Corbin's Treehouse - Corbin Dunn, Santa Cruz, CA
Plug Bug

Photography: India, Fruit delivery

India, Fruit delivery


This guy didn’t look too happy, but I took his picture as he was walking by on the road. India, February 2014.

LED Cyr Wheel: Longer video

Another slightly longer video from last weekend. I have about 3 or so hours on it, and it seems to be holding up well!

LED Cyr Wheel v2: Test Video

Some quick test footage from last week; I used the wheel for the first time last Friday for an event put on at the Exploratorium by Velocity Circus. It works AWESOME!

How to build a Cyr wheel or roue cyr: materials needed

A lot of people email me asking what materials to use for building cyr wheels (also known as a roue cyr, rouè cyr, or simple wheel).

I like “thick” wheels (also known as a “fat boy”), with 1 3/4″ to 1 7/8″ outer diameter. Your preference may vary. Length of materials may vary depending on your wheel circumference.


* Main tubing: 20′ of 6061 Aluminum Tubing, 1 1/2″ Outer Diameter (OD), 1/8″ sidewall thickness. I buy pieces from a local supplier, Gorilla Metals which used to be Metal Supermarkets. You can buy shorter pieces online; factor in the fact that you what a 5 piece wheel, and will probably have 6″ to 1′ of an under-bend on the end of each piece (again, varies by technique and how you roll or bend the wheel).

* Inserts (solid): 5′ of 6061 solid Aluminum Tubing, 1 1/4″ Outer Diameter (OD). Turned down on the lathe to a bout 1.230″, or, use sanding discs on an angle grinder after you bend it, and sand till they fit (not ideal, time consuming, but it works). Same suppliers as the Main Tubing. Inserts are anywhere from 10″ – 12″ long.

* Inserts (alternative, tubing): 5′ of hollow steel tubing, 1 1/4″ OD and 3/16″ sidewall thickness. This is heavy, but works for my LED cyr wheels. Inserts are shorter (but I haven’t stress tested them), usually 8″ long. Turned down on the lathe to 1.230″ (or so). Same supplier as above.

* PVC Covering: 25′ of PVC tubing, 1.5″ outer diameter and 1/8″ or 3/16″ thickness. Bought online,, search for Tubing. This or this. I’ve tried both..and one is superior and lasts longer, but I forget which one.

LED Cyr wheel: on/off switch.

I’m using a 6amp 120V on/off switch. I might be running up to 16 amps through it. We’ll see if this works..



The switch is fished through with wire and fishing line.

LED Cyr wheel: casting…more info

I learned some more things about casting the EpoxaCast 690. On one of the last pieces it was really opaque:


I called Smooth-On the next day and asked about it. They didn’t know, so I figured I needed some new stuff and ordered some. When I got home from work that day, I decided to microwave the Part A and this made it start to become clear again. So, apparently it is rather sensitive to temperature, and it being 50-55F in my garage wasn’t good for it. I warmed it up in front of my powder coating oven and the next casting was clear.

LED Cyr wheel notes: casting

Here’s some notes on casting…maying for myself when I go back to do this again and try to remember what I was doing.

First, to get the curve, I would cut the LEDs every 6 and re-solder them together:


All us techie programmer know a curve is an approximation of straight lines; and mine are 6-pixel straight lines.

Next, casting: I had two materials to cast with (see Casting experiments). Smooth-on EpoxaCast 690, and some Polyester casting Resin. The resin is fairly cheap, and they sell it at my local Tap Plastics store (I discovered that after I ordered it on Amazon). I experimented with both, and they seemed about equal. However, the Resin was harder after it dried, while the smooth-on was sort of more plyable. I’m not sure if that is good or bad (yet).

My first official wheel piece didn’t go so well. I used the EpoxaCast, and tried taping over the groove horizontally (in line with the wheel curve), but this was a mess. I kept getting a lot of bubbles and it was just not going well.


In the above picture, you can see I bought a scale, as it is a two part epoxy measured by weight.

Eventually, I realized it was better to spread the tape perpendicular to the wheel…but even then, I got lots of air bubbles under it. Worse, the next day a bunch “drained” out, probably from capillary action:


Hmm..not so great.

