Hi currieddog. There is a very common (yet incorrect) term around here that I considered factual of "a coil either works or it does not", which I then in my pinball experience found to be incorrect when my 1962 Williams Vagabond had a 10 point score reel coil was burnt just enough to sever the connection inside of the coil, and due to the crazy vibration, it was arcing internally and allowing the coil to work sometimes, but not all, and rather than meaning days or months apart this sometimes was within 2-second 50 point pulses. Seriously, that frequent. I would get like 30 points out of the 50, or 10, or 20, whatever. No rhyme or reason.
I wonder if you may be experiencing a "freak" issue of the same thing.
I have an idea in mind that can diagnose both THIS problem and MANY others, including nearly all functions within the score reel.
You should go out and buy a cheap ANALOG multimeter, and hook it to the two lugs on the score reel. Honestly at this point before going any further you need to verify if it is your score reel that is not receiving the correct instructions from the rest of the machine, or if the machine is telling the reel to do the correct thing and it is failing to do so. An analog multimeter (in contrast to a digital) uses an old-school meter (they still make modern ones) which you can watch with your eyes which allows you to see quick pulses with your own two eyes. In theory a digital multimeter might do the same if the signal was long enough for it to process, but would be very easy (or even not impossible) to miss especially when it measures every voltage between 0 and 30 on the way there as it rises and drops, especially with 5 pulses instead of 1. So that is out the window and using one of those would give you the worst results known to man and get us nowhere. Analog meter it is!
----
OR:
As described in my EM repair manual, I actually had a similar issue to this yet again except on a ball-count unit (which also requires many pulses) on my 1969 Gottlieb Skipper. I invented a little procedure that goes a little something like this. In hindsight, I think I invented this earlier on my exact Vagabond problem mentioned above, so I guess I used it to fix that too! I hope it helps you well. Instead of using an analog multimeter (which I don't own), I simply hooked an alligator clip test lead up to the score reel next to it. This would force the voltage provided to the reel (or in my case, ball count unit) into the next one and you can see if it matches up to the actions of the other one like a copycat, and to even go a step further since they're beautifully numbered score reels it's kind of funny, because instead of counting the pulses, you can just compare the two printed numbers and see if they match up (as long as you start them in the same place)!
Essentially, you're copying the data the problem reel is receiving and transmitting it directly to a different reel by force at the same time. This will allow you to count the pulses on a reel that you know is working properly, and thus distinguish the problem as the score reel, or instead the thing that controls the score reel. If both score reels are only adding 100 points for a 300 point pulse, you know the problem is elsewhere than the reels.
One word of caution, on multiplayer machines I usually hook the reel to one of the other player's for one simple reason, if you run the reels over to the next place (i.e. 700 -> 800 -> 900 ->.... 1,000!) you will add another "1" pulse to the digit you are using to test which will mess up your results. Even if you hook it to the reel before itself, it will eventually roll the problem reel as well, they'd both do it to each other and unfortunately you only have 3 reels to choose from with your problem reel smack dab in the middle of both. So just keep that in mind when utilizing this method and don't go getting yourself confused. The whole "rolls over to the next place" circuit could cause a lot of mental problems and confusion. Best case, test it with all the digits at 0 so you don't even have to open that bag of worms. One time will tell you everything you need to know, so you'll stay far away from 1,000/100.