Wednesday, October 30, 2013

Daft Punk Helmets: Lots of Lights

Previously, I covered the construction of both of the Daft Punk helmets in a series of posts. Starting with cardstock, they were assembled with superglue, reinforced with fiberglass, molded with Bondo, painted with metallic paint, and visors were made from heated plastic. They were nice and shiny and ready to be worn.

While the black-on-metal look matches the current era of Daft Punk helmets, they haven't always been so simple. Over the years they have gone through a series of visual changes, mostly involving the embedded electronics. You can find examples of when both helmets had an unreasonable number of LEDs hidden behind the visors doing all kinds of flashy things. The silver helmet can have a red LED panel stretching across the inside of the visor with other colored lights at the ends, while the gold helmet can have rainbow blocks of color running down the sides of the visor with a large yellow LED display panel covering the center. Mimicking these elaborate setups would take far too much of my own time and money, so I decided to boil down the lights to what I thought would represent the 'core' of the display.

For the silver helmet I decided to make a 40x8 red LED panel that would hide behind the visor and display words and patterns. For the gold helmet I went with RGB strips lined up down the sides of the visor. Instead of trying to diffuse the colored lights to make a solid block of color, I went with gluing the lights right to the back of the visor to create a slightly harsher and more modern look.

I'll start with the work that went into the 40x8 red LED display, since that one was much more difficult to assemble. The first step was to decide on how to control all of the LEDs. I went with a few MAX7219 controllers, each capable of controlling a single 8x8 LED panel. To prototype the controller, I set up a spare Arduino and pre-made LED panel:

Thanks to numerous examples found online of how to interface with the controller, I set off assembling the red LED panels. Since the display would sit right in front of my eyes inside the helmet, I needed the panel to have enough empty space between LEDs in order to allow me to still see out. To do this I set up a grid of holes in some cardboard and used it as a frame to solder the LEDs together. I made 5 8x8 panels where each column had common cathodes and every row had common anodes.

After quite a few hours of soldering, I had all of the LEDs soldered into panels. I wanted the LED drivers to sit on their own small protoboards near the panels.

With every 8x8 panel connected to a driver, I was able to hot glue all of the panels together side-by-side to make the single large display. I put little wooden sticks at the joints of each panel to allow the panel to be glued into the helmet without having to secure the LEDs themselves to the helmet.

Since the drivers could all be daisy-chained together, there were only 5 wires needed to fully control the panel. Two for power, one as a clock, one as a latch, and one as a data line. These would be hooked up to an Arduino controller, but more on that in a later post.

The lights for the gold helmet were someone easier once I decided to buy an addressable RGB LED strip from ebay. The one I picked was 144 LEDs crammed into a single meter with a built-in driver on each LED. As long as every LED was daisy-chained, only one data line was needed to set every LED to a unique color. I cut the strip into segments that were 6 LEDs long and re-soldered them back into a chain with wires to space them out.

This way, I could orient the strips to run parallel by bending each wire segment by 180 degrees, yet still only use a single data wire to control them all.
Since the LED drivers were all on-board, I was able to test the LEDs at many stages of the assembly.

The first thing I really realized was how bright the strip could get. With every RGB LED running at full brightness, the helmet would pull over 5A of current just to light it up. I was looking for a colorfully lit helmet, not a head-mounted search beacon. I decided the best way to solve this was in software, so I just kept the brightness in mind while testing the hardware. I marked off on the visor where the lights needed to be glued and went to work.

With all of the strips attached, I hooked up the power and data lines to a testing Arduino and lit it up.

At this point, both helmets had lights installed and ready to be driven. The end of the project was in sight, with only the control electronics, power source, and coding to sort out. Once again, more on that later.

Friday, October 25, 2013

Daft Punk Helmets: Vacuumed Visors

Previously, I covered the process of shaping the helmets and getting them ready for painting. In doing that I removed the opaque visors, since they would be replaced with clear copies. The goal was to mold clear plexiglass on to each visor so that I could control the shape of each precisely. Once I had the clear replacements, I could shade them black. After estimating how much plexiglass would be needed for each visor, I collected my materials.

The basic procedure is to heat up the plexiglass until it is malleable, then while it is warm, let it 'slump' on to the opaque visors. As the plexiglass cools back down, it hardens into the desired shape. The felt is to keep the plexiglass from sticking to anything. I had seen many people on the internet using vacuum formers to aide in the slumping process, but I had a hunch that it wasn't necessary. I was wrong, but more on that later. First I did the simple visor.

