Lionel 6-38043 Frisco Ten-Wheeler erratic behavior

You may need to remove the boiler. I had one where the small metal wire clip was installed directly over the motor. The clip came loose ,dangled and hit the flywheel sensor causing erratic performance. I hope that’s it cause it’s an easy fix. 

Clearly, the issue is in the locomotive and not the tender.  Your symptoms suggest a TMCC reception issue.  I'd carefully check the handrails (TMCC antenna) to see if they're damaged/bent and touching the boiler shell.  Next, probably a swap of the R2LC is in order.

Thank you both so far!

Pardon my inexperience here, but what is an R2LC?  Swapping suggests a replacement.  Is the part available from Lionel?

The R2LC is the TMCC receiver board, it plugs into a motherboard inside the locomotive.

I'd "borrow" one from another locomotive for testing if you don't have any spare parts.  Obviously, since I see a lot of TMCC, I have these parts available, does make it easier to swap them out for diagnostics.

Hmmmm.  I don't have another stock TMCC engine.  We have two others with ERR installations and one Legacy.  I guess that board is pricey?

The ERR R4LC will work in this engine for a test.  It's pin and code compatible with the R2LC in that engine.

I have spares...it's easier than taking the shell off to test and they are not expensive to stock considering the cost of new Lionel Legacy. Something like $40. Besides, you may need it anyway.

OK, here's the latest, which is good news but also a mystery.

I went to the layout tonight wondering what would happen if I switched over to conventional operation.  Well, the engine just sat there as I dialed up the power.  Hmmm.

THEN I noticed that the wire from the Command Base to the Powermaster had come loose.  Ooops.  Hooked it up, and switched back to Command, and the engine runs fine. 

Duh.

But there remains a mystery:  why did it respond to the commands somewhat/sometimes?  I guess maybe the wire was coming off but was not all the way off, but I have no idea what loosened it or knocked it off.

Anyway, thanks for all the help, guys!

While it works best if the wire is securely connected, it's possible for the signal to propagate to the engine without a good or any physical connection. I have run an engine on completely isolated test track on the bench while the base was on and only connected to the main layout.

What he said.  I've run command engines on my test bench with a layout a few feet away with no direct connection.

You both certainly diagnosed the kind of problem accurately.  Again, I appreciate the help.  Even the fact that it behaved better near the control panel is explained.

I find it interesting that the other engine ran just fine.  Apparently, it receives better.

Correct, not all TMCC boards are equal in sensitivity.  That's a common issue, people have one or two engines that have issues on a layout, but not others.  Also, the antenna installation and proper contact with the track & wheels plays a key part in how well the TMCC signal is received.

Makes sense.  This one has always been real picky about clean track and rollers.  The other engine I used will roll just fine despite a fair amount of gunk.