Williams Solved Black Knight (Williams, 1980) Subtle problem with Williams System 7

[SQLGuy]

Hi folks,

Recently I bought a Black Knight machine that was mostly working but had a few problems (a few bad switches, worn flippers, broken bumper wires and skirt, a bit of miswiring from previous repairs, a bad driver transistor, etc). All that's fixed, but game play still seemed a bit odd...

For background: Black Knight has four banks of three drop targets. When you knock down one, it starts a timer, and you have maybe 30 seconds to knock down the other two before they all reset.

On my machine I was finding that the game often didn't recognize when you knocked down the third target within the timeout. However, the switches all made good contact and the diodes and wires were all well soldered. Every time I pulled the glass and tried the targets manually, though, it would immediately detect when the third target was dropped. Finally I noticed the issue: there are a pair of target banks at the upper playfield and a pair at the lower; if you drop two targets from each of a pair, then the machine will not detect when you drop the third target. This is true for both the upper and lower bank pairs. The switch test mode also confirmed this... as soon as the second target is dropped from the second of the pair, the system stops reading the closed switches on the first bank.

I've verified the forward and reverse diode drops of all involved diodes. I've also verified that it doesn't matter which two targets are dropped. The upper pair of banks share a switch matrix column, the lower pair also share a column. The two columns come off different 7406's. I'm guessing that, once there's enough load, the 7406's are no-longer able to pull down the current fed through the combined pull-up resistors going to all the switches, but I can't think of what could have caused this situation. Both 7406's and both 14049's are original.

Anyone seen this before? Any suggestions?

Thanks,
Paul

 

[SQLGuy]

I checked this out with an oscilloscope today. I can see a nice clean signal from the switch, pulled down to about 1V when only one switch on a column is closed. As I close more switches on the same column, the signal gets smaller. By the time five switches are closed, the voltage is only being pulled down to about 3V. Apparently not low enough to read as a "low" to the 14049.

I'm still trying to come up with a theory as to why this would have ever worked. 5V is reasonably within tolerance and clean. Ground seems reasonable too... but I'll probably look a bit more closely at that.

 

[SQLGuy]

Yeah. I think I found the problem: the schematic shows wires (W13, etc) in series with the 7406's and 14049's, but my driver board has resistors there, and these resistors are dropping the voltage that's blocking the system from detecting more than four closed switches.

I looked at the user's manual, and there's a note that there are two possible driver boards, but if the earlier boards (D7997) are used, the series resistors in the switch matrix must be replaced with wire jumpers. Apparently someone swapped driver boards in this unit at some time without paying attention to that note.

I'll pull the board and swap the resistors later, and post an update from there.

 

[SQLGuy]

Swapping the 1K resistors for jumper wires seems to have fixed the issue.