I heated the 0.125" thick plexiglass in my oven at 285F for about 10 minutes. I could see that it was ready for slumping when it started to sag in the oven. I carefully took the first sheet out and placed it on the first visor. Using some extra felt to push the plexiglass down in the corners, I made sure it would solidify with the right shape. Once it cooled, I had an plexiglass replica of the opaque visor (along with some extra material around the edges)

Using felt to avoid scratches

I then carefully cut off the excess plexiglass from the edges, then sanded where I had cut for a smooth edge.

Since this visor was easy enough to make, I went ahead and tried making the more curved visor with the same method. I stuck a sheet of plexiglass in the oven, and prepared to mold it by hand.

It failed. The plexiglass was too thick and would not bend in two directions without folding around the edges. I decided the plexiglass itself was the issue, so I bought a few sheets of 0.06" and 0.02" thick PETG. This plastic becomes malleable at a lower temperature, and I figured the thinness would also help. After heating up the oven, I followed the same procedure as before.

My process still wasn't working, but I could see I was on the right track. Kind of. I decided it was time to use a vacuum former for help. Since I was running out of time on the whole helmet project, I decided to first look at local plastic companies that do vacuum forming. I contacted one about my project, and they said both that it would cost nearly $2000 to do it, and that they simply would not attempt it. This is not specifying a type of plastic or tolerance level, just the job of vacuum forming anything around our model visor. With few options left, I decided to build a vacuum former as fast as possible and for as cheap as possible.

If you search "diy vacuum former" you can find quite a few good examples of cheap vacuum formers that can be made with scrap wood and a shop vac. After glancing through a few instructables on the subject, I went to work. I was lucky enough to be able to borrow a nice 6hp shop vac through some friends, so that would keep the cost down. I was able to get the rest of the materials at Home Depot for around $25.

The basic idea behind a vacuum former is to have a flat surface that creates a uniform vacuum to pull a piece of heated plastic down. The model you want replicated sits on the vacuum surface and holds the plastic up away from the bottom. The plastic is heated so that it will bend and stretch around your model while being sucked down around it. Once the plastic cools, it will retain the shape of the model.

The vacuum former has two parts. The first is the frame which holds the plastic while it is being heated. The second is the vacuum box which supports the model and creates an even vacuum surface. I built the frame out of some 1x2 wood pieces and a couple nuts and bolts to tighten the two sides around the plastic. I built the vacuum box out of MDF, 1x4 pieces, and some rubber weather strips.

The edges of the box were sealed with duct tape. I drilled around 200 1/8" diameter holes in the top of the box on a rough grid I traced out and taped the shop vac hose to a hole I cut in the side of the box. The rubber strips had an adhesive backing, so I just stuck them right where the frame would sit.

With the vacuum former ready to go, I loaded a new sheet of plexiglass into the frame and put the frame in the oven. This time I got a little fancier with my baking and did 8 minutes at 300F and finished with about 3 minutes of broiling. The plastic was sagging by about 3" when I took it out and had just started to create tiny bubbles in a few places. With the shop vac running, I quickly placed the frame over the vacuum box and pressed down to ensure a nice seal between box and frame.

Yet another failure. At this point I was seriously worried for the helmets. Without a visor, the gold helmet was pretty useless. I took some time to think through how this attempt had failed, and luckily saw a reason to try again. A big issue for this attempt was that the plexiglass had fallen free of the frame in one corner during the heating process. I found this was because the plexiglass has slipped to one side while tightening the frame, so it was not being clamped down on that corner at all. This could be fixed by being more careful during the set-up. I also added some cardboard in-between the frame and the plexiglass in order to get a better seal. I wondered if I had drilled enough holes, so I computed the collective cross-section of every hole together and compared it to the cross-section of the shop vac hose. The shop vac hose was about 5 square inches, while my holes were only adding up to around 2. I went through and widened most of the holes in the top of the vacuum box to bring that figure much closer to the hose area.

With these modifications done, I set up for another attempt. Luckily the plexiglass sheets were only costing me about $7 a piece and I could get them cut to the right size at my local hardware store.

It finally worked. Well, mostly. Under the pressure of the vacuum, the model visor actually bent out of shape, causing the bottom of the visor to become wider than I wanted. This was surprising because the model was cured Bondo reinforced with thick fiberglass. I decided this visor would work well enough with some bending, and I was too sick of molding plastic to try again. I cut out the visor and prepared it for tinting.

