Rewriting yesterday's remarks now that all Idea-Tek games except for Venice Beach Volleyball
have been redumped. These games share a similar IC (with different markings) as the one described for Mapper 36
. But they are connected in different ways.
This board connects the IC's data bits four bits rotated from the Mapper 36 description
, so $4103's Mode bit is at bit 0, not at bit 4. $4102's D0 (IC Pin 3) is CHR A13, but $4102's D1 is NOT CHR A14. I cannot quite see which pin of the IC is connected to CHR A14 (I'd say Pin 4 or Pin 22), as the lead goes under the chip, but CHR A14 is always the inverse of the IC's XOR flag, e.g. a XOR flag of 1 selects CHR banks 0 and 1, while a XOR flag of 0 selects CHR banks 2 and 3, regardless of everything else. $4102's D3 must be latched to yield the expected $4100 read value.
小瑪琍 (Xiǎo Mǎlí, Nei-Hu and TXC versions):
The GoodNES dumps have only 8 KiB of CHR-ROM. It turns out that the game actually has 32 KiB CHR-ROM, with the first two banks empty, and the second two banks identical. Normally, this would be considered an overdump. In this case, it's part of the copy protection: The game flips $4102 D1 all of the time but always keeps the XOR flag at 0, so CHR banks 2 and 3 are always selected. Incorrect emulation will cause the empty CHR banks 0 or 1 to be selected when D1 goes low. This applies to both the Nei-Hu and TXC versions. Additionally, the Nei-Hu version checks that the value read from $4100 has the correct post-adder value, otherwise PPU rendering is turned off. The C&E version on the other hand is simple NROM with no protection.
F15 City War/Puzzle (original Idea-Tek releases):
Both use four CHR banks. In accordance with the above description, their bankswitching code uses a XOR flag of 1 for banks 0 and 1, and a XOR flag of 0 for banks 2 and 3. Puzzle is the only game that writes a value other than $00 or $FF to $4101: a value of $01. Given the $4100 result it expects, $4101 value $01 seems to still invert the entire byte written to $4102, so $4101 seems to be a "XOR $FF", or "invert" flag register rather than a XOR value register. $4102 D3 is expected to be latched rather than return zero or open bus.
撲克精靈 (Poker Jīnglíng):
The same, except that this game for some odd reason also has a Mapper 3 compatibility write, so our redump matched GoodNES' dump, which we had believed to be a Mapper 3 hack. It will not run as Mapper 3 on emulators that emulate bus conflicts (such as FCEUX) though, nor will it run on a Nintendo CNROM board, for the same reason.
Identical to Mapper 173, except that $4102's D1 is in fact CHR A14, allowing all four banks to be selected independently of the XOR flag value. Also unlike Mapper 173, $4100's D3 is expected to always be low even if a high value was written to $4102's D3. This may be the result of the adder being four bits rather then three bits wide, as I believed before and wrote yesterday, or more likely the result of D3 just not being connected and returning open bus (which would be zero if the address is $4100).
四川麻將 (Sìchuān Mahjong,
aka Mahjong Trap), 蝶變 (Diébiàn,
aka Galactic Crusader), 未来小子 (Wèilái Xiǎozi,
aka Joy Van Kid,
aka Metal Fighter µ):
The original Joy Van releases of those games use this mapper, checking not only the adder result, but that $4100 D3 reads back clear even when previously written with D3 set.
Rad Racket (TXC version):
We dumped this cart with all four banks at the same XOR flag value, indicating that the board we used for dumping is Mapper 136, not Mapper 173. Yet the game's bankswitching code still uses a XOR flag of 1 for banks 0 and 1, and a XOR flag of 0 for banks 2 and 3, so it's obviously written to be compatible both with Mappers 136 and 173.
麻将方块 (Mahjong Block, TXC re-release, CRC32 0ACFC3CD):
The current GoodNES dump of Mahjong Block is set to Mapper 173, which cannot be right because the game uses the same XOR flag value for all four CHR banks, which makes it Mapper 136. This GoodNES dump must be a TXC re-release of the game, even though the Super Mega copyright is retained on the title screen, as the actual Super Mega release, with a PCB that says Super Mega, uses Mapper 172 instead. It's not uncommon that TXC would re-release games without removing the original copyright --- the early TXC re-release of Xiǎo Mǎlí
retains the Nei-Hu copyright, and the TXC re-release of Rad Racket
retains the American Video Entertainment licensee.
This board is used by 1991 賭馬 Racing
and the original Super Mega release of 麻将方块 (Mahjong Block)
. It may have been used on the Super Mega Release of Venice Beach Volleyball
(the current dump of which is a Mapper 3 hack). It is similar to Mapper 136 except that the IC's data bits are rotated by four bits, and are thus similar to Mapper 36, and that the CHR bits are reversed. In the end, §4102's D4 is CHR A14, and $4102's D5 is CHR A13. Games using this mapper take the desired bank number and reverse its bit order in software by using six LSR/ROL opcode pairs before writing it to $4102.
Like Mapper 36 except that D0 is CHR A13, D1 is CHR A14, D2 is PRG A15. Used by 戰國四川省 (Zhànguó Sìchuān Shěng,
aka Tiles of Fate) from C&E
and 棋王 (Qíwáng)
Why did Joy Van and one C&E game use an Idea-Tek mapper? Because Idea-Tek was founded by former Joy Van employees
, and C&E's Tiles of Fate was initially distributed by Idea-Tek.