Leon Borre Williams System 11 Repair
Repair Williams CPU-board type 11.
Of this type of board exist multiple models. To learn more about the differences, click here The test program works on all boards.
The goal of this article is to verify and if necessary to repair the cpu chip, the memory chip and all connected PIAs. These PIA's (there are seven 6821) are used to send all signals to the external circuits, to the displays, solenoids, lights and via the switch matrix to all switches. When the CPU and PIAs are fine, it's almost certain your pinball will start and you can find any other problem using the internal tests. I assume that all voltages are ok, be sure to test these first using a DMM !
Start
So your pinball machine does not start, but you've checked all voltages and they are fine.
Remove the cpu board from the machine. If you are not sure where to put all connectors, mark these first !! Once the cpu board is out of the machine, connect it to 5 volts. It is also necessary to make a temporary connection between the right side of zener diode ZR1 and 5 volts.
The whole test setup. The red arrow marks connector 1J17 where you should connect 5 volt to pin 4 and ground to pin 3. Also visible is the extra check led (usage is described in this document) between the 5 volt and en pin 15 of cpu chip U15. A temporary connection (black wire) between 5 volt and the right side of zener diode Z1. (This diode is just left of the batteries)
Usage.
You need a test eprom which goes into socket U27. (First remove the present game-rom and store it safe !!) This test-chip has a non-blocking test which sets all outputs of all 7 PIA's to high ( 5v) and then low (ground). Non-blocking means the program will continue even if one of the PIA's is broken. Now you can check (using a DMM or logic tester) which PIA outputs go up and down. If none of them works the problem is on the cpu chip or the selections. Add a control led to address line 6, and it should light in the rythm of the test and is our visual check.
The check led, a regular led in series with a 1000 ohm resistor. Isolate the connections before use, and check polarity the first time you'll use it. Put the green side to ground, the led should light, if not then inverse the led connections.
The program is written into an 27512 eprom . You can download the program here............. ...... if you don't have an eprom programmer you'll have to find someone who can do it for you.
Since 10/5/ 2005 there is a new version of the test eprom ( Version 5). It is the same test eprom as for the Data East CPU board. The memory test is better and more severe. There are added extra commands at the PIA test. Now we can test all the output drivers right up to the connectors. I added a extra paragraph at these pages . To be sure you have the newest version print the image with " WORD" between the strange signs you will find the text "version 5." (Note: the 6821 lines CA1 and CB1 (pins 40 & 18) are Interrupt Inputs only and can't pulse)
Now we're started, if you're lucky the check led will blink, and if the PIA in IC 51 is ok, the diagnostic led of the board which is connected to this PIA , will also blink..If you have an early type sythem 11 there is a little display , there a number 7 will blink.
There are 7 PIA'S ( U51, U10, U38, U54, U9, U41 and U42), all pins from 2 to 17 have to go from 0 to 5volt in rythm. There are a few exceptions !! At PIA U38 are PA0 to PA7 connected as input (pins 2 to 9) and to see these move it is necessary to connect pins 1 to 9 of connector 1J10 to ground. Then you should see these outputs also move. At U9 it's output PB0 to PB7 which don't move (pins 10 to 17). These are connected with the D/A (digital to analog) convertor U2 and take care of the sound. But repairing the sound is something which is out of the scope of this article. And anyway the pinball will start up fine without sound...Finally there's pin 9 of U51, which is connected via jumper W7 to ground and it will also not move...
If one pin doesn't move, short it with the pin next to it. If both go up and down then the PIA is broken. If both don't move, then or the PIA is broken, or there's a short in the output which didn't move in the beginning. To know which of the two problems we have, you'll have to disconnect the output (you can cut the trace, be sure to repair it later !) and check if it works now. If it moves, then there's a short, if not, it's the PIA which is certainly bad.
......In the worst case...
If the program doesn't run, we'll have to look at the basics and that's the CPU-chip-6808 itself. First replace the 6808 which is the fastest thing to do, the chip is socketed! Use a 6808 of which you're sure or a new one... Still doesn't work ? Then check these pins: 2, 3, 8, 35, and 40 they all have to be positive (around 4 volt). On pin 39 there's the clock signal, on pin 4 IRQ and on pin 5 the VMA signal, and on pin 37 signal E (syncro for external parts) these all have to be around 2 to 3 volts (measured with a DMM). They are moving signals, here you find some examples of them.
The clock ....... and VMA signal
The E signal is almost identical to the IRQ signal...
If you have differences with these 8 signals, then you've found your problem and using the schematics you can trace them back until you've found the source of the problem. This is not a big problem, the circuits aren't very complex.
The long road... or not !!
What rests are the other outputs of the cpu, address lines (pins 9 to 25, except pin 21 which is ground !) and data lines (pins 26 to 33), but also the selection circuit of the test-program-chip U27 and the selection circuit of the PIA's. If the program doesn't work it's possible the test-eprom hasn't been found, or the CPU or PIA's cannot be found. How do we solve this problem ? It's very easy !!!
