The combination reg-rec units I have seen are similar to the Kawasaki and Suzuki regulators of the late 1970s. They consist of a pair of triacs or SCRs that shunt the unneeded current back into the alternator, rather than the battery, hence the term "shunt" (this is an electrical design term). There is another form of "shunt" regulator, a brute-force shunt, that just shorts out the extra current as heat. These types usually have sizable heatsinks to handle that heat. Old Brit bikes have these, as a zener diode shunt. The CX500 Honda also had these, and had pretty short life, as did the alternators on those bikes, from the heat.
The type that are overheating these SOHC alternators are the ones that an electrical designer would call a "Shunt" regulator. I imagine the general marketing industry would call them "solid-state" regulators.
The SOHC regulator system is [electrically termed] a "voltage limiter" system. It monitors the voltage across the coil of the regulator itself (which in this case is the same as Ignition Circuit voltage), using the fixed resistor on the back of the regulator to convert the voltage to a specific current. The coil in the regulator is just a specialized relay that pulls itself open partway, based on the amount of current entering its coil. If it pulls far enough to open the "relaxed" contact, the alternator's field only receives about 1/2 the normal current, so it drops to about 1/2 the normal output power. If the voltage gets too high (above 13.2 volts on these bikes) the coil pulls in all the way and shorts out the field. This makes the alternator output drop to a few tenths of an amp.
The "Reg-Rec" units on many bikes after 1979 consist of a bridge rectifer that uses 4 diodes and 2 SCRs (or Triacs, in some units, which act similarly). These fancy devices act like a diode when turned ON, but act like a high-ohms resistor when OFF. They operate on AC current in this case, and if the Ignition voltage gets above 14.5 volts (typically on these units) they turn ON for the pulse of power from the alternator phase to which they are connected. Their other end is connected to the next winding on the alternator, which at that moment is at a lower voltage than the one being measured (because 3-phase alternators have 60 degrees of AC phase shift from one coil to the next), and it short-circuits the extra power into the next coil. This makes the next coil reduce its output and it "fights back" magnetically at the rotor, which then causes heating in the windings of the coil. Most of the reg-rec units I have seen have just 2 of these phase-shorting SCRs or Triacs, but a few have three as well.
In any case...the field gets turned ON at full current with most Reg-rec units, which is why the output voltage jumps up so high. One solution might be to install one of the Honda resistors from the back of the OEM regulator into the field wire: this would at least reduce the heating overall. The peak voltage would then drop down to something like 13.5 volts instead of the hot 14.5 usually seen. I suppose if I ever get one of these units, I could find a way to test this possibility...
The stock Honda regulators can be adjusted to generate less voltage limiting, which in effect increase the alternator output, too. It would do so without overheating the alternator, too.
...that's why I've started "refurbishing" some of the OEM units: I'm adjusting them for a little higher output, to cope with the [largely] cheaper batteries we are all seeing from China lately. These batteries are coming to Wal-Mart and some auto stores with the Champion label: they have recycled lead in the plates. This makes them store less energy and they "leak" more internally. The result is lower system voltage on vehicles with "voltage limiter" alternator regulators (including old '60s Fords and Chevies!).
