Note: Controller numbering on PCB is opposite of numbering on case. Number on case appears to be the correct one, so I will use case notation. There is a three position switch allowing selecting the mode: 2p,5p,6p The only other components (besides resistors, or capacitors) are two transistors, an LED, and a 20 pin IC. ;-------------------------------------------- SNES connection: Here are the connections from the SNES (and the pins on the trace side of PCB from left to right) rounded tip GND IO D1 (to serial port) separator D0 (to serial port) latch clk Vcc square end Note: D0,D1 are inverted internally in the SNES main chip, so that Gnd -> 1 and Vcc -> 0 for reads from $4017 (or $4016). Bit value in $4201 goes to IO pin (remember, this is open collector/drain so reading $4213 is effectively an AND of what controller and $4201 have for this bit ... allowing controller device to communicate back to SNES through this line if $4201 has the IO bit set high) Bit value in $4016 goes to latch (no inversion). Writing values in $4017 don't appear to change any signal on controller side. Clock pin is normally high, and is strobed once per read to $4017 (or $4016 if using controller port 1). Clock will occur regardless of values written to $4016 or $4017. ;----------------------------------------- 20 pin IC From trace side of PCB, chip is upside down with pin one on the upper right. From chip side of PCB we have: ___ ___ SNES, clk - 1 V 20 - Vcc SNES, IO - 2 19 - SNES, D0 P2,d0 - 3 18 - SNES, D1 P2,d1 - 4 17 - P2,latch deals with switch - 5 16 - P2,IO P3,d0 - 6 15 - P3,clk deals with switch - 7 14 - P4-P6,clk P5,d0 - 8 13 - P3-P6,latch P4,d0 - 9 12 - P6,d0 Gnd - 10_______11 - "inverted" latch Pins 1-9 are pulled up to Vcc using a SIP resistor pack. pin 11 is pulled high by a resistor, and goes to a transistor. Transistor is NPN (part S9013). Collector connects to pin 11. Emitter is grounded. Base goes to two resistors, one to ground and the other connects to the SNES-latch signal (which is also pulled high by yet another resistor). This is acting as an open collector inverter for the latch signal to pin 11. ;----------------------------------------- Player connectors: P2: This is special, as it is the only controller port that has traces going to all 7 pins. It probably was designed to pass all lines through for 2p mode for compatibility, but in the end they were cheap and used a connector that only uses Gnd,d0,latch,clk,Vcc (leaves out d1, and IO). For this port the SNES-clk is a direct connect through. For some reason it doesn't share the latch line with the other conrollers either. Vcc connects directly to SNES line. P3-P6: latch line is common Vcc is controlled by the switch and a transistor. P4-P6: clk line is common ;----------------------------------------- Switch: pin 5, pin 7 of IC mode 2p Vcc Gnd mode 5p Vcc Vcc mode 6p Gnd Vcc The switch and a transistor also controls the Vcc of P3-P6 ports. The switch connects to base of NPN transistor (part S9013). Collector connects to Vcc from SNES. Emitter connects to controller P3-P6 Vcc. The switch (base) line also connects through resistor then LED to Gnd. When in 2p mode, this switch line is connected through a resistor to Gnd. In 5p and 6p mode it is connected straight to SNES Vcc. Pin 5 is pulled up to Vcc, and the switch does nothing except for 6p mode where it pulls it to Gnd. Pin 7 is pulled up to Vcc, and the switch does nothing except for 2p mode where it pulls it to Gnd. ;----------------------------------------- 5p and 6p modes: I have yet to find anything that distinguishes between 5p and 6p modes from the SNES side. Nothing seems to give data on the player 6 controller. The latch lines are all set according to $4016, regardless of the IO pin setting. With latch line low: IO=1 snes-d0 = controller 2, d0 snes-d1 = controller 3, d0 controllers 4-6 don't get clock signal IO=0 snes-d0 = controller 4, d0 snes-d1 = controller 5, d0 controllers 6 gets clock signal controllers 3 doesn't get clock signal (controller 2 always gets controller signal due to how they designed the device) SNES-d1 will be pulled low (will read as 1) if the latch line is held high (regardless of IO pin setting). However d0 still responds, but since latch is held it is basically just continually reading the B button. This setting of d1 was done on the official multitap as well to allow detection of a multitap vs. a controller. ;----------------------------------------- Consideration, clues, and speculation: The official multitap uses combinatorial logic (ie there is no "state" to the system, the data on the input pins completely defines the output). This cannot be done here. There are no inputs to the SNES left except the IO pin, but the IO pin needs to be considered an output to the tribal tap to select which controllers are read. And there are no other outputs from the SNES in which to select additional controllers (the latch could have be used, but its action has already been explained above and it shows that is not the case here). I was guessing the extra data was sent back via the IO line, as it could allow all 5 controllers to be read in two passes. This can't be done purely with combinatorial logic because the IO line will necessarily change. Something needs to "trigger" a state in the logic, in which the player 6 data is used on the IO line. I have not found any such trigger yet. Another possibility is that after so many reads from one port, the data from another would be placed on d0 or d1. I have read 256 bits from each controller... nothing special happens. The only remaining option is some more general "trigger" that allows the player 6 data to appear on the SNES d0 or d1 inplace of another controller's data. I have not found any such trigger yet. This device was made by Naki, which still exists (www.naki-world.com) so it is possible (albeit unlikely) that someone could be found at the original company that can provide specifications or some kind of datasheet. There is also the chance that some games were capable of using this for 6 player support (Nach says "There are 2 or 3 soccer games that use it IIRC. They were 6 player." A seller on EBay claims the manufacturer told them some European and Japanese games used it.) ... I don't have all the games, and no emulator supports the tribal tap, so the only way to test this is to look through the controller reading code of every possible game while looking for bizarre controller access code. I checked a couple candidates, and didn't pursue this further. (I could also run each on a copier, with the tribal tap attached, but not only is that time consuming, but I can't read Japanese or most European languages... and the European games often need to be patched just to run on the copier.) This company made some very sophisticated devices for the SNES (such as a save-state device), so I trust that this device works some how, some way. I unfortunately can't figure it out at the moment. EDIT: I also ran some tests with all 5 controller ports filled just in case you need the first four attached to activate the 5th. Still nothing noticeable. ;---------------------------------- Request for help: If you have some ideas and the means to try them out, I can probably help out and may even be willing to buy you a tribal tap. Please contact me if you are interested.