While searching the custom chip, I found that there is little information on it. I can only come to the conclusion that it is a custom memory mapper AND parallel to serial data conversion IC.
The three ICs on the first board from Acclaim are the 74ALS138 decoder used on many many other boards for memory expansion, a 74ALS74 Dual D-type flip flop and a 74ALS125: Quad bus buffer with three states. I cannot prove this theory yet, but I believe the Acclaim LZ95A53 is all three of these ICs built into one. 20 pins could easily achieve this since many pins are shared and many others are not used at all on the other chips.
Here is the first circuit board with all four ICs. Four ceramic capacitors, one electrolytic and two resistors. The board originally had only one mask ROM so I assume it was a game of 2MBytes.
And here is the circuit board with only two ICs. The custom "Acclaim LZ95A53" can be seen at the top right. By reducing the three chips into one, they also reduced the required space on the board, reduced component count and most likely cost. They also switched from their own board to a board made by Liteon. There are several improvements that I can see on the board when they made the switch. Not only is the copper much more smooth but the solder mask is shiner and they even tented the vias. I rarely see tented vias on game boards. The drills are also smaller and less sharp. The previous board has splintering around all of the drills.
Looking at the revision designators on the boards, the liteon is newer. One more thing that I noticed about both boards is that neither have break marks to show that they were panelized. Acclaim chose to finish off the sides very nicely which is odd. Game boards were technically not supposed to be seen by the end user, but they took the extra time and effort to clean them up as opposed to (looks over at other boards on desk) konami and Capcom. A third Acclaim board that I own which is yet older than the two in detail above has also been finished on all four sides.
In conclusion, Acclaim seemed to have cared a little more about the games that they produced. Going the extra mile to make their boards of higher quality and to develop proprietary ICs. I can respect them for this though I never had any doubt about them. EA on the other hand...made boards that I despise. I may post about them another time. For now I will be following the pins on the LZ95A53 back to their origins and proving whether or not it is simply a combination of three 74-series ICs. Granted it is, I will have a pinout shortly there after.
I have since begun following pins on the board from the custom chip to other locations. So far it would seem that I was correct, mostly. The chip is definitely decoding ROM address and possibly RAM addresses (granted your board requires parallel RAM) and has connections with the Serial RAM; however, it is also making connections to the !AS pin on the 68000 and the !LDSW and !UDSW pins. these pins are beyond me, but I have read they have to do with writing only a single byte at once rather than the full 2-bytes (16-bits) that it capable of.
I will continue to edit the diagram below once I have more information:
__ __ A20 1=| |=20 VCC A21 2=| |=19 NC /C_OE 3=| |=18 !OE2(ROM2) /C_CE 4=| |=17 !OE1(ROM1) /AS 5=| |=16 NC D0 6=| |=15 NC /RES 7=| |=14 NC /LDSW 8=| |=13 NC
/UDSW 9=| |=12 SDA(24lc04) GND 10=| |=11 SCL(24lc04)
Also note that pin 12 which connects to SDA of the serial eeprom is also connected to VCC via a 10K ohm resistor. I assume this is a pull-up resistor for data