I keep tinkering with this. Some new results.
Measured the 5V supply from the solenoid driver card (Alltek board in my case). It is rock solid, no changes. I think this eliminates the Vcc (5V) sag at least in my machine. The factory Bally card may be different especially if someone's 5V has drifted over the years (change those bad caps).
I was given a sniffer coil to try out. It is one those coils you used to put on older phone receivers to let you record phone calls. It is a very small coil with no core. I attached this to my scope and started sniffing. Held it up to the transformer and got a nice strong 60Hz signal as a baseline. The scope itself created a big high frequency signal at what I believe is the Atmel controller internal clock frequency. All these things drop to near zero more than 1inch away. Most of the space under the playfield has nearly no measurable magnetic fields, you have to be up tight to another wire or coil to see anything. Unfortunately that is why these machines have problems. The original design was to tie wrap everything together.
Then I held it up to a flipper coil and fired it a few times.
1.jpg
This was the result, keep in mind this is the result of a magnetic field interacting with the coil I am holding. This means the trace is the rate of change of current flow in the sensor. The actual numbers are not directly in the pinball machines circuits. These traces varied a lot, some looked like mounds, others were decaying sine waves.
2.jpg
Then I held it up to the bumper coil.
3.jpg
Here is the result. Although lower by a lot, in terms of signals transmitting over distance, pinball coils act like excellent repeaters. Because Bally chose to ground coils to fire them, they all have the same 43 V power at the same time. Ripples in the flippers cause ripples in every coil and the large magnetic field pulse happens at every coil, they look pretty much the same all over the machine.
4.jpg
So, once again I pulled the coil fuse on the playfield.
5.jpg
This is the result. A hugely reduced spike made it to the coils.
6.jpg
What this seems to mean is that although I got a huge improvement by creating dedicated wiring to the flippers separated in space from the main harness, the remaining crosstalk is still conducted around the playfield by the power lines to the coils which of course are all over the place and need lots of wiring that inductively couples into the switch matrix wiring.
Conclusions:
- To fully fix this I would have to power the flippers from their own 43 V source that is not split off from the coil circuit as I have now.
- This also means putting shield or braid around my new flipper wiring will not help since this is not a radiated problem (nothing is being broadcast across the inside of the machine.
- The EOS spark could broadcast some noise but I am not measuring anything from the sparks (at least not is the frequency range of my little scope).
I have one more thing to try but I am pretty close to the end of the road on this.