I've been able to exclude it from the network (reports "unknown device removed", roughly). Including it again after that produces no new results.
Based on advice on the data sheet I've even moved the HE physically closer to it to try pairing. No change in behavior.
I'm identifying it for pairing through toggling the attached switch 5 times (or sometimes more), not by pushing the Program Button on the relay. The built-in wires come out the opposite side from where the program button is, which means that in its current installed position, buried in a ceiling box, the program button is flat against the top (back) of the box, completely inaccessible. And several of the built-in wires are in wire-nut connections with 4 or even 5 wires. (A set of three fixtures, plus an unswitched outlet powered from the same feed, with the lights controlled by a switch loop, all the wiring in one box.)
Which is to say, if I actually end up having to dig it out of that box and disconnect everything, I will then smash it with a hammer and throw it out. This shouldn't matter for inclusion/exclusion, and they almost work even (enough that I'm in no doubt it's identifying the right module). It does mean I can't try the "factory reset" step without a lot of trouble (and a hammer).
I'm pretty sure I'm pairing the next module before I install it, even if that means I have it running on a breadboard with alligator clips supplying 120V to it .
Oh -- for local control of the lights it's working fine.
Have you tried rebooting (not resetting) the Z-Wave radio? Sometimes it gets confused during bad pairings.
Shut the hub down cleanly to red light. Power down the hub at the wall, not at the micro USB, which is fragile and can easily break the solder joints to the PC board. After 30 seconds, power back up and boot to green light. Rebooting the hub does not reboot the Z-Wave radio. As far as I know, this is the only way to reboot the Z-Wave radio.
Also, you only showed a screenshot of the bad device. Do you have any other devices that are ghost nodes (no in, out clusters)? If so, they need to be removed.
Just a thought, I see you got the discover button, did you try clicking it and putting the device back into inclusion mode? In theory, it may be able to complete the pairing at that point if the moon and stars align...
I've been through inclusion and exclusion at least half a dozen times
[ETA: Wait, you're suggesting that discover button on the device may do something different from the main discovery path? I did go through it once, and it just takes me back to the main page (and I went through discovery again from there, so if it was doing anything different I did it). Didn't help.]
As I said in the original message, I moved the hub closer after failing to pair it in the initial locations. Since this required unplugging the power supply from the wall and moving it (along with the hub) down into the basement, yeah, I've done that.
No other anomalous devices; everything else is up and working. It's a new, small, setup.
When I initiate discovery (from my phone web browser, since that's easy to carry to where the device is) I get this:
Sitting hasn't helped (but I've gotten a few other things working). I'll probably be pulling it in the next few days and seeing what I can get it to do on a test bench.
I'm wondering if it's worth finding a way to put the antenna outside the grounded metal box the module is installed in. I don't know that the problem is in fact connectivity, it shouldn't be by any range estimators, but I don't know that it isn't. If it works on a test bench and doesn't work again in a box, though, I may try that. Send it out of the box through a hole, and just tape it down to protect it? That should at least be able to communicate one direction fairly clearly!
A switch module is exactly what I want here. I really don't want dimmer functions on this set of lights. It's harmless, but annoying to pay for :-). And all the no-neutral devices are dimmers; installing a switch module up there in the box is the obvious way to do this.
Not literally; it's buried in the back of a ceiling box and wired in with the built-in wires on the module (no screw connects). But pulling the light ring off the box and leaving a couple inches gap there would at least somewhat increase radio access, so, something like that is worth trying. I guess I'll also try to turn it over far enough get at the physical reset button; if I can do that without disconnecting it, that's worth a try.
I'm going to be setting up a small test-bench for some other modules I just bought used, to be sure I understand them, and if I can't get the switch module this post was about working before then, I'll pull it and see how it behaves there. And maybe get it paired before re-installing it.
Also, I've been installing a number of other Zwave modules, and probably yet more today, and each of those at least raises the possibility of happening to be in a position where it can relay effectively; the direction in which the metal boxes leak most is probably not obvious visually. So, maybe it'll just suddenly start working :-).
It is working locally, controlled by the dumb switch, which gives me some hope. And also makes getting it fully online less urgent (having no lights in the laundry room would reduce my approval rating in the local polls!).
I have a similar setup for devices that I'm going to be installing in inconvenient locations so that I can power them up and pair them before doing the install. Generally a best practice to pair in their final location, but there are times that I really don't feel like climbing up a ladder and hanging out over space while trying pair a device (I don't recover as quickly as I used to from a gravity impact event.)
Found this other thread, which recommends rebooting your hub when an inclusion fails in this fashion: Z-Wave inclusion issues
And, I had this same kind of inclusion failure with a Fibaro motion sensor. I rebooted the hub, and then the motion sensor included fine.
Later today, I hope, I'll go back to the test bench and see if I can get the switch module included using that technique. It sounds like from the other thread the problem has been reported, but it still seems to be in the code.