Williams System 3-6 Sound Board Repair

12 December, 2010 (14:48) | Pinball | By: admin

Earlier this year I did some work on a Williams System 3-6 sound board for a friend. At the time I thought I would document the repair and also repair a few more of these boards that I have awaiting repair and do a more in-depth analysis of this board but I never got around to repairing the other ones and probably won’t get to them any time soon (no projects currently needing this board) so I figured I would share the original repair and some general information about this board.

I’ll skip right to the good part and reveal that the problem I found with this board is that someone had removed the key pin from the power supply connector and connected it to the sound select inputs header. I don’t know this for a fact but the owner did tell me that there was no key on that connector and it seems pretty obvious based on what blew up.

The first thing you’ll need to troubleshoot one of these boards is a clear copy of the schematic. Looking at the schematic below we can see that it’s a fairly basic Motorola 6802 based computing system. Not unlike an MPU.

In order to test the sound board out of the machine you’ll need a power supply capable of supplying +12VDC and -12VDC. I use an old AT power supply for this but most computer power supplies should be sufficient.

These are the specs for the power supply I’m using:

Below we can see the schematic for the power supply section of the sound board. In the machine the board is supplied with 18VAC and the voltages needed to run the board are created by the on-board power supply (+5VDC, +12VDC and -12VDC). The power supply is exactly what you would expect in a linear power supply. A rectification bridge, some filtering and a voltage regulator.

To test the board out of the machine we will supply +12VDC and -12VDC to the pins which originally accepted 18VAC and of course ground will be connected to the power supply ground. The +12VDC and -12VDC can be connected in either polarity across pins 1 and 9 with ground connected to pin 5.

Follow these steps to connect a power supply to test the sound board:

1. Connect +12VDC to connector J1 pin 1
2. Connect ground to connector J1 pin 5
3. Connect -12VDC to connector J1 pin 9

This is what the connections will look like. In this image +12VDC is yellow, Ground is black and -12VDC is white.

Once the power supply is connected you can check the voltages produced by the power supply. I like to test these voltages at the filter capacitors. Unfortunately there are no test points on these boards. With a voltmeter set to DC and the black lead connected to ground you should see approx. +5VDC at the + side of C27. The +12VDC and -12VDC should be approx 1V lower than the supply voltage. The board will work with these voltages as low as 9-10VDC or even lower. The +12VDC can be checked on the + side of C29 and the -12VDC on the – side of C30. Remember that the filter cap for the -12VDC is filtering a negative voltage so the + side of the cap is connected to ground unlike all of the other caps in the power supply section.

The AC ripple across the filter caps should also be checked. There are a couple of ways to check this. The best way is to put an oscilloscope across the voltage and observe the ripple. Another less accurate way to check for AC ripple across a filter capacitor is to put a meter across it set to AC. There should be very minimal AV voltage (no more than about 0.1VAC). If the capacitors are original it’s a good idea to replace all of the electrolytic capacitors on the board. This will also improve the sound quality.

This image shows the locations of the capacitors mentioned above.

Now that we know we have good power coming out of the power supply we can check the basic things required for the CPU to start running code. The reset and clock circuits. At this point we’ll need a logic probe. The reset pin of the CPU (pin 40 of U1 – Motorola 6802) should go low for a fraction of a second then high when power is supplied to the board. If the reset does not start low and then go high the CPU will never start running.

If the reset is either staying low or never going low I would suspect either one of the transistors or the zener diode have failed. The board I’m working with does not have a reset problem but if it did I would most likely install a Dallas DS1811 reset IC in place of the original reset circuit.

Here is the schematic of the reset section:

The next thing required for the CPU to boot is a stable clock signal. The clock circuit on the sound board consists of a 3.58 MHZ crystal oscillator connected across pins 38 and 38. Below we can see the schematic of the clock circuit. These 2 pins (38 and 39) should be pulsing when checked with a logic probe.

Here are some examples of what the clock signal pins will look like with an oscilloscope.

Pin 37 (enable) should also be strobing. This is what it looks like:

Once we’ve checked the reset and clock circuits we can assume that the conditions required for the CPU to start running are satisfied. At this point you’ll want to take a look at the CPU socket. There is nothing worse than chasing intermittent problems caused by a flaky socket. If you have any doubt about the CPU socket just replace it. More and more I’m finding that these old sockets are just not cutting it any more. This logic applies to all the sockets on the board.

Now we can install Leon’s test ROM and test the rest of the board starting with the 6820/6821 PIA. Note that the sound board can be strapped for 2 different types of ROMs. The sound board will be strapped for either a 7641 PROM or 2716 EPROM. In order to use the test ROM we’ll need to jumper the board to use a 2716 EPROM.

More information on Leon’s test ROM including his own repair guide for these boards can be found here.

The picture below shows the jumper settings for a 7641 PROM.

This picture shows a 2716 EPROM installed but the board is still jumpered for a 7641 PROM.

These pictures show the board jumpered for a 2716 EPROM.

Once the test ROM is installed and running the following PIA pins should be strobing from low to high once per second: Pins 2-17, and 19.

