A bump just to say I think I have this problem licked. Finally.
Figured the problem had to be in the long-throw optos, and they seemed to be firing properly but they didn't register 100 percent in test mode. The switch would stay closed at seemingly random times, which would suggest it often thought there was a ball in the right or left gate when in fact there was none. So I held off on pulling the transmitters and receivers.
Decided to remove the 24-opto board and have a look. Sure enough, the inductor had a broken post on it. It was just barely broken, so depending on the vibration, it could have been making occasional contact - enough to function every now and then but not all the time. Never would have seen it without taking it out.
The original board is on the left. The inductor is the mushroomy component just up and to the left of the bottom-most screw in the picture. With the board out, I shut off the mist feature (which I should have done a long time ago), and sure enough, the recount problems disappeared.
At this point I have to give major props to Mike at homepin.com. He has re-engineered 24-opto boards and got one out to me - see the above pic. Popped it in, it works great, and the bottom line is the miscounting problem has stayed away. Haven't had a miscount since we pulled the old board last week, but of course time will tell. Absolutely fantastic customer service, as well. Highly recommended.
Plan to ship the old board off to Coin-Op Cauldron when I refill the machine fund.
Thanks to everyone here who contributed to getting the problem fixed, especially The Count! It was much appreciated and I learned quite a bit!
