Thanks for the considerable energy expended here. To correct any assumptions: From transformers to TIU and TIU on (buss wires) all is 12 gauge copper stranded, there is some 10 gauge as well for a few segments. Drops are frequent and all 16 or 18 gauge to inner and outer rails, with outer rails tied together equally often at drops. I use no isolated outside rail trigger circuits, so all track has both rails contributing to the circuit. Track (GarGraves) has double soldered pins, the usual ones plus 12 ga bare copper inserted in the bottom opening of the rail ends and soldered, as well. TIU to track drop is less than 2 volts at the farthest stretches from origin. The problem is the power supply internal resistance mentioned in Vernon's post above. It seems to drop those motors right into their critical narrow speed variation range.
Like it or not (I'm fine with it), I have a number of "sow's ears" from this standpoint that are silk purses for operations to the railroad's CEO. The CP-now-C&NW custom F3 set with 4 Pullmors, the Pride Lines City of Denver with 3 open frame prewar-style motors and 2 ERRCo command board sets running them, 3 K-Line TMCC Trainmaster chassis with C&NW renumbered Lionel shells (can motors but no cruise control), and I'm sure other locos I'm forgetting in the inventory that will be brought into action as operations increase. To fix each of those locos with remotoring and Cruise Commander boards, or just Cruise Commander M boards for DC motors, involves much cost and labor and no little aggravation, from considerable experience doing it. Even if it penciled out slightly less costly than the low-drop Variac model I mentioned, the hassle would still make the Variac a better bargain. I already know from tweaking the voltage while running that bringing it up within 1-2 volts of baseline solves the problem of extremes of race and crawl speeds to an acceptable level. Getting rid of the 3-4 volt drop at the power supply posts will substantially solve the problem. If I haven't made the point clear why I'm chasing more stable source voltage, it should be now.
I am interested in the isolation transformer safety point above, as long as that transformer doesn't have its own internal resistance problem, however, I suspect it may and it wouldn't take much to defeat this strategy with the 0.8 volts already expected as load-related drop for this model of Variac. Otherwise, I will make the Variac safe by limiting the travel of its dial internal to the cabinet to a few volts above the 19v desired. A second-level redundant safety in the form of a voltage-triggered breaker is the Zener diodes already across my TIU inputs for spike protection, which will fail in dead short (they have 37v ratings as I recall) with overvoltage, then tripping the transformer's own breaker (12 or 15 amp placed in the cabinet where the two Variacs would live) and preventing unsafe voltage from reaching outside the cabinet. I have plenty on hand, so will probably add Zeners at the cabinet outputs right outside the rapid breakers so high voltage can't escape even for a moment. No, this will not be UL listed, but input as to the safety of the strategy is welcome here. I have grandchildren, too.
I'm also curious whether Vernon can say for "that big L 620 Watt ZW-L" that my 6+ amp 4-motored train loads won't pull its output down the same 3-4 volts. I'm not aware of the technical differences. My large "The MAX" is unlikely to have lesser transformer windings or greater internal resistance than whatever is in the ZW-L, but I'm speculating based on how cool it runs regardless of load. So if actual measured volts/amps say differently, please advise.
I realize most of you wouldn't be in the same situation, so advice in hindsight that would have altered loco decisions decades ago will not help now. Advice about how to do this and do it safely will be helpful and appreciated!
