Stuck with Atlas switch machines

It's been a busy work week, which has slowed progress somewhat. But the real culprit has been the problematic switch machines.

I have set up a board with 16 switch controls, wired and joined together. That's 116 screws and 10 sections of wire needing insulation removal and joining to terminal screws. Not that I'm counting. I've connected 5 switch controls. That's 45 wire ends requiring removal of insulation, 30 wire ends to be connected into butt splices, 15 wire ends connected to screw terminals, and 10 labels to affix. Not that I'm counting. Only 8 more switches to go until I can return to the bridge and the upper level.

Unfortunately, I've run into some problems along the way. Remember that all of the Atlas switch machines were tested prior to setting on the layout. I'll need to be more careful in my testing next time. 2 of the 5 that I've connected so far don't work correctly. In both cases they consistently switch in one direction but not the other. I've taken apart and tried to diagnose, but haven't solved the problem yet.

In the past I've set problems like this aside and moved on. Usually that's been a mistake. I really should understand the problem and have a good sense of the innerworkings of the machines before I move forward deploying 50+ more of these, because I may find that the best corrective action is something preventative that I can do before deployment.

One really frustrating thing is that these things get so hot so fast. The instructions are clear about never holding the button down for more than a second, and I never do. However, when testing anything I'll typically repeat the operation for several iterations to see how it behaves -- but if you do that with the Atlas switch machine you can literally melt the plastic encasement, as I found with the 4th switch machine I connected.

So, when I get back to the layout tomorrow my sole task will be to figure out what's wrong with these two faultly machines.