So, I decided to try the Resin for the second casting. It went on really well, as I put the tape perpendicular to each piece and seemed good. But the next day, the resin (which dries MUCH faster than the Smooth-on Epoxy) was sort of “blotchy” under the tape. Grr. (Note: later I fixed this by putting some epoxy on top of it).

So, I tried the smooth-on epoxy again, and with the horizontal technique it was working pretty well, despite some spots getting air bubbles and dripping out.

One thing I would do is hot-glue fill in the edge to keep the epoxy running out:


After I taped it, I discovered I could “cotterize” it with more hot-glue to keep it from dripping:


The glue is fairly easy to cut off with an exacto knife.

Next thing I learned: tape makes a difference. I was using Heavy Duty Scotch shipping tape to “cast” the LEDs in; it pulls off pretty easily and cleanly. I ran out, and used the weaker/cheaper ACE tape.


Bad idea. It reacted funny with the epoxy:


Luckily most of the white opaque mess peeled off, but some I had to scrape off. Plus, this tape left a sticky residue. It was not the right tape to use.

Other notes: the temperature should be warm..ideally 70F or higher. The stuff takes forever to cure when thin; it says 24 hours, but it seems much longer — at least 36, but probably 48 hours. And I think it takes a week before it is fully cured. Outside temps have been 50-60F (or up to 70F during the day).

LED Strip test and resin/epoxy

Each time I wire up a new piece I test it to make sure I didn’t mess something up. Here is a test of 4 pieces, rainbow pattern:

I did one section covered in the smooth on didn’t work so well. It was to liquidy and probably need more catalyst than what it says in order to cure a small thickness. The test piece sort of felt rubbery, and I forgot to add more. Rubbery might actually be good. The other resin I got was rock solid hard, so I’m trying it on a second piece (so far, it seems better, but it will need to dry overnight).

On the first test piece with smooth on, I would add the epoxy and then try to cover it with clear packing tape to mold it into a curve. This was a mess….and frequently captured air bubbles underneath. Using the other casting resin worked much better (no pictures yet) with the same process.

Worse, the first test piece sucked a bunch of stuff out via capillary action overnight:


Luckily it dries really slow (24+hours) so I could wipe off the overflow with acetone on a rag.

LED Cyr wheel: Wiring up the wheel

I have a 5 piece wheel, and I have a “top” side for reference, and the numbers go from one to five in a counter clockwise direction. This is piece 1 to 2, with the 1 insert on the right side. This right side is the location I will push the chips into. The insert can be used to push them in, and usually the insert bolts will only be removed on the right hand side.

Below: The wires coming out the hole: red/black are 5v. They will be split to power each side of the wheel’s LEDs. The white wire is the data line for one side of the wheel, and the yellow wire is the data line for the other side. They (for now) tie in to the same input on the Teensy, but in the future I could control one half separately from the other if I need to.


Close up of splitting the 5v and ground wire:


The white wire goes to one side of the wheel and the yellow to the other side. I’m only doing one side for now. To the left of the main hole you can see the button; one end of the button goes to a pin on the Teensy, and another to ground. I can then tell when it is pushed and do stuff.

Here’s a close up of the input into the wheel, which has to be far enough down to clear the insert:


The 5v+ and ground can be put anywhere into the strip, but the white data line needs to be on the first pixel (based on the direction of the arrows in the strips). Note that I squeezed in a 1N4148 diode from the DIN (data in) to the +5v red wire (with the negative end of the diode on the +5v); I’m hoping this will help keep the LEDs from dying, because if they do I’m in trouble. The white wire above runs to the DIN of the first pixel, along with another diode:


The other end of the wheel has the DOUT (data out or DO) line that has to feed to the next wheel segment: it runs from back to a hole past the insert:


Again, the yellow is for the other side.

The LEDs themselves half a sticky backing that helps them sit in the groove. Every 6 pixels I cut the DO and 5v line (but not the ground) and tweaked it to match the curve. Then soldered them back together:


Subsequent sections are wired a little differently than the first one. Section 2-3 is below:


I have a 5v (red) and ground (black) 16 gauge wire going through the entire wheel (minus the last section). I tap off 5v to power each section with 18 gauge wire (note to self: 20 gauge wire would have probably been fine, and use flexible wire for the next one!).