In my case I found several pins were stuck high. I had a Bally MPU handy so I swapped the PIA into the Bally MPU and tested it there with Leon’s test rom and found the PIA to be good. That made me suspect at least one of the MC4050B buffers was bad (IC8 and IC9). I checked them both with a logic probe and found a bunch of outputs stuck high. This is when I started thinking that someone had connected the power connector to the sound inputs header.

I desoldered some resistors and caps to isolate the buffers for testing.

At first I though only one buffer was bad but they both turned out to be bad. Here we can see one buffer replaced and the other one being tested.

Fresh new buffers!

Looking at the schematic below we can see how the sound select inputs work. Each of the sound select inputs are connected to a PIA I/O on port B as well as one of the 8 inputs on a CD4068 8-INPUT NAND gate. If any of the inputs go low the output of this gate will go high sending an interrupt to the CPU via the PIA. This is how the CPU knows that it’s time to make some sound. The interrupt signals the CPU to make sound and the CPU decides which sound to make based on the inputs which are buffered through IC8, IC9 and ultimately the PIA. This logic is all pretty basic and you can troubleshoot it with a logic probe.

The next step is to connect a speaker and test the sound select inputs and audio outputs. To do this we’ll need to connect a speaker across pins 1 and 4 of connector J2 and install a game ROM. Some of the earlier games had no self test in the game ROM but others included a self test which would run when the switch on the board is pressed. For testing it’s best to use a game ROM that has self test code (I’m using Phoenix in this example).

Once the speaker is connected and power is applied to the board pressing the test button should cause a bunch of sounds to be played through the speaker. This will test the PIA outputs to the D/A converter, the D/A converter and the audio amplifier section. In my case all of these things were working.

Note that there are 2 versions of this board. One that has connector J4 and one that doesn’t. On boards that have connector J4 a jumper will have to be connected across pins 1 and 2 of J4. This connector is also a handy way to isolate the power amplifier from the pre-amp in the amplifier section for testing.

This picture shows a speaker connected to the sound board with alligator clip leads.

All of the sounds that come out of the board are generated by a MC1408 8-bit multiplying D/A converter. The CPU drives the inputs of the D/A converter through the PIA and the output is then amplified. That’s pretty much all there is to how the sound is created.

Once we know the D/A converter and amplifier are working we can test the sound select inputs. Grounding each of the sound select inputs one at a time should create a sound. Some games do not use all of the sound select inputs so you may need to consult the schematic for the game in question to see how many of the sound select lines are used.

This schematic shows the D/A converter and audio amplifier.

Once the buffers were replaced I was able to test the rest of the circuit and everything seemed to be working fine. I let the board run for a while on the bench and found that it would stop working once in a while after a few minutes. After some troubleshooting I found that when the board would stop working it was because IRQ (CPU pin 4) was stuck low. The cause turned out to be the output of the CD4068 sticking high occasionally causing the PIA to constantly send an interrupt to the CPU. I figured that either one of the capacitors across the sound select inputs was shorting intermittently or the NAND gate was bad so against my better judgement I just shotgunned those parts.

I set up my logic probe so it was resting on the interrupt pin of the CPU so I could monitor it over time. Sure enough it would stick low after running for a while.

I did a few other things like replacing sockets but for the most part that should cover the repair of this system 3/4 sound board. I also consumed a lot of these:


Comment from Richard
Time February 20, 2011 at 7:23 pm

Great write up on fixing the WMS sound boards…

Any chance you can help with a question that keeps coming up on RGP? Can you use the old style PROMs for a Williams “Type 1″ “rectangular board” in the newer “type” 2 square board? Seems like either the prom or a 2716 Eprom with the same ROM code on it don’t work. I’ve tried changing the jumper settings and am fairly Ok with the boards.

Seems the Type 1 “rectangular board” can be jumpered for the factory Prom OR a 2716 Eprom. The type 2 “square board” can run Eproms of various types (2532, 2716) but although there is a setting for a Prom (Big Strike, a shuffle bowler with sound only) it doesn’t work with a pinball Prom (from my FLash which works on the earlier board)…

I tried it with those jumper setting and my Flash PROM… Seems like it should work in theory… but I wasn’t smart enough on the day to figure it out.

I’ve looked in some detaila gain tonight, and believe it comes down to the memory addressing. Even if you made an adaptor for the PROM to pin it out like a 2716 (which is possible) and using CS1 / CS2 as the chip selects- I still think it would still fail. You would have to do the hardware changes and still likely to be stuck with a programming job.

Any ideas? -Richard

Comment from admin
Time February 27, 2011 at 9:47 pm

Hey Richard,

Thanks for the comment! You inspired me to solve this problem and I’ve posted the solution here:


Comment from Richard
Time February 28, 2011 at 8:04 am

You are the man. Do you havr the ability to bend spoons with your thoughts, as well? :o ))

The schematics and jumper settings were doing my head in. I knew there was a way to do this, because the boards are actually quite close in design.

I usually prefer a hardware solution, rather than change code. It’s a shame it’s such radical surgery rather than a ROM adaptor made up of two chip sockets or something like that. But at least you solved it for the rest of us, and it works. There are plenty of the “square type 2″ WMS sound cards out there, and personally I think they are more reliable and sound better.