Doug : I agree. The ECM is doing exactly what it is programmed to do, but not what it should normally be doing, if that makes sense.
I see what you said about the SECOND coil trace pulse being a bit less wide than normal. But look at the top coil trace. When the second trace is shorter, the top trace never goes low, but there seems to be some ringing, only. It looks to me like what is happening at the top coil is causing the signal at the second coil to terminate. So why isn't the top coil being pulled low by the ECM ? Since the ECM was swapped, we know it's not a faulty ECM.
It looks like the ECM may attempt to pull the line low, but when no primary coil current is detected, the cycle is terminated for both coils. That may be why the second trace goes back high prematurely at that point.
And the interesting thing is that there is no trouble code for "no coil current" or "open coil". So even if a fault is detected, there isn't a code set for it.
So, here is a question for Jeff :
When the second set of coils were swapped in, where only the individual coils swapped, or the entire coil assembly including the wires going to the coils and the two 2-pin connectors ? If only the coils, there could be an intermittent open in the top coil negative wire ( in the traces ) that is occurring at only a certain frequency ( RPM ).
There is a report of a ST1100 with a similar problem.
See :
http://koczarski.com/mmartin/Intermittent.htm
"A close inspection of the harness revealed a break near the connector to the coil
, but inside the insulation."
BTW, I had an old Moto Guzzi with a Dyna ignition that had a miss at only certain RPM's. The problem turned out to be a bad solder joint inside the pulse detector between the Hall Effect device lead and the wire that was suppose to be soldered to it. It was fairly easy to troubleshoot down to the pulse detector by switching wires around for/between the two cylinders. When I went to replace the Hall Effect detector, I found the bad connection. I simply resoldered it ( didn't replace the sensor ) and it ran great after that repair. The connection was buried in epoxy.
At this point, just for the heck of it, if an intermittent open in one of the coils wires isn't found, then besides disconnecting the instrument cluster connector, ( that was suggested before by a poster ) I would connect a 0.47 uF cap at the Service Check connector between the +12 volt pin and Ground pin, just for S&G's.