Williams System 3-7 technology
Williams started making solid state pinballs in 1977, when Hot Tip and Lucky Seven
games were available as both EM and SS models. This made it possible to compare
durability and profitability between different technologies. Unlike Gottlieb,
Williams did not make any more games in both versions, and although a few EM
games were still produced, their era was definitely over.
The control system of the first machines was named System 3. Before these,
Williams had tried completely different systems in short proto series of
solid state Aztec and Grand Prix, they were then Systems 1 and 2. System 3 was followed by
Systems 4,6 and 7 - System 5 was used in shuffle and baseball games but never
in any pinball. All these are very similar in basic operation. Afterwards
they presented Systems 9,11 and eventually WPC and Pin2K, but those differ
much from the older systems and are not discussed here. There was also
System 8, again used only in one baseball game.
To refresh your memories, a list of Sys 3-7 games:
System 3 (11/1977 - 8/1978) :
Hot Tip, Lucky Seven, World Cup, Contact, Disco Fever
System 4 (10/1978 - 3/1979) :
Pokerino, Phoenix, Flash, Stellar Wars
System 6 (7/1979 - 10/1980) :
Tri Zone, Time Warp, Gorgar, Laser Ball, Firepower, Blackout, Scorpion, Algar, Alien Poker
System 7 (11/1980 - 2/1984) :
Black Knight, Jungle Lord, Pharaoh, Black Knight Limited Ed., Solar Fire, Barracora, Cosmic Gunfight,
Warlok, Defender, Time Fantasy, Joust, Firepower II, Laser Cue
1977-1980 were years of fast evolution. System 7 had enough memory capacity for
quite complicated games, so it remained in use for over three years.
These systems are based on Motorola 6800 (system 3,4) or 6802 (system 6,7) CPU. PIA chips
6820 or 6821 are used to drive 7 segment plasma displays, first with 6, later with 7 digits
and also lamps, solenoids and switches. RAM memory onboard was first 256 bytes and later
512 bytes. Bookkeeping and settings were kept in a 256x4 bit CMOS RAM. Program storage was at
first 512 or 2048 byte mask ROM, and later up to 4 kB ROM or EPROM, that the different systems
could have 4-6 chips onboard. No special or custom IC's were used, so even after 25 years
it is possible to repair the boards, although the old Motorola chips are getting a bit
difficult to find.
Inside a System 3 game (Disco Fever). Upper left is the CPU board, below it the driver board and
at the right is the power supply. Sound board in System 3 and 4 games is at the left side of
lower cabinet but in System 6 it took its place from the backbox, just over the power supply.
A smallish transformer can be seen at lower right. Note the driver board connecting directly
to CPU board with a long row of pins.
System 4 looks the same. Most noticeable difference is in game software. Settings can be entered
with front door buttons instead of using DIP switches as in System 3. The CPU board still has those
switches so it is interchangeable between Systems 3 and 4. 512 byte masked ROMs were not used in
System 4 anymore, so they have one less ROM socket. This socket can be added to board but today it
is wiser to burn the software into 2 kB EPROMs that can be used in all CPU boards.
System 6 machine (Alien Poker). Sound board is now on top of power supply. There are some changes
in CPU board layout, e.g. the battery holder is in different place. There are no significant changes
in driver or power supply boards.
System 7 machine (Black Knight). Again some changes in CPU board. Driver and sound boards are the same.
A couple of rectifier bridges that were previously attached to backbox, have been moved to power supply board.
Power transformer is now put back in lower cabinet so it doesn't build up heat in backbox anymore.
Power supply board
The power supply stayed almost unchanged between System 3-7. There is a 3 amp 5 volt
regulator in a big heat sink. It generates supply voltage to logic circuits. Incoming
AC from transformer is rectified with two diodes. The regulator itself is well protected
and doesn't usually fail. However the diodes do sometimes fail. A shorted diode blows
the fuse, an open diode causes 5V voltage to drop and prevents game from starting. Common
problem in all power supplies is drying of electrolytic capacitors. The regulator input
voltage is filtered by C15, a 12000 uF electrolytic. When it begins to fail, it may take
longer for the game to boot up, or the game may reset spontaneously. Depending on
temperature, lifetime of these caps is 5-15 years so they are definitely getting old
on these games.
