I'm back to it. Having recently struck upon what seemed like a reasonable means of increasing the gear reduction, I launched into the modification project with high hopes. The concept: add another reduction gear by splicing two gearboxes together.
This turned out to be much more viable than nearly any other idea I'd entertained up to now, most all of which involved worm/worm gear sets. While I have some mighty small worms (from Mikroantriebe), none of them had the right pitch.
The gearbox expansion project got off to a reasonably good start; it appeared there was just enough play in the middle compound gear for another one to engage it, so I chopped up two gearboxes and shaved their sides by nearly microscopic amounts until the gears meshed.
I knew going in that there would be no way to simply glue the two gearboxes together: they're made from a type of engineering plastic to which nothing will bond. But I continued working, confident that an idea would pop up eventually.
Sure enough, an idea emerged, a kind of brute-force approach: using CA and splicing plates the full size of the gearbox sides to produce the largest bonding surface areas possible, in the hopes that it would compensate for an otherwise very weak bond. Attaching thin stainless steel strips cut to fit the gearbox with CA, I began to suspect that my crazy idea might actually work. It was strong enough to survive the stresses of replacing the gears, which required considerable flexing of the joints—the acid test.
After a bit of trimming, fiddling and fitting, I managed to get the gearbox working nicely. The next task was getting it to stay in place on the chassis without the shell, since that was going to be replaced anyway. I found that the direct approach worked quite well: I simply glued the gearbox right to the chassis; thus the gearbox itself served to splice the two chassis parts together. To ensure that I hadn't bungled the project, I applied power directly to the motor, and smiled with satisfaction (tinged with some relief) as the gears all spun merrily.
At this point I began learning some valuable lessons. My plan was to eliminate the contact springs, and hard-wire the truck electrical tabs right to the motor. But as I replaced the powered truck—which required some twisting forces to be applied—I discovered that the springs do more than conduct current... they also hold everything together and keep things properly positioned. Without the springs, the pickup parts inside the truck sides are free to pop out the top. Furthermore, the truck can slide side to side, instead of just rotate, and the springs keep it more or less centered.
And at this point I started becoming progressively more impressed with the designers of these "toys." A remarkable amount of very clever engineering went into their development. They solved a great many problems that I have no hope of addressing myself, even as I struggle to come up with entirely new drivetrain designs. So, I temporarily attached the original metal strips installed in the shell directly to the chassis so that I could make a test-run. And that's where things began to go south quickly.
First, the unit would run only in one direction; even though it tested fine on the workbench, something was binding when the throttle was reversed. Worse, the direction in which it did run was with the power truck facing forward, which resulted in some slippage. Plus there was a tremendous amount of vibration; the unit literally shook itself off the tracks a couple of times.
Hoping that I could spot the binding problem, I picked the unit up off the track and promptly dropped it on the floor, whereupon it shattered into several pieces (which is why there are no more photos). And that brought an end to this project. Quite a contrast to the wildly successful switch project!
Although the disappointing experience has not destroyed my plans to try again with a different approach, it certainly set me back. Convinced in advance that I'd aced the drivetrain modification problem, I became aware of the need to reign in my hubris. It also further deepened my admiration of the designers: they overcame a great many obstacles, and I was even more amazed that the things ran at all.
So, what's next? For the short term, I'll be looking to address performance issues with a pulse width modulation type throttle, which has been demonstrated to get the minimum speed down to a satisfactory rate.
Ultimately I'm thankful for what I'd sometimes felt was an unnecessarily large collection of T Gauge trains (three full sets); if I didn't have more than one set, I'd have nothing left that was functional, until another order was filled. At least I still have something to run!