Repairing Williams C.P.U. boards.

System 3, 4 and 6.

Test-IC to check basic functionality of the CPU and driver 'out' of the pinball machine.

Improved test-chip enables " flash test" of memory-chips. ( since 13/3/2002.)

Since8/10/2003, version 3 was released, another improvement on the memory test is made. Thanks to the cooperation with Oliver Kaegi from Switzerland. It was his idea to use the "on board" LED's as a better ´indicator´ for the memory test.

Again a change was made on 8/8/2011 I did change the type of Test-Eprom from 2716 to 2732, there were two cases where a 2716 did not work properly on a type 6 board where a 2732 works on all types all the time. The image of the test program was adapted to a 2732 type Eprom.

On "Mark's pinball Page" you can find an adaption of this repair method, as this is an american site I assume the ´english´ is much better, so don't forget to look there to!

To my amazement I found almost no information about repairing these CPU-boards. Even the most known site for repair manuals, Clay's pages at marvin3m.com, doesn't provide any help for these first types of solid state Williams pinball games. ( Since 10/11/2002 there is a complete repair guide for these models 3/7 at marvin3m.com, in the guide my Test-Eprom is recommended and used) Even repairconnection.com which does repairs, doesn't accept System 3 and 4 boards anymore. Only on the french site of Arcajeux there is one very good article about these boards, but they recommend to treat them VERY carefully if they still work. I quote the french text: " On ne touche pas un flipper Williams qui marche. Le démontage ainsi que le transport sont une source d'ennuis quasi infaillible!!!" which says: Don't touch a Williams pinball game if it works. Transporting it or taking it apart is asking for troubles!" The article at Arcajeux can be found here:

As said, I was amazed hearing all this. These pinball machines have been in bars and arcades, where they weren't treated well and moved a lot. So don't touching them sounds pretty strange.

By coincidence, I don't have any of these first models of Williams in my collection (of 15 machines). I lately bought a Williams Test Bench, which didn't work, and also some CPU and driver boards, so I started to check myself where all the problems came from.

The beginning:

Between 1977 - 1984 a lot of different models of CPU boards have been used. The oldest, type 3 and 4, can be identified easily: System 3 does have space for IC14 but there's no socket installed for it. System 4 has an extra IC: IC26, and the socket for IC-14 is present. On the pictures you can see these boards, but on System 3 the socket for IC14 is present, because a lot of these boards have been adapted by their owners. It's mandatory to add IC14 which I'll explain later.

System 3 (with socket for IC14). I numbered the IC's so you can see where they are located. They do not have to be all present at the same time! Also take notice of the orange dot at the bottom right, this is an added 33pF cap. Why, is explained further below in this text.

System 4. You see an extra ic socket for IC26. Here also the IC's are numbered for clarity, but not all of these IC's must or may be present at once on the board.

There is also a rare model in between the two, which has IC14 + IC 26 but does not have the sockets installed.

All of this all is not very significant as we'll see that it's possible to convert all these boards at least to System 4 using some simple modifications.

 

System 6 can be recognised because the battery holder is at the right of the board.

A System 6 board with the battery holder on the right. I numbered the IC's so you can see their location. Not all of them are always present on the board! Note the jumper in the left top corner, it connects resistor R27 and capacitor C23 (both upper sides) and is necessary to work on the test-fixture when only the 5 volt is applied (as explained under ´manual´).

Practice

The first difficulty with these Williams pinballs is, when the CPU does not work, in a lot of cases, all of the coils will energise and all lamps will come on. As a result the fuses blow or the coils or their drivers burn. The same thing happens if you turn on the pinball without his CPU-board in place. And again the same if only one signal 'Blanking' stays 'low´. This is enough to make a "bad reputation" for thiese pinballs. Repair men had to remove the CPU/driver boards out or had to remove some fuses to do 'research' and there where almost no indications. Only the two LEDs on the CPU and their ´information´ was not reliable at all. So most of the time it was necessary to put the boards on a test-fixture what means that the repair could not take place 'on the spot'. My solution to this problem is twice:

A) Do some modifications and "pre-repairs" on the board, and get rid of ´aging-problems´.

B) Use a simple test-fixture (only a 5 volts supply) and a special test IC , which I offer for free download.

 

My tests have resulted in some basic rules:

ALL IC SOCKETS MUST BE REPLACED!! Don't think you can skip this rule. There are at least 5 on the CPU board and 3 on the driver board. Even whilst the boards I have looked "good" without any signs of battery damage or any other damage, there were a lot of bad contacts in the IC sockets which make the CPU board and driver board unreliable.