Plasma displays require +100V and -100V. These voltages are made with two similar circuits
on power supply board, one has NPN and the other PNP transistors. The series pass transistor
is subject to failure. Replacement must be capable of handling at least 200V. The original
types are not available anymore. Suitable replacements are MJE340 (NPN) and MJE350 (PNP)
but with these you must twist two leads over each other.
Heat sink compound is recommended between transistors and heat sink. Filtering the display
voltages are two 100 uF electrolytics. When those dry, the displays go dim or start to flicker.
It might not be a bad idea to replace the 100V zeners with 91V ones to lower display voltage a bit.
It makes displays slightly less bright but increases their lifetime much. The display tubes are
getting hard to find.
Driver board connects directly to CPU board with a 40 pin connector. Processor bus goes thru this
connector. It is important for the connector to be clean and solid because of the high speed signals,
otherwise the game will not function correctly.
There are three PIA IC's on the driver board. IC11 is used in switch matrix reading. PIA B-side
lines drive the matrix via 7406 buffers. A-side reads the matrix via 4049 buffers. If there is a
fatal short on playfield, causing coil voltage to enter switch matrix, it is these buffers that
usually fail. The PIA itself may survive. There is a 1k resistor on wires going to playfield but
it is not always enough to protect. So when servicing game with metal tools, keep power turned off!
Most common way to blow switch matrix chips is to make shorts under playfield with a screwdriver.
Another PIA, IC12, controls lamps. Like the switches, also lamps are connected as 8x8 matrix.
PIA B-side outputs +18V to matrix columns via 7408 gate and two transistors. A-side pulls matrix
rows to ground via 7406, two transistors and a thyristor. Current in lamp row is measured over
0.4 ohm resistor and when it rises enough, the thyristor is triggered and row turns off. This prevents
light bulbs burning out even in case the CPU suddenly stops. This clever way of lamp controlling is
still in use.
A third PIA, IC5, controls solenoids. They are divided in two groups, normal and special solenoids.
Normal solenoids are controlled only by CPU. Special solenoids are used in bumpers and slingshots.
They must act so fast that on these machines the CPU could not control them quickly enough. Closing
of a playfield switch then enables solenoid power directly, but it can also be controlled by CPU, as
is done in solenoid test. PIA's A and B sides drive 16 normal solenoids via 7408 and two transistors.
Special solenoids use 7408, 7402 and two transistors. Solenoid operates if playfield switch pulls 7408's
one input low. The other input can be pulled low by CPU with PIA's CA and CB outputs.
The Williams driverboard is amazingly durable. On the lamp side there are seldom problems. Sometimes
a TIP42 fails. Those are cheap and easy to find. Solenoid driver transistors fail if the solenoid
stays on for some reason. Common darlington TIP122 is suitable replacement. Usually the 7406 and 4049
buffers protect PIA in switch matrix faults, but as the PIA's are getting rare it is best to get a few
of those now.
System 3 CPU component placement
System 3 CPU schematic
System 6 CPU component placement
System 6 CPU schematic
There are some differences between CPU boards in various systems. Basic operation is the same for all of them.
Besides CPU, there is one PIA on board, that controls displays and is used to read DIP switches and one
push button (The other button generates NMI interrupt to CPU).
CPU boards don't fail often. Most common problem is batteries leaking. Stuff oozing from old batteries
will corrode foils on PCB and destroy IC sockets. The sockets are second most common problem.
They are cheap and use only tinned connector spring against IC lead. They oxidize and lose tension.
The worst sockets are of SCANBE brand. If you are fixing a dead CPU, first check if it has SCANBE sockets.
Replace those with proper good quality sockets before even trying to diagnose other problems.
Battery leaks can be cleaned with vinegar solution and toothbrush. After that start checking if foils
have been damaged. It takes long and is tedious work. You might or might not get the board working.