Both if the visors were tinted black using Nite-Shades. This is a spray intended for vehicle lights to make them look black when there is no light inside shining out. I tested the spray on the many failed visor attempts to make sure I wouldn't accidentally ruin the visors I had finally made.

Once both visors had been tinted, I began attaching them to the helmets.

The second visor did not fair so well. After a day or two the plexiglass became brittle and cracked down the side. I carefully glued it back together with special plexiglass adhesive, but once I tried bending it again, another crack formed. The visor was doomed. After 6 failed attempts at making a visor for this helmet, I was pretty let down. Luckily, after 6 failed attempts I was getting better at making them. I ordered a sheet of PETG to make a new one, with the intention of vacuum forming it just like the plexiglass.

I'll stop here now that I've covered my process for making and tinting both visors, even though the second one is not 100% done. I'll try to update with a picture or two when it has been fixed.

UPDATE: It seems like 0.03" PETG is the way to go. I've formed a new visor that is reasonably strong and fairly flexible. The material was much easier to heat up (~300F for ~3 mins) and actually allowed me to redo the vacuum forming process twice. On the first attempt the PETG didn't get pulled down as well as the plexiglass did, probably due to human error in placing the frame on to the vacuum box. The sheet cooled into an almost-visor shape, but I put it back in the oven and watched the plastic return almost to it's original shape. This was a life-saver.

I reinforced the fiberglass/Bondo visor mold this time to prevent the bottom from bending out, but the visor was still slightly misshapen. Luckily the formed PETG is flexible and allows me to bend it back into place. Next step is to attach lights to the visor, then glue the visor on to the helmet.

Thursday, October 17, 2013

Daft Punk Helmets: Icing a Toxic Cake

At the end of my last post, both Daft Punk helmets had been formed with papercraft and fully reinforced with fiberglass. Once the fiberglass had cured, the helmets were rock hard, but fairly lumpy on the outside. The fiberglass cloth was applied to the inside, leaving the hard edges and seams of the original papercraft on the outside. To smooth out these bumps and produce a nice surface for painting, I used Bondo. Bondo is a putty-like substance that starts as a two-part mixture, goes through exothermic curing once mixed, and ends up as a hard clay-like substance. The final state is relatively easy to sand, allowing for fine tuning of the surface shape. The idea here is to iteratively shape each helmet by applying Bondo, letting it cure, then sanding it down. With each iteration of this, the helmets get closer to their final shape.

It's a lot like icing a toxic cake.

With some 60 grit sandpaper, I was able to shave off a lot of the goopy unevenness of the first layer of Bondo. My goal was to sand it down to the correct shape, and if I hit fiberglass, add some more Bondo everywhere to allow for more even sanding. I enlisted some help for sanding, since doing two helmets lends itself easily to two people working.

Helper with tiny power sander was a life saver.

Not quite what I had in mind..

The biggest issue we ran into was air pockets. I was not at all careful with my first layer of Bondo, so as we sanded it down we found many dents and pockets caused by uneven application of Bondo. At first we tried filling these holes with more Bondo, applied with a small stick for precision. This took a long time and was not nearly as precise as I wanted. In some areas we also needed to add an even layer to get away from the fiberglass beneath, but would end up with even more pockets and dents using Bondo.

Fixing past mistakes and adding future ones.

You'll notice in the pictures that the visors have been cut out already. In order to create the clear plastic visors later on, all we need is a solid block that mimics the desired shape. Since the mold doesn't need to be perfectly smooth, I went ahead and cut them out using a Dremel rotary tool. The key was using the right cutting bit and being very careful. I ended up breaking two different cutting bits in during the process, so I won't claim I know which one is best to use.

To get the rest of the helmets smooth faster, I tried a two-part solution. I bought some liquid nails that claimed it was easy to sand once cured and used a small nozzle applicator to fill in dents and holes. Unfortunately, each application was taking days to fully cure, and I didn't have that much time to waste. The next part of the solution was to use filler primer spray paint. This is a type of primer that has a decent amount of volume to it when applied, so it helps fill in small dents and holes. Once the helmets had all of the major bumps smoothed out, I switched to primer.

The difference in surface area was surprising.

Dat back.