Coming into the section will be the yellow and white data line (used 20 gauge wire; 22 gauge would have been fine since it is just a signal wire — but you don’t want it too thin so it breaks when moving inserts around). In the picture below I’m using the black/red 18 gauge wires to “fish” the yellow and white wires through the hole.


Fished through the hole in section 2-3, and wires getting ready for a connector:


LED Cyr Wheel: v2: Chips etc

Here’s some of my progress and notes on making my new cyr wheel. This is partially so I can look back and remember what I did if I have to take it apart for some reason.

The chips that will go in the wheel: Teensy 3.1 is the main processor (black). Pololu voltage regulator (converts the LiPo 7.4v to 5v for the LEDs and Teensy). Accelerometer/gyro combo (middle right), and SD card reader from the guy who sells the Teensy.


I made a cradle out of acrylic from TAP Plastics. The circles are turned on the lathe and fit perfectly in the wheel. This is the voltage regulator mounted; the wires on the right are flexible (aka: high wire count) 16 gauge wire, which will run throughout the wheel center to deliver 5v. The right side 2 wires will connect to the 7.4v battery, in a different wheel piece. The 2 black/red wires on the left go to the main “5v” feed into the wheel, where each section will tap into (to avoid voltage loss, I’m using 16 gauge…plus, according to my calculations it might need a lot of current to power the wheel). 18 gauge wires are tapped off for the first section LED strips..


Looking at the above picture, the bottom wire is a tap on the source voltage so I can read it with the Teensy. One resistor is already covered in shrink wrap, and the second is the second half of the voltage divider (I forget which side is which; I’d have to look it up). In code, I have:

#define RESISTOR_Z1_VALUE 10000.0 // 10k resistor

#define RESISTOR_Z2_VALUE 4680.0 // MESURED value of 4.7k resistor

static float readBatteryVoltage() {

float readValue = analogRead(g_batteryVoltagePin); // returns 0 – 1023; 0 is 0v, and 1023 is max ref voltage

  float vRef = 3.30; // 3.3v ref

float refValue = analogRead(g_batteryRefPin);

  float voltageReadValue = readValue / refValue * vRef; // 16 bit resolution,

  // The read value is with regards to the input voltage

// See voltage dividor for reference

float voltage = voltageReadValue * (RESISTOR_Z1_VALUE + RESISTOR_Z2_VALUE)/RESISTOR_Z2_VALUE;

  return voltage;


The white wire feeds into my g_batteryVoltagePin on the teensy.
The above code snippet is how to safely read an arbitrary voltage with a voltage divider. Don’t ever feed a pin more than 5v or you will kill your hardware.
Note that I used two pins..I think I could use some internal ref voltage instead, but I need to look it up more. One tip: to get more stable readings in a Teensy set higher resolution and longer averaging:

  analogReadAveraging(16); // longer averaging of reads; drastically stabilizes my battery voltage read compared to the default of 4

analogReadRes(16); // 16 bit analog read resolution

Cradle full

The next picture below shows the cradle with everything on it. The gray wire to the right goes to the SD card reader. The teensy is “sort of” in the middle (slightly biased to the left). On the far left is the accelerometer. Between that and the Teensy is a little board with two 4.7k pull-up resistors for I2C communication with the accelerometer. The teensy needs the resistors. Reference:


I originally wanted the plastic cradle to be glued to the inert to allow it to easily be pulled in and out. But I wanted the SD card reader to be easily accessible without pulling the whole thing out to change the programmed pattern. So the gray wire goes to the SD reader which floats inside the insert. I wanted it fully on the right side, but I thought it was too much and would interfere with the main plug. So instead, you take the insert out on the left side to change the SD card, and on the right side to plug in the main power. This will make more sense when I take a picture.

I used some hot glue and screws to hold things in place, and wrapped it a bit to get it to not short out. I will use heat shrink tubing when I know I’m done messing with it.


I should have bought the really flexible “noodle” wire for other parts; it would make sliding the thing in a lot easier. Right now the 18 gauge wire to the LEDs is not very flexible.

(c) 2008-2012 Corbin Dunn

Corbin's Treehouse is powered by WordPress. Made on a Mac.

Subscribe to RSS feeds for entries and comments.

14 queries. 0.704 seconds.