After you have good sockets and known good ROMs, the board will test itself pretty thoroughly. Nice thing
is that the board can be operated on workbench, just connect 5V to power connector J2 (J2 1-3 = 0V, J2 4-6 = +5V).
Observe two LEDs flashing on then off. This shows that CPU, clock and reset circuits are OK and at least
part of ROM is working. If the LEDs do not flash, check for 1 MHz clock signal on CPU (IC1 pin 3). Bus buffers
IC9-IC10 may fail. They can be bypassed. From System 6 onwards this has been done at the factory. Just remove
the chips and connect jumper wires between 2-3, 5-6, 9-10 and 12-13. The buffers are not needed if there are
no 512 byte masked ROMs onboard. 8T28 buffer chips are hard to find.
When the board starts so that you get the LEDs flashing, push the diagnostic switch (the lower button) once.
The LEDs should flash twice. If this happens, the board is probably OK. It might still have display problems,
if that happens check PIA IC18. Most other components are OK if you got the two flashes. If either or both LEDs
remain lit, the test program has found a problem:
Lower LED on: ROM failure. Check sockets and ROMs
Upper LED on: RAM failure. Check IC13/IC16, sockets and IC8 (7408).
Both LEDs on: CMOS RAM (IC19) failure. Also check IC12/IC7.
On System 7 boards there are no LEDs, but a 7-segment display instead, and more exact test routines.
When you power on the board the display should show 0 for a moment, then blank. If it doesn't, check the same things as in the
older boards: CPU clock, reset and bus. After pushing the diagnose switch the display should again show 0 for
a while and then blank again. If other numbers appear, there is a fault:
"0" CPU does not start.
"1" IC13 RAM.
"2" IC16 RAM.
"3" IC17 ROM 2.
"4" IC17 ROM 2.
"5" IC20 ROM 1.
"6" IC14 Game ROM 1.
"7" IC26 Game ROM 2.
"8" IC19 CMOS RAM or memory protect circuit (front door switch).
"9" Memory protect circuit (front door switch).
Bad IC sockets have been mentioned. It is really a common problem in these games. Also bad contacts
between CPU and driver board can cause strange things. The connector should be cleaned well, and it is
also wise to resolder all pins. The solder joints are very often cracking, especially if you have
removed the driver board many times. The same goes for all pin connectors. Take some time to reheat all
pins and add some solder. It gives you trouble free operation for the next 20 years.
A fact is that all electrolytic capacitors dry out. You can estimate capacitors condition by
measuring voltage over it on both DC and AC ranges. The AC reading should be much less than 20%
of DC reading. For example, if the capacitor shows 20 VDC, you should see less than 4V (typically under 1V)
on ACV range. For this measurement to work, your multimeter should block DC when measuring AC. Most do,
you can check by measuring a battery voltage on ACV range - should be 0.
If the game fails to boot, first check all fuses. Check all power supply output voltages - see the schematic.
There might be dried capacitors. A short on driver board may prevent CPU starting. You can remove driver
board and see if the CPU then starts and displays come on. If this happens, there are problems with your
driver board. A short, bad connector or failing PIA.
Be careful when testing a non-starting game. Some coil may be constantly energized. This will burn
both the coil and its driver transistor. Until you get the CPU alive, it might be wise to remove
coil and lamp fuses F2 and F3 from power supply board to minimize risks. Check power supply +100 V and -100V
outputs so that you do not burn out the displays with too large voltage. And of course the power supply
must be fully working before you can even start to diagnose other boards.
I have seen the battery connection diode D17 short. Then the game is trying to "charge" the batteries
when turned on. Alkaline batteries do not like this at all and may even explode! So, if you start working
on unknown machine that has been long stored, check the diode. If it is OK, just insert new alkaline
batteries and forget about them for two years.
Typical problem in drop target banks are the switch diodes vibrating themselves loose. Target stops
scoring then. Use short leads on diodes and mount it with heat glue gun or silicone rubber compound.
Some games have small capacitors on bumper and drop target switches, same goes for those also.