Always convert your board to a type 4, the modifications are limited. Using IC14 replaces IC 21, 22 and 26, so you don't have to replace these sockets because they aren't used anymore.

Make a modification to the reset-circuit, it's only 10 minutes work (not necessary on System 6).

Check the oscillator circuit, if necessary put a 33pF cap over the cristal, 5 minutes work.

Search and repair cracked solder-joints.

With these modifications I can state that these CPU boards work very well and reliable, I can treat them normal, put them on the workbench, solder something in or out, without having any problems.

1) the IC sockets. They have to be changed, their quality did not resist time. Take your time and be careful. Just lift the black plastic by putting a small screwdriver beneath it and pull the plastic out. Then you can desolder the metal springs, one by one. Insert a new socket of the best quality, they don't cost much more and are really better. If you add IC14 (which you should), you don't need to replace IC21, 22 and 26. Also the sockets of the CPU (IC1) and PIA (IC18) have to be replaced. It takes some time, but it's really necessary.

2) How about IC 14 ? Well, Williams has "Game Proms" and "Roms". Roms are 2716 Eproms which are in IC17 and IC20. The "Game Proms" are 2 or 3 and they are in IC21, IC22, IC26, these are 7614 roms. These 2 or 3 can be added into one 2716 Eprom (which has 4x capacity of a 7614). This 2716 Eprom has to be placed on IC14 and replaces the two or three others. On all websites where you find rom-images, also Williams itself, you'll find the image for IC14. The content of the obsolete 7614 can't be found anymore. If your 7614 Proms are still good then you can leave them, but of course don't forget to replace their sockets! If left you do not need IC14.

If you want to use IC14, you can convert a System 3 board to System 4 by adding a socket for IC14 and converting the selection circuit for IC14. If you want to use a System 3 board in a System 6 game, these modifications are mandatory, as System 6 games always use IC14. So how do the modification?

Like this:

Find IC15, there's a trace between pin 6 and 7 which goes to a round solder pad. Cut this trace and connect the round solder pad with pin 1 of IC 15. That's it !

3) The reset - circuit. This modification was presented by Williams and goes like this: (not neccesary on model 6).

Search the top left corner for 5 transistors and several resistors. There are 4 horizontal resistors with just above them a capacitor. It's ok to remove this cap (C27), also resistor top left (R30). Add a new resistor (Rx) of 10 kohm like shown in the drawing, between the top side of the zener-diode and the left solder pad of the removed R30. That's it.

4) The oscillator circuit. (System 3 and 4)

These Williams boards use a special chip, designed to work together with the 6800 CPU. This chip takes care of the oscillator-cicuit, reset circuit and another signal: "BUS 02". In some 'old boards' there are 2 parts less used, L1 and condensor C68, I think because of this the oscillator doesn't always start correct, resulting in a pinball machine which doesn't boot at all. How to check this: if you have a working CPU board, you should be able to put your hand on the crystal and chip and the oscillator must not stop, signals Q1 and Q2 must be present. You can check these at pin 3, 36 and 37 of the CPU 6800. These should be about 2,5 volts. If the oscillator stops when touched, or if it doesn't work and start when being touched, then the components aren't reliable enough, and it's advised to solder a 33pF cap parallel to the cristal. (On System 6 were 6808/6802 CPUs used and the clock circuit is completely different).

5) Check all the solder joints of all the connector pins, as well on the CPU as on the driver-board! Use a magnifier glass. I'm sure you will find some cracked solder joints with the crack just around the pin. Heat the joint and add a little new solder to flux!

With these modifications you should have good working CPU boards !

The test IC:

You can download the (version 3 ) image of Test-Eprom 2732 and burn it with an Eprom programmer: or buy it here

If you don't have a Eprom burner and don't know someone who can do it for you, you can buy them from me, for the cost of the Eprom (15 Euro, postage included). mailto:leon.b1@scarlet.be

This Eprom has a small program which does the following:

Put the outputs PA0 to PA7 and PB0 to PB7 of PIA chip IC18 (CPU) and of PIA chips IC11, IC10 and IC5 on the driver board, HIGH and then LOW, over and over again. Same for ports CA2 and CB2 of these chips. This goes in a rhythm. This program does not lock-up. This means if any output of a chip is bad, the program will continue to work. This way you're able to use a LED in series with a 2K ohm resistor, a multimeter or a logic probe to read all outputs of chips and see if everything works. A LED in series with a 1000 ohm resistor connected to adressline 6 , will flash if the program runs ok. The 2 LEDs on the CPU-board will also flash, if at least the PIA in IC18 is ok, because they're tied to the signal CA2 of IC18.( on System 7 it's the numeric display that blinks) The test only checks the CPU 6800 (IC1 model 3 and 4 )6808/6802 model 6, and the 6820/21 chip and is not dependent of mempory chips (IC13 IC16 IC19) and certainly not of program Eproms (IC26 IC22 IC21 IC20 IC17 or IC14). Of course clock chip IC5 must work correct..( exists not on model 6 ) also selection-chip IC15 and the buffers of the address and data lines, but this is easy to check (see below).

