Tuesday, October 29, 2019

3D printing a LED driver enclosure - 2 - Work in progress


Long ago 😅 I said I was planning to design my own enclosure for a led driver and led controller. Of course, reality interfered and lots of other things got in the way.

Still, I made some progress. And hit a couple of snags on the way...


Yes. I am perfectly aware I could have bought something like this:


But where's the fun in that? Besides... it wouldn't allow me to add a 'led controller' inside the box. And it's a transformer, thus probably higher losses. It might have made sense cost-wise and certainly time-wise but...  well... you know...

(Also, I'm pretty damn sure these were not separately available when I went looking for one. Unfortunately, I'm also pretty damn sure I sometimes make mistakes, so... 😉)


For the basic concept little has changed. The only thing I've changed is the addition of a fuse (not entirely sure it is required, but I still think it is a good idea).

Designing something, even a little box like this, takes a couple of iterations. Every time you think you have it all figured out something else comes along.

Oh, there must be a hundred ways to do it better, and I'm pretty sure I've completely overdesigned this, but part of the project is to learn.

Fusion 360

I suspect the fastest way to design this was a. measure components, b. build models for these components, c. subtract them from the final object, d. subtract a little more. Sketchup would have been great, but unless you pony up it doesn't offer Boolean tools anymore.

I tried different packages / design tools, and settled on Fusion 360 for now, which is (still) free for private use.

Once I was pretty sure the basic concept was right and had some sketches ready I set down and started to learn Fusion 360. Which works well, most of the time...

Simplify 3D

The best things in life are free. Unfortunately, that does not always (never?) apply to software. Fusion 360 is a great tool for designing a box, but I also needed to 'slice' the model and feed it to the printer.

The original Dremel software was okay, but I found Cura to be better. Then Dremel released a new version of their own software which is actually a descendant of Cura.

I also tried Simplify 3D, a commercial alternative, and I must say I like it. It isn't cheap (149 euro I believe) but it's very easy to use, very fast, has great support for the Dremel 3d20. I'm impressed. Now I have to decide if I want to spend some money on it, or not...


Okay, I got distracted. Played a little Dungeons & Dragonswrote for Dapper, modified some 123-3D printer spool holders, tried to make my Dremel 3d20 more quiet, went all nerd so I could monitor the print process from my living room... Needed to figure out how to connect the led lamps / DC wiring, learn about crimping...

And in between try to find a new job. And a car. And a life.


Design considerations

  • As small as sensible
  • No solder or crimp connections on the AC side.
  • Cooling / heat
  • One inline fuse
  • Strain relief
  • Cable entry should be possible from the top (mounted on top of a flush mounted installation box) as well as side
  • Enclosure should be 'complete' before installing, however it should be possible to remove one or more components for easier access to the wiring behind
  • AC and DC as much separated as possible
  • Any other safety concerns

Proof of concept

I actually printed a proof of concept, improved it, reprinted it, etcetera. My ol' Dremel 3d20 was quite busy as these things are (as 3d prints go) fairly large, at least for the limited print bed of the Dremel. Here are three iterations, printed in black PLA, each is 132 x 132 mm.

Power consumption

I want to deploy three enclosures, each with its own LED driver / Milight controller. 2 upstairs,  each 12 V, 1 downstairs running 24 V.


The individual spots I've tried range from 2 to 4.5 W. The best ones were 4 W.

5 spots per rail, each 4 W, totals 20 W at 12 V.

The Meanwell APV-35-12 delivers 35 watts and seemed a good match, perhaps a smidgen too large but I could not find a better candidate.




Downstairs I am going to use 10 times an Ikea 24V Omlop spot.


Why? I needed absolute minimum depth due to the construction on the ceiling, as well as very little heat. I didn't plan any spots at all in the beginning, they were a kind of afterthought / emergency measure... More about that some other time...

As for life expectancy: Ikea claims 25000 hours. At 3 hours per day that's 22 years, but in real life it's probably going to be closer to 8 hours a day, so let's say 8 years... As these are non-standard lamps I would need to replace them with some other product, or redo the complete ceiling, in 10 years (I'm pretty damn sure Ikea is not going to carry these LED lamps for umpteen years). But that's okay, I guess... Yet to be sure I'm going to pick up a few spares, just in case.

Update. It looks like Ikea upped the prices of these spots, 13 euro now. Pfew. That ain't cheap...

Unfortunately, I need a few of 'm. Main reason: after refurbishing and finishing the room on the ground floor I decided to add Dolby Atmos ceiling speakers... and those required speaker cabling. Unfortunately, the only place to run those cables was through a very small alcove. Which already contained all manner of cabling...

But I managed to cut holes without damaging the existing cabling! Took some effort, but still...

Interestingly Ikea does not mention how much power these consume, so I measured that myself:

(Yeah, that's a new Korad KD3005D bench power supply. Seems to do its job well, though its current metering is a bit off, and the transformer is pretty noisy under load.)

24 V x 56.3 mA = 1.3 W / unit
10 x 1.3 = 13 W

The APV-35-24 has the same dimensions as the APV-35-12, so I would only have to design one enclosure. Again, it's a little too big, and again I could not find a better match. Oh well. We'll see how things work out.


Any electronic component dissipates some heat. According to the APV-35-24 datasheet its efficiency is 84%. Let's assume the driver has a loss of 20% so it will generate some heat:

20 W x 0.2 = 4 W loss

I have NO idea how much space is required to use natural aspiration to cool 4 W, but a single 4 W spot gets lukewarm.

Just to be sure I planned some ventilation ducts in the casing, but I'm pretty sure it is complete overkill, and I'm going to test a complete enclosure on maximal load before committing to all 3. Don't forget: these things sit on the ceiling, which is the warmest place in your house.

(And if the ventilation channels are too large I can always reduce them before the final version.)

Here's the latest design, with projected LED driver (green), LED controller (blue), inline fuse (red) and screw terminal block (white).

Safety... How about fire?

Whilst considering all safety aspects I just realized something... Yes. PLA is a plastic, and is a great insulator, but is it safe? What if a short circuit causes a spark, what if the LED driver goes up in flames? What will happen to the enclosure?

It turns out PLA, the most common and cheap filament, is inflammable and NOT self-extinguishing. In fact, it keeps burning ever so slowly, dripping flaming drops...

Bummer. I don't think I want something like that on my ceiling. Perhaps I shouldn't be too worried, as the box could also be mounted underneath a wooden ceiling, there is a fuse, and the driver circuitry is inside its own plastic casing, but still it feels wrong...

I'll continue for now, and should reach a stage where I have a complete printed enclosure in PLA I'm satisfied with. Then I'll have to decide to to either...

  • Accept the risk (if it's really a risk)
  • Choose another material (and thus buy another printer, sigh)
  • Find someone to print these for me, or...
  • Find a PLA (compatible) material which is fire retarding



No comments:

Post a Comment