case was 80h).

I decapped and visually dumped G&W Boxing BX-301: http://www.seanriddle.com/gameandwatch.html

Turns out that my electronic dump got all the bits, but shifted the bytes one position in pages 32-63. I'll look at the logic capture and see if I messed up the conversion. Or maybe my timing's still not quite right when I send the opcodes.

I also digitized the segments: http://www.seanriddle.com/bx301svg.7z

And scanned the background: http://www.seanriddle.com/bx301bkg.tif http://www.seanriddle.com/bx301bkgblur.tif

Do you have a higher resolution image section of the die shot, for extracting the melody ROM?