To make it clear I add here that the PIA's are the output-chips who control all functionalitys, coils, lamps, displays...

The control LED ( in series with a 1000 ohm resitor) must be connected to + 5 volts (= capacitor C23 upper or lower side depending on type CPU) on the left of the CPU board with the red grip, and the green grip connects to pin 15 of the CPU chip IC1.

Do not forget to isolate the two wires on the LED, and control the LED connection by first putting the green grip on masse , the LED must light up, if not inverse the two LED leads.

With the test you can launch a memory " flash" test by pressing the " diagnostic button" present on the CPU board.By pressing this button the normal test stops and you will see tree "flashes " at the control LED if the 3 memory chips are ok. By counting the number of "flashes" you can tell which of the 3 memory's is bad. ( see further)

In version 3 , all rules above still are the same BUT there is a suplementary indication: If memory chips are all three ok you will see following behavior of the " on board" LED's:

First the upper LED lights up once = IC 3 ok, now the lower LED lights up once = IC19 ok, at last both LED's light up = IC16 ok. When one of the memory chips is bad there will be another behavior..( See under "Memory -test").

 

At the same time this means that if the program does not work, only a few components on the CPU board can be responsible. (read the manual..)

Manual.

The program is used with the CPU board outside the pinball machine, connected to +5 volt.(Pin 3=gnd pin 4=+5 on connector IJ2.) Do not forget to adapt the reset-cicuit on model 3 and 4 boards, for model 6 make the jumper between the upper sides of R27 and C23.For System 7 underside of C23 and R3.

Remove program chips IC 26,22,21,20,17, batteries are not needed. Put the test chip in IC17, connect the control LED ( C23 + 5volt side and pin 15 IC1 ) and power up.

If the control LED flashes in rhythm you're lucky and are able to check the outputs of IC18. If he doesn't flash then first try a new 6800 (6808/6802 in model 6 ). Still no result, then try a manual reset by connecting pin 40 of the 6800 (IC1) or 6808/6802 (model 6 ) to ground( for a short wile). Check then if there's +4,5 volt present at pin 40. If this isn't, the problem is in the reset-circuit. Most of this is IC5 for model 3 and 4 boards. For model 6the reset circuit is made by the transistors Q 1/2/3/4/6/7/8 and 9.

Next step if it doesn't work is check signals at 6800 (IC1), pin 3 should have a clock signal. If not, check and follow this missing signal back on IC5. Same for the second clock-signal on IC1 pin 36 and 37. (clock signal is about 2,5 volt). Next signal to check on 6800(IC1) is pin 2, should be +5 volt zijn. Finally check the voltage on pin 5 which should be about 2,8 volt.For model 6 you find the clock on pin 37 ( 2,5 volt ) and you must find on pin 5 also 2,5 volt.For all models 3/4 and 6 check if the buffers of address lines and data lines work, these are chips IC4, 3 and 8 for the address lines and chips IC9 and 10 for data-lines. On the schema you can see the inputs and outputs, which are drawn opposite each other, i.e. A0 is pin 2 IN and pin 3 OUT. A1 is pin 4 IN and pin 5 OUT, and so on.. the input signal must be the same as the output-signal !! One of the signals we're checking now must have been bad if the 6800 /6808 / 6802 has been replaced and is ok now, you trace any bad signal back to its source and take away the short or the other source of non-function. A final option could be the selection of IC17 or IC18 not working, check if you get signals on the address, data and chip-select lines of IC17 pin 20 ) , and selection signals on IC18 (pin 35, 36, 23, 24, 22 and 25). Now all outputs of IC18 should go high and low. If there's an output on chip IC18 which doesn't go up and down, connect this with the output next to it. If this other one also fails then there's a closure on this output, check for that first. If both outputs go up and down after connecting them, then you can be sure that this chip (IC18) is broken. A final test is putting the normal program chips in their sockets (except IC17), their presence shouldn't interrupt the test. Should one of the chips change the test, then there's something wrong with it. Conclusion: if your outputs all work correct, you can start checking memory chips

