If all of the switches are open it is unlikely to be the typical diode problem unless one of the caps is shorted or the cap lead is touching part of the switch.

Measure the voltage at each of the columns and see if there is any variance. The symptoms don't exactly match a locked on column, but it's still possible.

A logic probe would be better if you have one.

The column signals are pulsed. If the signal is locked on (or stuck high) it will cause symptoms similar to what you are seeing. The sequence of pulses is what tells the system which column has been triggered. If the column is always high then the game will think every switch in the column has been pulsed, which in one case matches your symptoms.

If the signal is high rather than pulsed, that column will read a higher voltage than the other columns. Again a logic probe is the preferred method of test, but a meter will give some indication of an issue.

The problem at the moment is that the symptoms are very inconsistent and conflicting to some degree. Some symptoms point at a column stuck high, some point at a diode problem and some point at a row problem.

So let's start with the columns and then move through the possibilities.

Have you worked on the game recently, either repair or installed led's?

Nice explanation of the problem, btw.

Do you have a better image of the switch matrix and I'll point out where to test. From what I can see it looks like the rows and columns are laid out differently than wpc (for example), but I can't read enough to tell.

Yes.

OK, let's change our terminology to send (normally called column) and receive lines (normally called row). The send lines are on the left and the receive lines are at the bottom.

Now that I can see the matrix better (plus the additional symptoms you've spotted) I'm suspecting it is a receive (row) problem.

Remove CJ4 and jumper from pin 10 to pin 13 and see which switches indicate closed. This will narrow it down to a board or playfield problem, although it really looks like a board problem at this point.

No problem, best to be careful when jumpering things. Pin 1 should be labeled on the board. Yes I mean physically connect the two pins with a jumper. Just jumper it for a couple of seconds and you should see the switch closure indications.

By jumpering from pins 10 to 13 we are simulating the closure of the Rocket R switch. If you look at the matrix you will see pin 10 goes to one side of the switch and pin 13 to the other side of the switch.

If you get multiple switch closures then we know the problem is on the board since we've bypassed all of the playfield wiring.

This is really odd. It's not the board based on your tests and it doesn't sound like the typical "box" type problem, unless one of the caps is shorted, but then that should show up a a closed switch. I'm thinking out loud here.

Check each of the send (column) wires at the connector for a short to ground. I'm not sure that could cause these symptoms, but it's worth checking. Also check the receive lines at the connector for a short.

If you could pick up a logic probe there are some other tests we can do. I can walk you through using the logic probe.

http://www.amazon.com/Elenco-Electronics-LP-560-Logic-Probe/dp/B000Z9HAP4/ref=sr_1_1

Yes it is. Especially the part where the number of false indications will fluctuate each time the test is done.

Game off, meter set to diode/continuity and check from each column wire to ground and each row wire to ground. Yes at CJ4, pull the plug and test.

Connectivity would be a bad thing.

The switch matrix will either make you a better man or break you. In spite of that I'm sad to see it go away on Spike.

We're kind of left with Robert's suggestion about the caps--not much else under the playfield. I would think a shorted cap would show up as a closed switch, but I could be wrong. Just test across the caps for a short as you did before.

Yes, and the layout on the matrix is not helping at all. I've found myself getting confused a couple of times because of it, and have tried to call them send and receive instead of row and column.

Good idea Robert.

Just radials. I think they should be checked as you suggested though, at this point we're looking for anything to provide some direction.

One lead on each side. Inline is fine as long as the switch is open. When you first put the lead on the cap it should read low and then increase. If it's shorted you will just get a low reading just like you would checking a wire.

Should be good, but I would double check by testing from the ground braid to a ground on one of the boards. You never know if someone has disconnected part of the ground braid. As a general rule I always make sure I've got a good reading before moving on to a questionable reading.

No problem. I'm not sure if the bally boards have test points, but that would be the first place I would look for a ground. If not check the schematics, there will be ground on most of the connectors.

Robert beat me to it, and his answer was much better.

Exactly this^^^

But just in case, make sure the cabinet has a good ground. In the situation above you would be providing a path to ground. The opposite possibility is that the cabinet ground is floating/shorted to something besides ground and you are injecting a signal into the switch matrix.

Also be aware that based on this info when we tested whether the board was bad by jumpering that was an invalid test. I still think Robert is right on, but we now know that we did not eliminate any possible board issues.

I'm just thinking out loud.

It is.

Remove CJ4 and check continuity from the board side of the connector to ground. Check all send and receive lines.

If the problem relates to current flowing through his body to ground, that would not likely have occurred unless when doing the CPU jumper test he was touching the paper clip (or whatever the OP ended up using) and also touching ground. In fact this would be a good test, by the way. I don't mean to sidetrack things, just wanted to mention that our earlier thought may have been invalid based on the additional info.

I'm off on the road for a few days, then at a wedding and then back on the road, so I'll leave it in Robert's very capable hands. I'll try to check in, but no promises.

A short is bad. Some continuity is normal since youre reading through other components. Sorry if i wasnt clear earlier. Roberts right--no problem. Dont worry about it. You, and anybody following along, now better understand the switch matrix.

All of the rows are held high. The columns are then sequentially pulled low. If a switch is closed then the column pulls the row low. When you touch the switch yow are providing a path to ground which pulls the row low. Since your finger does not provide a pulsed signal the cpu thinks multiple switches are closed.