He's pretty close to the truth, too!
In the mid-1970s I analyzed all of this and hand-built a few transistorized ignitions for the CB750. Mine had one (until someone stole it off my bike one nite at a motel
) and my friend that I learned CB750 with at school got the other one.
The CB750K2 (and probably later ones as well) system draws 4.1 amps per coil, steady-state with new points and 13.6vDC battery voltage (not in the bike: see below). The open-point back EMF, measured at the points without a condensor, is just under 400v peak. With the 0.24uf stock condensor in place, it reaches a max of about 48v at the points' connection: this 10:1 PRV reduction is typical of Kettering system designs. Coil collapse without the condensor produces 5kV peak plug voltage in open air: with the condensor it reaches a little over 7kv in open air, with 2 plugs on the coil. The condensor lets the coil collapse faster. When the condensors get old, the capacitance drops: at about 0.18uf the points burn quickly, too.
The single black wire in these bikes that feeds both coils is approximately 23 AWG in cross-section. This limits the dual coils' available current, much like the resistance wire in Detroit's Kettering systems. It typically measures close to .25 ohms between the battery + terminal and the coil terminals where the bullet connectors plug in: those connectors are also quite high resistance. The result overall is about 6kv at the plugs at 3.6 amps coil current with fresh tuneup parts on a nice summer day, with good battery. That's the stock setup. After about 1000 miles, the coil current drop to about 3 amps, as measured with my trusty inline Borg-Warner racing ammeter. For reference, Honda required 7.5kv from a "good" coil in their testers, which ran them at 14vDC with a 4000 RPM simulated mechanical oscillator running the coil.
When the 3 ohm coils are added, the solder-spliced connections in the Honda harness that feeds the black wire to the coils can heat up and crystallize the solder, causing low voltage in many other places downstream. That's why those coils should have a separate feed wire from the key's circuit, or through a relay that is run off of the old coil V+ connection. I recommend using LED tailite and lower-wattage headlight (no more than 55w high beam) when running these coils, too, lest the system voltage suffer overall, and if you have a K3-K6, remove the front running light circuit from the turn signals. These coils will draw a little over 4 amps apiece, which is 25% higher than Honda's security blanket.
In a nutshell, that's the stock system's dynamics.
To make a transistorized unit that will run from the points: use a PNP transistor rated at 80Vceo with the highest Vbco you can find. Gain (Hfe) should be low. Use Honda's own condensors for suppression. Close the gap on the points to about .010" (you'll have to slot the points plate to get the timing in position at this gap). Add about 100-300 ohms inline base resistance to the points and bias to ground to suit the Hfe you have, tune the rise with capacitance you like (I think mine was 0.022uF) and locate the module on the rear fender, under the taillite, with appropriate heat sinks. Use larger (16 AWG) wire to feed the coils from this module, and have fun: when mine was stolen off my bike, the points had worn the blocks off so far that I almost couldn't get .012" gap back on them to finish my trip home. The faces were still a mirror finish at that point, and they probably had over 20,000 miles on them.
Toyota, from 1972 through 1980-something, had an "Igniter" module that assisted the points in their cars. These also work nicely on Yamahas, I have been told, by adding one junkyard unit per set of points to your bike. I suppose Hondas wouldn't object, but I've never tried one...