If you also want to check PIA's of your driver board, then connect the driver board to your CPU-board, put test-chip in IC17 and power up the board. The LEDs on the CPU board will start flashing and also the relay on the driver-board will start clicking. This relay is connected to CB2 of PIA (IC5). If the relay clicks this is also a test on the most important signal BLANKING.. which occurs in many different circuits. It is HIGLY RECOMMENDED to replace ALWAYS C31 and C32, two capacitors wich fix the timing and the output of the BLANKING signal, very important for the driver board ! The blanking signal enters the driver via pin 37 of 1J1 if it is not there follow it back on the CPU where IC 23 is the source of it...Somethimes the relay clicks not at all , because it's slow , and do not react at the short blanking pulse. so if you do not hear the relay mesure with a probe on his connections points, if there is a pulse it's ok.

Now check if outputs of the 3 PIA's go up and down. Caution the outputs of PIA IC5 model 3 and 4 or IC11 model 6 , PA0 to PA7 are always forced low. To bypass this connect all pins of connector 2J3 to ground (7 pins. use an alligator clip to ground) and then these outputs will also dance like the others..

If there are some PIA's who don't work at all, again you'll have to check the selection of the PIA on it's selection-pins, these are 35, 36, 23, 24, 22, and lok at pin 25 for 2,5 volt. If one of these signals is missing, follow it back to its source, using the schematics and logic-probe or oscilloscope. If it's only some outputs who don't work, connect it to a working output, if they both DON'T work then there's a short on the non-working output, if they DO work then your PIA is defective.

Here's the setup to test the driver board. Connected to the CPU, in IC17 is my test-Eprom, and you'll notice both sockets next to it have been replaced..

Connector 2J3 on the driver board has been connected to ground with all its pins..

 

Conclusion:

Once you know the outputs of the PIA's work, from there on you can use the standard lamp- and coil-tests incorporated in the pinball tests...Or you can use the driver board test also available on this site.

Memory test:

Once the PIA test passes ok, we test the memory's. Push the " diagnostic" button on the CPU board. ( ATTENTION , I had the case the " diagnostic" pushbutton did not work.. i stead you can connect pin 6 of the CPU chip briefly to mass, this creates the same effect as pushing the " diagnostic button.) The The PIA test will stop and the LEDs on the CPU board will remain on or off depending of the moment we push, in any case this doens'nt matter. We look at the control LED and afther a short timing where the control LED is off. The "flashes" will appear. The first is for IC13 ok , the second for IC 19 ok and the third for IC16 ok. If you had tree " flashes" the test automaticly returns to the normal PIA test and every thing will start flickering again. IMPORTANT! a " flash" means the control LED goes on AND off again. Example; if the LED flases once , then come on again and stays on this is ONE flash..If you, push and the LED comes on ( not at full strenght) and stays that way , this means NO flash.

No flash means IC13 bad, one flash IC19 bad, and two flashes IC16 bad.

To make things easy, you may also get our MEMORY-Tester if you need to test common RAMs on Pinball boards.

VERSION 3:

If the memory chips are all three ok you will see the following behavior of the "on board" LED's. First the upper LED will flash once, = IC13 ok, then the lower LED will flash once, = IC19 ok , the both LED's will flash, = IC16 ok.

If one of the chips is bad ;. BOTH LED's will light up and stay "on" = IC13 bad, BOTH LED's will flash once and stay on, = IC19 bad, BOTH LED's will flash twice and stay on = IC16 bad.

In 85% of the cases its the memory chip.so replace this first, put an socket in place handy for a next time.If afther replacement the test still indicates the same memory bad, you have to control the selection signals on that chip. As the test will continue to test the chip as long as the result is not ok, it's easy to control the different selection signals which you can control with a logic probe, or a voltmeter.( Logic probe is best).

For the 6810 memory you have to find; pin 1 = ground, pin 2 to 9 included =data= 1 to 2 volt , pin 10 =2,5 volt, pin 11= 0volt, pin12 to 16 included =3 volt , and 17 to 23 included =2 volt, pin 24= 5volt.

For the 5101 memory you have to find, pin 1 to 6 included =2 volt, pin 7 and 8 =0volt, pin 9 to 16 included about 1 to 2 volt. pin 17 = 3 volt, pin 18 = 0,5 volt, pin 19 =3volt, pin 20 = 3,5 volt, pin 21 = 2 volt, pin 22 =5volt.

I hope that this very technical document will help a lot of people and wish you a lot of success!

Any questions ? Email me!mailto:leon.b1@scarlet.be

Back to home page...(LEONs homepage is no longer operational.)