A larger version of the V-block clamp accommodates the 35 mm = 1-3/8 inch octal base of a 5U4GB Full-Wave Vacuum Rectifier tube:
The evacuation tip nearly touched the inside end of the base spigot!
I had to cut the shaft and half the body off the shell drill in order to fit it into the space above the tube base and below the chuck:
A slightly larger shell drill would still fit within the pin circle, but the maximum possible hole diameter in the base really isn’t all that much larger:
The getter flash covers the entire top of this tube, so I conjured a side light for a rectangular knockoff Neopixel:
There’s no orientation that doesn’t require support:
A little prying with a small screwdriver and some pulling with a needlenose pliers extracted those blobs. All the visible surfaces remained undamaged and I cleaned up the curved side with a big rat-tail file.
I wired the Arduino and Neopixels, masked a spot on the side of the tube (to improve both alignment and provide protection from slobbered epoxy), applied epoxy, and taped it in place until it cured:
The end result looks great:
It currently sends Morse code through the base LED, but it’s much too stately for that.
The Smell of Molten Projects in the Morning 
You have a 5V supply, and the 5U4GB takes 5V to light the filament (that’s where it gets the “5” in its name). I’d suggest sending (slow) Morse code by blinking the filament, but that tube takes a whopping 3 amps for the filament (which is more of a ribbon than a “filament”). If you had a spare computer power supply, you could light it up.
I am so not firing a 15 W filament in a thermoplastic socket! [grin]
A 1H5’s filament only draws 70mW (1.4V @ 50mA) but probably doesn’t provide enough glow to be very visible. If I didn’t think you already had plenty of tubes available, I’d be tempted to send/bring you some that would make for interesting light sculptures.
Despite the craptastic LEDs I’m using, I think lighting a filament would be even worse: they’re buried inside all the plumbing, not out where they’d be visible.
Ya never can tell, though. A top light on that 0D3 regulator turned out to be a dead loss, because the mica didn’t reflect any of the colors at all. It’s still blinking Morse and nobody can tell!
Why is support material sticking so far out of the print?
I’d try printing it without support. With cooling fan on, that bridge should come out good enough and it’s not really visible.
Slic3r loves it some blobby support, that’s for sure. I think the lump outside the box provides support for the curved edge, but it’s really hard to tell what’s going on in there. I started designing a support structure and decided to find out what automation could do: it’s barely Good Enough.
The model won’t build without support, because the open side is curved and there’s a big hole in the top: the threads can’t bridge across the entire top. I thought about building it with the ferrule downward, but then the surface around the ferrule needs support. This way, at least, all the support scarring happens inside the box where nobody will ever see it.