It's useless to give you the signals because without a working board you won't have these. So I've found this workaround: remove the test-eprom from the board and start the board again. The cpu will now work without a program, and it will execute all NOP's (Non Operatieve Instructions) so it goes through all its addresses from 0000-0000-0000-0000 to FFFF-FFFF-FFFF-FFFF over and over... With this we can check: all address lines have to move and can be measured on pins 9 to 25 (except pin 21 which is ground) and should give 2 volt. (pins 24 and 25 are a bit less, around 1 volt). Now we've checked all address lines and the chip should work, or we've found the problem. An address line which doesn't move has a short. To test this, bend the concerning pin up, and plug the cpu chip back in. If it works then there's a short on this address line. You can trace this by cutting the address line at several parts and locate the defective chip which causes the problem. Another signal to check is R/W on pin 34 and also afther the buffer IC13 on his pin 18. What else can go wrong: only the selection of the program chip or PIA's.
Chip selection
We continue to work without program chip. Because the cpu goes through all addresses, it will also come across every selection address of every PIA and program chip. The selection pulses come in at these pins: for PIA's check pins 23, 24, 25, 35 and 36 which should give alternating signals (around 3 volt, with a testlight both leds are on, one a bit more then the other but always both on). For the test-ic in U27 check pins 20 and 27 (2 to 3 volt). If there are pins which don't get pulses but always 5 or ground then the selection has a problem. Check the schematics to find the ic where this signal comes from.
The selection continues within the chips and uses lower addresses, but we've already checked all address lines so we're sure they work...
One last case: there's something wrong with one or more data lines. Data lines of the 6802 pin 26 to pin 33 must move and have around 2 to 3 volts. If one is missing look for a short. Don't forget the 6802 has already been replaced so this can't be the problem, the short has to be on the data line... This can quickly be checked by bending the pin of the outputt up and see if there now is a signal out of this pin. If so then there's a short and the buffer is broken, because all signals go through chip U16 which buffers them. Inputs are pins 11 to 18, outputs pins 2 to 9. If all signals go into the buffer but not out of it, then it's certainly broken. It only one or two signals don't come out then there's probably a short on these outputs. With a short you'll have to interrupt the data line temporary and trace the ic which causes the short. This make take a lot of time, but this really is the worst case scenario.........
If you have found your problem and solved it, plug the test eprom back in and check if the test runs fine and all outputs of the PIA's move.
Remark:
Why don't we have to check the addresses when they arrive at the PIA's ? Or the data lines ? Because there are 7 PIA's, it's almost impossible that none of them will arrive, as we check these addresses when they leave the cpu....
Memory test.
When this first part of the test is ok , we proceed with the memory test.The only thing to do is to push on the diagnostic push button. If the memory chip is ok the flashing of the control led will continue, if the chip reads bad the flashing will stop. I did measure the memory signals with a memory that had one pin bent up to simulate an error. So in case of error the signals are to be;
Data pins; pin 9,10,11,13,14,15,16,17 between 1,5 and 2,5 volts.
Address pins ; pin 1,2,3,4,5,6,7,8,22 and 23 around 2 volts, only address A10 = pin 19 is much lower , 0,5 volts.
Futher pin 12 and 20 = 0 volt. Pin 21 = 3 volts, pin 24 = 5 volts, pin18 = 3,5 volts.
If you have these signals it must be the chip itself that is bad.
Extra output tests.
In the new version 5 are included the PIA outputs CA1 and CB1, and an output command at address 2200 , a direct databus to Solenoid drivers command.. This way we will create outputs on all external drivers.
The CPU board includes the drivers for solenoids, display's, lamps and the output for the switch matrix. We can check these right to the output connectors. We are using the well know LED strip plugged in at the various connectors.
The LED strip;
A number of LED's every LED in series with a 470 ohm resistor, and the common to a red wire , that we will plug in the 5 volts. Plug in the connector at the CN that we have to check. If some outputs are missing use the schematics to follow the dancing signal from the PIA ( 6821) passing the drivers to the output connectors.You can use a voltmeter or a logic probe.
Output switch matrix.
Coming out at 1J8 , all 8 LED's should blink.
Display output.
Here we need a LED strip with another ( small) connector. Do not hesitate to make such one. He will serve you for different other boards. Bally lamp driver , Zaccaria CPU board , Data East ,Atari boards and some others.
The LED strip with a smaller connector.
Outputs are on 1J22, all the LED's will blink. At pin 1 the LED will be always " on".
On 1J21 LED at position 1 will be "on " all the time , from 2 to 10 will blink.
Solenoid outputs.
