David Haywood's Homepage
MAME work and other stuff
December 27, 2012 Haze Categories: General News. 13 Comments on The Post-Christmas Update

Not a great deal to report here, but I just thought I’d mention that Manhattan has been improved since the last batch of screenshots and the background scrolling is much more correct now. I’ve also done some general purpose cleanups.

The reference videos we have (YouTube) show the colours cycling between levels, that doesn’t happen yet in MAME, and to be honest I can’t see any registers for it, I’m starting to wonder if it’s a different version. I’ve asked Charles who dumped the cassette if he can test the version we have, but we’ll have to see. It could just be a missing emulation feature, considering the clown-car credit lists often associated with Deco Cassette the driver is still in rather bad shape but I’m running out of places to look!

While on the subject of the Cassette system it does seem to have been designed primarily with the ‘Highway Chase’ game in mind, and in many senses the feature-set of the hardware is very similar to the standalone Mad Alien / Highway Chase hardware, but seemingly a bit weaker because the actual Highway Chase released on the Deco Cassette system is a worse looking, and rather glitchy game if the references we have are to be considered accurate; it even has bizzare pink/blue desert scenes and water scenes with green water, despite the correct palettes being right there! Furthermore the dividing line in the road isn’t present in any of the reference videos despite the writes for it and emulation of it in the driver?!

Anyway, the background ’tilemap’ is just one of those things that turned out to be very close to Mad Alien, instead of being one simple tilemap it’s actually a left ‘edge’ and right ‘edge’ (which can have independent scrolling to bring them closer together / further apart, while sharing a vertical scroll register for the overall vertical movement speed, clearly designed to draw a road capable of becoming wider / narrower..) Furthermore, there are multiple sections to the tilemap and there is a register to control looping of the top part only, or the bottom part only, or doing a transition between the two, again very much designed to do different sections of a road, allowing for tunnels and the like; that feature was missing hence Manhattan not looping the correct section of the tilemap properly. Fixing this also fixed the annoying background looping in Super Astro Fighter (I also fixed the colours there, the driver was failing to actually store the written palette values in RAM, but then expecting to be able to read them back)

Clearly it was a cheap system for low-budget games, trying out ideas, and creating new revisions etc. without breaking the bank, but that does sometimes make it hard to know if bugs in the driver are real game bugs, or actual emulation bugs, especially when you’re not 100% sure your reference material is the same version as the dumped version. Sometimes there are obvious clues that things differ, the Astro Fantasia videos we have show a completely different main boss, the Graplop videos we have seem closer to the Starcade version with an actual Graplop title drawn whereas the ones we have are the Cluster Buster variant, and what seems like a very early version of Graplop before any polish was added.

I should have another previously unknown 80s game to show here in a couple of days thanks to the same guys who uncovered Planet Probe, but more on that when it happens :-)

I was looking a bit at the protection on the Pgm 3-in-1 but haven’t been able to make any logical conclusions about some of the ops, even what should in theory be the simple ones controlling the movement of the cards. I’ll have to do some hardware tests.


You can follow any responses to this entry through the RSS 2.0 feed.

Haze, I just saw your submission for DECO CPU 6 and I just wanted to let you know that Pro Golf does indeed use this cpu. (I have no clue if it is used in any other games.) I can provide pictures if necessary.

Ok, it’s more the comment about Zoar using it I find confusing, the Zoar encryption is unrelated to what Pro Golf is using (Zoar seems the same as Burger Time, a more complex scheme where the encryption is applied based on a prior write occuring)

There are still some weird bits and pieces tho, Zoar and Treasure Island both decrypt at least one opcode incorrectly, or have some non-standard m6502 inside the cpu? there is a manual patch for Zoar in the driver, and another for Treasure Island that dosen’t work properly. I guess they could also be bad dumps, but IIRC Zoar is verified.

ProGolf (DECO 6 version) also has some weird behavior compared to the other set causing the ball to be misplaced after taking a shot, need to investigate what causes that in case it’s a similar ‘wrong opcode’ problem.

The 222 / C10707 encryption (both of which seem the same) is the least complex of all, just a bitswap on the opcodes, based on notes it seems some early PCBs (and the bootlegs) just implemented the same in logic, although it makes sense to just use the same code, decrypting on opcode fetch is a lot cleaner than decrypting upfront.

Do you know if the there is something wrong with Bump ‘n Jump when in test mode? If you go to test 5 (press the key mapped to coin 1 or 2 to advance the test screen) I see the message “BACK TEST 9H ERR”. I checked the Bally-Midway manual for the game and found it competely useless. I’m putting together a jamma harness for my boardset to see if I can check the real thing.

looks like the same / similar thing to Super Astro Fighter on Deco Cass, it wants to read back what it writes to Palette Ram (and other addresses), which are currently hooked up as write only.


The bootleg for Pro Golf also uses the original code and just implements the DECO CPU 6 with a standard 6502 and additional TTL logic.

The ProGolf parent set isn’t implementing CPU6, it’s implementing 222, which yeah, even Data East implement in logic sometimes. I’m not sure it’s a bootleg. Interestingly that set does work tho, and doesn’t have the weird ball behavior. Are we confident the Deco-6 set is dumped properly?

Shoot Out is a curious case, we have an encrypted US parent set on dual CPU hardware, then a non-encrypted Japan version on much weaker single CPU hardware, and an encrypted Korean bootleg on the same weaker hardware, so I think there are bits of the puzzle missing there.

I dumped two original Pro Golf boardsets and one set matches set 2 (progolfa) and the other did not match either set. Both boardsets are using the Deco CPU 6 so set 2 appears to be legitimate. I also dumped a bootleg Pro Golf which is slightly different than all of the above. I sent the dumps and board pictures to the DU. If you are interested in looking at what I dumped just send me a PM.

I forgot to mention that I did a jamma harness for the bootleg Pro Golf boardset and it worked flawless while the two originals were dead. At the start of the game there are two tones before the title screen appears and when you play the course map on the right hand side is not transparent but the fairway portion it represents (the light green) is black and the trajectory of your shots is drawn onto the mini-map.

if you can make some screenshots for visual comparison that would be good, original references are always useful, especially when we know the board they came from.

your ‘non matching’ deco 6 set is a bad dump

G4-M.A2 FIXED BITS (xxxx1xxx)
G3-M.A4 FIXED BITS (xxxxxxx1)
G3-M.A4 BADADDR x-xxxxxxxxxx

the bootleg copies the DECO6 logic, the only real changes to the program rom vs the original DECO6 based sets are the copyright string -> 1982 (the changes in g04.a4) and the arrangement of the tiles which make up the title (g01.a9 changes)

either way, it’s good to know the basic dumps of the deco6 based set are good (confirmed by your other board matching perfectly, and the bootleg matching perfectly aside the mentioned changes), and that the ball positioning glitch isn’t down to a bad dump.

It’s probably not worth you (re)dumping the bad roms on the 2nd board, it’s likely also a 100% match when dumped correctly.

Haze thanks for looking. I dumped the non-matching original boardset roms three times and got the same results so I assumed they were probably different and did not think to run them through romcmp.

Here are two videos I made with my digital camera of the Pro Golf bootleg board in action.


By continuing to use the site, you agree to the use of cookies. more information

The cookie settings on this website are set to "allow cookies" to give you the best browsing experience possible. If you continue to use this website without changing your cookie settings or you click "Accept" below then you are consenting to this.