Switch is likely the issue, try replacing it or if you can get to the contacts (unlikely) try cleaning them.
Hood

How could it be the switch if the switch is clearly being activated (causing the software to tell the motor to change direction) and the reset isn't tripped until the Z has moved a distance away from it?
Glenn

By setting a high debounce interval you are making Mach ignore signals that are shorter than that length of time, the higher the debounce setting you have the lnger the signal duration Mach will ignore.
Simple way to see, if your switches are normally closed (should be for safest option) then disconnect the wiring from the switch and short it together, start homing and open the wres and leave them open for a while and see if the limit trips, if it doesnt then you have confirmed its a bouncing switch. Remember to put them together again to stop the axis backing off
Hood

Hood,
The switch is NC and I 've tested it as you suggested.  I guess I don't understand what's going on because the switch works when tested for continuity.  When I start homing with the wires shorted the Z moves toward home and when I pull the wires apart it reverses direction, and when I put them back together again it trips the reset.  I also experimented with different debounce setting with no apparent change.  What exactly does that tell us?
Ever been to Penicuik?  I've got some friends there.
Glenn

Can you attach your xml, will look to see if there is a config issue.
Hood

Here it is.  I'll be done for the day soon.
Thanks for all your time.
Glenn

Dont see anything wrong in the xml from first looks, do you have a spare input on your breakout board that you could connect the Z switch to and reconfigure Mach for it?
Hood

I will have to look into that.  Again, I'm pretty ignorant about the electronics but I'm trying not to be.  I'm using a Bob Campbell breakout board and I've attached a photo of it.  Can you tell if there's a spare input? 

Hood,
A student discovered the problem.  Our Y axis homing switch was damaged and I guess was constantly in open position and causing the malfunction.  The student assumes that when homing the machine just activates one switch at a time so the Y switch break wasn't detected until that switch was activated after the Z had finished its operation.  In retrospect it makes complete sense.  He simply tested by separating the wires from the broken switch and the homing operation continued fine after the Z homed.
Thanks for all the help though.  It's all a learning experience.
Glenn

Glad to hear you found it
Hood