Week 7: Molding and casting

This week's assignment was to "design a 3D mold, machine it, and cast parts from it".

Fifty Shades of White: Color-programmable high-power LED lamp

During Week 6 I started thinking about projects that could be made on more than one week. I decided to design and build an LED lamp. Specs:

Cree LEDs are really common in LED lighting applications. I decided to use the XH-G series:

The XLamp XH-G LEDs are the first mid-power, ceramic LEDs that offer no-compromise performance and reliability. Unlike common mid-power plastic packages of today, the ceramic-based XH LEDs are designed to deliver the long L70 lifetimes of Cree’s other high-power LEDs, such as XP or XT LEDs. XH LEDs are optimized for fluorescent replacement lighting applications, such as troffers and panel lights, where high efficacy, long lifetime and smooth appearance are critical. The XH LEDs enable lighting manufacturers to create a new generation of more energy-efficient, longer-life LED lighting solutions without sacrificing cost or performance. The XH-G LED delivers leading efficacy levels of up to 170 lumens-per-watt at 65 mA, 5000 K, 80 CRI. Source: http://www.cree.com/led-components-and-modules/products/xlamp/discrete-directional/xlamp-xhg

On Digikey, I picked:

Assuming a minimum of 25lm/LED I need 20 LEDs (~10$). However, when CADing the lamp, I found it easier to use an even number of rows and columns. I used 16 of them. I'll have to push more current in them. I should be able to get from 400 to 1400lm total. That's really bright!

I'll wire them as four strings of 4. 4*3.4V = 13.6V. It's easy to find 18V power supplies. 18V - 13.6V = 4.4V to drop. At 250mA, I need a 17.6 ohm 1.1W resistor per string. I'll use PWM to change the intensity and the color temperature. This is what I designed last week:

However, the external diameter of the orange part was 73mm (2.87in): too big for the 3x3.5x1.5" blocks of wax. After last Wednesday's class I quickly modifed my CAD to get. a diameter of 64.2mm (2.53in). This is the result:

As you can see I didn't bother placing the LEDs in the assembly and I didn't redesign the diffuser plate. All my attention was on the casted part. The threaded tube that you can see on the back face is a standard 5.5mm barrel jack that I'll use to plug the lamp.

I designed a 2-part mold in Solidworks, using the Insert Part in Part feature. I manually added the walls and the vents. Special thank you to Matt Carney for fool-proofing my design and teaching me Solidworks tricks. Here are my two parts:

I started by making the Inner wax mold. Here's what I noted in my logbook:

Completed wax mold:

I decided to try the OOMOO right away:

Ready to remove the bubbles:

I added a plate of acrylic on the top to get a flat surface:

Fully cured OOMOO (~2h):

Me carefully extracting the OOMOO from the wax:

And here's the final result:

I followed the same procedure to machine and cast the Outer mold. My 3 mounting holes were too small; they didn't get machined. That's no big deal as I can easily post-machine them.

Here's the Outer wax mold:

Pulling the OOMOO mold was trickier than before. Although I waited the same amount of time (2h) as for the first time, the OOMOO was not as cured. I made a way bigger batch with Hasier: maybe we didn't mix it as well?

I left the un-molded OOMOO part under the hood for the night to let it cure. Here's the result:

I went to Reynolds to buy castable urethane plastic. My initial idea was to get either Task 4 (sold for thin walls) or Onyx Black (this one has an awesome flashy black finish). They told me that the Onyx was really weak for small features. They recommended the Task 3 instead of the 4 as it has a lower viscosity. I bought a Trial pack of Task 3 and a small bottle of So-Strong orange pigments. According to Smooth-On I need to use from 0.01 to 3% of the urethane volume in So-Strong. Putting too much can prevent cure. I want it dark and bright, but I also want to be safe: let's do 2.0%.

I started by assembling my 2 part mold, holding the parts with electric tape. I applied some mold release on the two parts. It’s clear that my external walls are too thin. I need to expect some deformation. Machining in a bigger block of wax would have been better.

Hasier and I made the Task 3 mix. I added 4% of the B part’s weight in Orange So-Strong. It will give us a little bit less than 2% of the total weight. It’s really really dark! I could have used lot less. You only know when you start stirring.

The Task 3 is not too viscous, I felt confident that I could pour it in my (way too small) holes. Turns out that I had to pour it drop by drop… the surface tension is too big on such a small tube. After maybe 10 minutes of slowly pouring it from 2 different holes (keeping the other 2 as vents) the mix started to get hot. I started to fill from the 3rd hole, hoping that I could fill the mold.

Hasier tried to fill his mold, without any success; it simply didn’t flow. We exceeded the pot life: it was like hot wax now. In a desperate attempt, I removed all the black tape, lifted the top mold and dropped a big blob of Task 3 straight in my mold. I then pressed the top part (Inner mold) back in place. A bunch of red urethane dripped on the sides. It looks like a scene from Dexter…

We left our parts cure in the shop for 90 minutes.

The results are not as good as I initially expected… but way better than what I expected after all the pouring problems! It was really easy to demold. Here’s the part after I removed the outer mold:

The molds seem intact. I’ll be able to use them again. Two-part silicone mold with casted urethane part:

After some manual cleaning with a sharp blade:

As you can see, the middle tube didn’t get casted. The small post that broke while machining the inner mold was useful after all... The plastic is also softer than I expected. Maybe it isn’t fully cured yet? Time will tell.


Final Project Update

My Week 1 page describes the first idea I had for my final project, the Dishpenser. Six weeks later I'm still thinking about this and I took the decision to go ahead and start designing the system. To make sure I can complete it before the end of the semester, I decided to start with a unit that only cleans glasses.

I sketched some rough ideas. I'll post them soon, with the initial CAD files.


Week 1 Week 2 Week 3 Week 4 Week 5 Week 6 Week 7 Week 8 Week 9 Week 10 Week 11 Week 12 Week 13 Week 14 Week 15

Jean-Francois (Jeff) Duval - jfduval(@)media(.)mit(.)edu - 2013
Top | MIT Media Lab | How To Make (almost) Anything 2013 | Index
"Hidden" keywords: Jean-François Duval, JFDuval, Jeff Duval, PCB, Dishpenser, MIT Media Lab, Biomechatronics, Electronics, Robotics, DIY, Hack, Hacker