To eliminate dents, we followed a simple pattern: apply two coats of primer, let dry, sand with 220 grit until Bondo shows, wash, let dry, repeat. After three or four iterations of this, most of the dents had disappeared. During this process I also cut out the mouth slit on the helmet that needed it, since the hole left by papercraft had been accidentally filled in during the Bondo process. Again I used my Dremel and switched between a few bits while cutting.

This whole shaping ordeal took a long time and a lot of effort. Most of the helmets are large smooth surfaces, so hand sanding wasn't too hard. For the smaller details around the ears, I used a Dremel to carefully sand them down while maintaining the angles I wanted. While the final shape was not 100% perfect, it was good enough for a costume.

At this point, both helmets were ready for real paint. In one of my next posts I'll cover what steps we go through to get the nice reflective silver and gold surfaces needed. Spoiler: there is lot's of sanding involved.

Tuesday, October 1, 2013

Daft Punk Helmets: Papercraft to Fiberglass

Progress has been made on the helmets. As mentioned in the introductory post, the first steps were to create full-size paper versions of each helmet and then reinforce them with fiberglass. I've been following some guidelines put out by a few sites like this one.

The idea behind papercraft is that you can create a folded-paper version of a 3D model that has been designed on a computer. There is software that takes imported 3D models and splits up the geometric faces into pieces that can be cut out from paper and glued back together to re-form the 3D shape. This software outputs a series of 8.5"x11" images that are printed to cardstock. In order to connect the various 2D pieces together to form the 3D shape, the software adds on flaps and fold lines to instruct what pieces need to be glued where.

After finding nice-looking 3D models of the helmets on the internet, I set out to print, cut, and glue. I didn't have access to a printer that could handle cardstock, so instead I just printed to normal copy paper and glued each sheet to a piece of cardstock.

Hints of a 3D structure.

Rising from the fold.

The whole papercraft ordeal seemed daunting at first, but then I realized it was 1am, had just spend 6 hours gluing my fingers together, and had no intention of stopping. When I got busy cutting and assembling (with the help of a friend) I noticed it wasn't as bad as I had anticipated; the mindless work was actually a little addicting.

Went through a few X-acto blades.

Hoping I can fix slight lumps later..

Each helmet ended up needing about 6 hours of work. I was very happy with the final product, even if my occasional sloppiness in cutting caused some unwanted lumps on the helmets. Both helmets were lightweight, but still able to support their own shape.

The next step was to reinforce the helmets with fiberglass. Fiberglass is a strong, lightweight material that is made of two components: resin (glue-like stuff) and fiberglass cloth (a cloth that is made of bits of glass). If handling a cloth made of glass shards sounds like a bad idea to you, then you are a reasonable person. Don't let the stuff touch your skin. I've made this mistake when installing insulating wrap on motorcycle exhaust pipes and had to cover my hands with duct tape in an attempt to pull the tiny glass fibers out of my skin. Also, the resin comes in two parts that are mixed together to initiate curing (hardening/drying). This is an exothermic reaction (gives off heat) and it can really heat things up. A thin coating of the stuff probably won't heat up noticeably, but a cup of the stuff can start to smoke and will burn your hand if you hold it. In all, don't try to make fiberglass things unless you are familiar with the hazards. Do it outside, wear protective gear, etc.

The first step in reinforcing the helmets is to coat the inside of them with a thin layer of resin. I mixed the two parts together in a mixing cup and used a cheap paintbrush to coat each helmet.

Outside looks wet due to the epoxy from the inside seeping through.

Once the helmets were thoroughly coated I let them dry in an outdoor closet. After a day they were much stiffer than before, but still not robust. This first coat of resin was primarily to ensure the paper helmet doesn't fall apart or warp during the application of fiberglass cloth. To efficiently apply the fiberglass cloth to the inside of the helmets, I cut a large sheet into smaller strips (about 3"x1") and coated the inside of the helmets again with a thin layer of resin for stickiness. I would place 4 or 5 overlapping strips inside the helmet, then dab more resin on top until they looked soaked. I found that my resin would start to harden in the mixing cup after 10 or so minutes, so I made small batches in order to not waste materials.

After being left alone for another day or two to cure, the fiberglass was hard as rock. My rough estimate is that each could withstand having about 40 pounds placed on top before buckling. Don't blame me if you try this and they collapse with less than that. I'm certainly not going to put my beloved helmets to the test.

Ready for the next step!

The next step we be to coat the outside in Bondo (or similar) and sand it down to fine-tune the shape.