Vacuum Tube LEDs: Ersatz Heat Sink Plate Cap

I wanted a slightly larger “plate cap” to fit a big incandescent bulb and it seemed a fake heatsink might add gravitas to the proceedings:

Vacuum Tube LEDs - large incandescent bulb

Vacuum Tube LEDs – large incandescent bulb

Yeah, that antique ceramic socket holds the bulb at a rakish angle. Worse, even though I painstakingly laid out the position of the heatsink atop the bulb, it’s visibly off-center. Which wouldn’t be so bad, had I not epoxied the damn thing in place.

After reaming out the M2’s filament drive, the entire blue base printed without incident.

A closer look at the cap:

Vacuum Tube LEDs - ersatz heatsink plate cap

Vacuum Tube LEDs – ersatz heatsink plate cap

Memo to Self: Next time, line it up with the vertical glass support inside the bulb and ignore the external evidence.

The boss has a hole for the braid-enclosed cable to the knockoff Neopixel:

Vacuum Tube Lights - finned cap - Slic3r preview

Vacuum Tube Lights – finned cap – Slic3r preview

The cupped surface perfectly fits the bulb’s 3.75 inch diameter. While you wouldn’t mill out a real heatsink, it definitely looks better this way and (alas) gives the epoxy more footprint for a better grip.

I built the fins with a 1/8 inch cutter in mind, so the fin root radius allows for a G3/G3 arc without gouging. I doubt machining a fake heatsink from aluminum makes any sense, but the cheap extruded heatsinks on eBay don’t look very good. Plus, they sport completely unnecessary tapped holes for LED mounts and suchlike.

A cross-section shows the wiring channel and cable entry:

Vacuum Tube Lights - fin cap solid model - section

Vacuum Tube Lights – fin cap solid model – section

I epoxied the Neopixel in place, applied double-sided carpet tape to the whole thing, then painstakingly trimmed around the fins with an Xacto knife:

Vacuum Tube LEDs - Ersatz Heatsink plate cap - tape

Vacuum Tube LEDs – Ersatz Heatsink plate cap – tape

That looked better from the top side (where it was completely hidden) and came heartbreakingly close to working, but after about a day the cable + braid put enough torque on the cap to peel it off the bulb. Obviously, the tape holds much less enthusiastically after that.

Part of the problem came from the cable’s rather sharp angle just outside the cap:

Vacuum Tube LEDs - Ersatz Heatink plate cap - detail

Vacuum Tube LEDs – Ersatz Heatink plate cap – detail

Rakish angle, indeed. Two of ’em, in fact.

Unlike the smaller cap on the halogen bulb, this time I didn’t bother with a brass tube ferrule, mostly to see how it looks. I think it came out OK and the black braid looks striking in person. Conversely, a touch of brass never detracts from the appearance.

Obviously, the cable wasn’t long enough, either. Part of that problem came from underestimating the braid length: it shortens dramatically when slipped over the cable, even when you expect shortening. Somehow I managed to overlook that, despite cutting the cable quite long enough, thankyouverymuch. There’s a tradeoff between gentle angles and having the cable stick out too far for comfort.

Memo to Self: Use a cable at least four inches longer than necessary, measure the combined cable + braid assembly after screwing the bulb in the socket, and don’t epoxy anything before all the parts are ready for assembly.

That’s why it’s a prototype made out of blue PETG…

Protip: running old ceramic sockets through the dishwasher greatly simplifies their subsequent cleanup.

All in all, I like it.

The OpenSCAD source code as a GitHub gist:


, ,

  1. #1 by Jason Doege on 2016-02-12 - 09:08

    Time for new toys! Use that as a mold positive, get yourself a foundry and cast an aluminum heatsink.

    • #2 by Ed on 2016-02-12 - 10:05

      Ahhh, the smell of burning PETG in the morning…

  2. #3 by madbodger on 2016-02-12 - 10:44

    Most plate cap radiators have stacked circular fins (for some reason, vertical fins like yours seem to make more sense), but those would be a deuce to print. Horizontally finned transistor heatsink

    [Ed: I dinked with the URL to show the image.]

    • #4 by Ed on 2016-02-12 - 11:12

      I pondered a pile of transistor heatsinks and came to the conclusion that I didn’t have to be that bullheaded… [grin]

      But maybe I should fake a cylindrical version of a TO-3 heatsink with vertical fins across the top:
      Fischer AKK191 TO3 Heatsink

      • #5 by madbodger on 2016-02-12 - 12:12

        Thanks for tweaking the URL, I keep getting it wrong (wordpress shows it fine in the preview). That’s a truly odd-looking top-mounting TO-3 heatsink. It does look cool, but it would take me a while to code that thing up in OpenScad.

        • #6 by Ed on 2016-02-12 - 13:13

          Perhaps extruding a polygon, rather than hacking solids? Doing it either way with an algorithm would pose a challenge, fer shure.

          I’ve enabled Markdown syntax in comments, but the syntax is (of course) completely different in the editor on this side of the monitor.

          Images should be:
          Exclamation Open-square-bracket alternate-text Close-square-bracket Open-paren URL-linkie-thing Space Double-quote title-text Double-quote Close-paren

          … for obvious reasons, I’m not even trying to figure out the escape quoting rules …

    • #7 by Vedran on 2016-02-12 - 12:53

      These circular fin radiators shouldn’t be too hard to print, as they are not structural parts. I’d simply slice them down the middle, print both halfs with fins in vertical position and glue them afterwards. Should work nicely.

  1. Vacuum Tube LEDs: Knockoff Neopixel Failure | The Smell of Molten Projects in the Morning
  2. Knurled Inserts: Epoxy Anchoring | The Smell of Molten Projects in the Morning