Attention! On the schematic we see that the solenoid ground at 1J13 is also grounded on the CPU board , this is not on all sys11 boards the case. Therefore if you have no output at connectors 1J11, 1J12 or 1J19 ground first pin 1 of 1J13 and try again!!
We find these at three connectors.
1J12 Output on all 8 pins.
1J11 Output on all 8 pins.
1J19 Output on 6 pins.( 3,4,6,7,8 and 9)
Lamp drivers :
The lamp drivers show some differences, this because they are foreseen to work on a 18 volt tension that is not present , on this bench connection.
We find the lamp drivers at 1J6 and 1J7.
To compensate the missing 18 volts , we temporally connect pin 1 of 1J4 to 5 volts. With the LED strip on 1J6 we will see all LED's blink once , then the test stops. To repeat that switch off and on the 5 volts. You can repeat that a few times to be shure all LED's light up.
Remove the temporally connection between 1J4 pin 1 and 5 volts. Now plug in the LED strip at 1J7. All LED's that will light up poorly will light up normally at the rythm of the test .
This ends the complete testing of the output driver circuitry.
Testing the sound section:
The sound section on the CPU board contains a separated CPU chip ( U24) a PIA ( U9) a memory chip ( U23) the sound eprom at U21, the amplifier is U1 and the D/A converter at U2 .
This is only the case with CPU boards of the first type or /A type boards . On the later types /B and /C the sound section is no longer present on the CPU board but moved to a separated sound board. Already on the /B board there is no connector 1J15 ( loudspeaker connector ) present. During the test we will solder the loudspeaker wires directly at the holes that normally have the connector pins.
To test this circuitry we place a new test eprom in U21 .
Find the image here , it is to burn into a 27128 ..... The check sum of this test eprom is : FC3E This same test eprom can be used to test the sound section on a Williams type 9 CPU board.
Before powering up the board, replace the normal game roms at U26 &U27 to avoid conflicts with the running test program. Afther power up check if the test is running at pin 15 of U24, the signal there is constantly changing from 0 to 5 volts. If not the program is not running. Check the basic signals coming to the CPU chip (U24 ).Like the clock and reset , these are the same as the ones at the " normal" cpu so as that one was running ok , these signals must be ok to . So if the test does not run chaznge U24. Now we suppose the program is running , test the output pins of the PIA ( U9) pins 10 to 17 they have to " dance" between 0 and 5 volts. Pins 2 to 9 will not move they are held at 0 volt by a direct connection to the outputs of PIA U10. If the pins 10 to 17 do not move, check first the selection signal of the PIA (U9) at pin 23, we need pulses there . If ok change the PIA . If not check the selection chip at U8. Inputs are at pins 1,2 and 3 you need pulses at all 3. The output is at pin 4. Only pin 4 can be at 0 volt all others have to be a steady 5. If there is another pin coming to 0 volt the input is bad , if all stay at 0 or 5 volt the chip is bad.
I assume your PIA is working fine now, We now can proceed to check the amplifier and D/A circuitry.
To produce some sound, and test both these chips, we need 12 and -12 volt , Use an old PC power unit this tensions and the 5 volt are all coming from such a power unit. The following connections are needed
12 at pin 9 of 1J17. -12 at pin 8 of 1J17, a loudspeaker ( 8 ohms) at pins 1 and 4 of 1J15.( solder) No need for a volume potentiometer, just tie pins 1and 2 of 1J16 together.
At power up if all is ok you hear this ........ Sound like a rapid heart-beat, that is the transistion from 5 to 0 volts ...
If not check first the amplifier, touch with your finger pin 2 of 1J16. A loud hum should be the result.
If not replace the amplifier U1.
If the amplifier is ok and still no sound replace the D/A converter at U2.
Checking the sound memory at U23.
Push the switch " sound" once. The outputs of the PIA come to an halt and the sound will stop for a short while, then restarts , this means the memory is ok. If the sound does not restart the test is running continuously on the memory test. Check the signals at the memory chip.
If the signals are ok , replace the memory chip.
Extra test for input switch matrix.
As explained in the general PIA test, every PIA is tested in "output mode". If a PIA works ok in "output mode" we can be sure the IC is ok. Never the less it is interesting to know that a input signal coming from the normal input connector 1J10 reaches the input PIA at U38. Testing is done as follows. We remove the test eprom an we keep the socket empty.Now we will put at 0 volts one by one the pins of switch matrix input connector 1J10.
Every time a pin is put at 0 volt we check , using the test led, the input pins from 2 to 9 at the PIA U38. When a pin is put at 0 volt the test led should light up at the according pin of U38.
Pin 9 of 1J10 at 0 volt , means pin 2 of U38 goes to 5 volts ,if the test led is connected here he will lite.
Pin 8 of 1J10 at 0 volt , mean pin 3 of U38 goes to 5 volts , and so on for the other pins..
The test led is connected permanently with his negative lead at 0 volts and we test with the positive lead on the PIA pins.