It's taken looooong enough, but I finally got some results to put in front of people. The chart is a summary of results testing the 8 coils I have gotten my hands on. I got most of them from John Vietti. Their pedigree was lost in the transfer, but the pictures in the second attachment may help if you run across similar ones on Ebay or elsewhere.
The testing was done with about the lowest primary current one could ever use. The battery was about 6.2 volts, and there was a total of 5 ohms of ballast resistance (1 ohm of that was a primary current sense resistor). There are two tests on each coil, one with a Rimfire Viper Z3 plug with a 0.012 gap and the other with a CM6 with a 0.025 gap. The next round of testing will use lower ballast resistance, more like one would normally use.
With this small primary current some coils could fire either plug, some could fire only the 0.012 gap plug, and some could not do either. I'll skip the measurement details in this post, but I probably need to write up an explanation so people can understand what I did and can critique it. I thought it was interesting that some of the coils put out more spark energy at this low current level than many of the CDI modules tested previously at full power.
All energy results in the table are millijoules. The energy in the primary is calculated as one-half the [inductance] x [current squared]. Energy in the secondary is the sum of all the samples of spark plug [volts] x [current] x [sample period].
The "mj per ampsq" parameter may sound a bit weird. Since primary energy is proportional to amps squared, the ratio of output energy to input current squared may turn out to be a useful parameter to describe the efficiency of the coil. To be continued . . .
Maybe "mjoule sec/pri" a bit more intuitive. That's the ratio of output energy to stored primary energy.
Anyway, this all is a lot more work than I bargained for. I'll follow up in due course with more tests, but I'd be very happy to hear your reactions.