Large 8 Door (Leaf) Iris for a 2 Foot Diameter Round Window

I've always wanted to do something with our 2 foot diameter round window in an upstairs bedroom. I'm retired now and found the time to act. If you're willing to follow along, I'll take you on my journey.

During the journey, you will encounter jpg files, gif files, animated gif files, and pdf files. All these will be thumbnails of bigger views. Running the pointer over a thumbnail image will produce a bigger view with a tag with text. Clicking on this view will open a still bigger view in a separate window (and to return press the back button.)


We had our house built in 1977, and since I opted to have a stairwell to the basement at the back of our attached garage, the garage had to be pushed forward by 4 feet, thereby requiring the east facing window on an upstairs bedroom to be changed from rectangular to round. The round window was 2 foot diameter single pane with six lead spokes in a square box frame. It opened when the top half was rotated. A half round screen was on the inside.

I always thought it would be interesting to have a mechanical iris over the round window, like those found on older (before digital) cameras. However, I thought the aperture (2 foot) was too big. A couple of years ago I came across this Steampunk Forum link which showed how eight leaves could slide and not overlap, so the challenge to construct the iris was thrown down.

I can program in raw postscript. It's similar to javascript. I enjoy postscript because it uses reverse polish notation like the HP calculators without any equal (=) button. I used postscript programming for line intersection and bezier curve calculations.

I have software to convert my postscript ps files to gif files and software to convert mutiple gif files to animated gif files. This proved invaluable to check designs and interference of the moving parts.

I also have software to convert postscript ps files to pdf files. This proved useful to draw full scale components and check for clearance.

Finally, I can convert a postscript ps file to an encapsulated postscript eps file. The eps file can then be imported as tool paths into a 2.5D CNC router.

Replacing the Round Window

The original round window was taken out and recycled at ReStore (Habitat for Humanity). I bought a new round window from

The new round window is thermal dual paned and swings out from a hinge at the top to open. It has an inside screen. It is also completely round so I had to build a box frame to install it. I covered the frame with drywall leaving just the round window. Here are some pictures of the new window.


Designing the Mechanical Iris

The mechanical iris consists of four layers. I knew that I wanted a base plate octogon shape with a 2 foot circular hole to guide the leaves as they moved. In the next layer, the leaves are guided by runners in the base plate and pins on the other side which are in a curved cam groove. The next layer was a cam ring with curved grooves so when the cam ring rotated the grooves would push or pull the pins thereby moving the leaves along their runners. The final layer was a cover plate to hold everything together.


After some animated gif experimentation I found that if one makes the line of the runners perpendicular to a radial line that adjacent leaves would slide continually in contact with each other and could open in either direction (theoretically!). I then changed the line of the runners to 91°. With this modification the adjacent leaves would separate as the leaves opened and would close to zero when the leaves closed. This would also eliminate friction between adjacent leaves. I attached ultra-high-molecular-weight polyethylene, UHMW (or UHMW PE), splines to one leaf edge which would fit in a groove in the adjacent leaf. These UHMW splines (from Woodcraft) have very low friction and they also keep the leaves in one plane as well as block out light.

The next challenge was to find the mathematical shape of the cam groove. Working backward I knew where the pin would have to be at any leaf position. Doing a lot of angle calculation I settled on a single bezier curve. A bezier curve is a curve which can be descibed by 4 control points. It is simple to get the slope at any position along the curve and I found a simple way to lengthen or shorten bezier curves. I plotted MANY iterations of the curve to give me a screw angle as constant as possible. The final bezier curve results in a screw angle (measured from line perpendicular to the line of runner) of 21.5° ± 1.44°. The result is that the cam ring only has to turn 72.26° from fully closed to fully open. Professor Frank Fronczak helped me clarify my thinking and introduced me to needle roller bearings for cam following. I purchased 0.5" diameter, for a 0.3125" shaft, and 0.3125" tall, closed end, cage guided needle roller bearing (BCE55) from:

This enclosed cup bearing size will dictate certain dimensions to follow.

Cutting the Mechanical Iris Parts

After finalizing the design I figured I would design templates to rout the cam dadoes and position of the runner dadoes. I'm very fortunate to have a mechanical engineer friend, Rich Frisbie, from college days. While visiting one weekend I showed him my plans and idea to make router templates. Rich suggested I had an "indexing" problem. That is, I had to accurately position and index the templates 8 times for each of the eight runner dadoes and each of the eight cam curve dadoes. He suggested that I see if I could find someone with a large CNC router to rout out the parts.

The next Monday I went to order the 3 sheets (each 5' x 5') baltic birch 5/8" (really 15mm) plywood from a woodworking supply store. The laminations are 1.5mm. While there I explained my project and asked if they knew a place that had a large CNC machine to rout my parts. The owner suggested I inquire at Drift Studio. I drove over and explained my project, and they were quite intrigued. They would squeeze in my project when they had time. Chris Hindle from Drift Studio was exceedingly helpful.

Drift Studio has a ShopBot and the software they use for CNC control is PartWorks. The software can import eps files for tool paths and drill locations. One has to specify to rout on the: right side; left side; or, centered on; the toolpath vector. A number of points to note:

Next are toolpaths for the three 5x5 panels for parts used (scaled pdf files) and some photos of the CNC routing:



Finishing the Parts

I still had work to do on the parts. On the back plate I countersunk for 3/8" diameter 3" long flat head socket screws. Since the holes were already CNC drilled I used a countersink (C24) with 3/8" drill rod to center the countersink. The drill rod was 4" cut off a 12" long 3/8" drill bit. The idea of using a drill rod to center the countersink was reinforced by Rich Frisbie. Our son Joe helped hold the back plate so I could use a drill press.


The cam ring had the two outer edges eased with a 1/16" radius round over router bit. This was done because the edges go over UHMW plastic (on and off) and I didn't want an edge to catch on the plastic. My wife Elaine steadied the cam ring as I routed the edges. Also to drill the holes and counterbores (using a Forstner bit) for the two handles fastened to the cam ring our son Joe helped hold the cam ring so the drill press could be inside the ring while drilling.


The cover plate had 60 mil (1/16") Slick Strips 12" long 1/2" wide attached at the octogon vertices. The half diamonds had 30 mil (1/32") Slick Strips 9" long 1/2" wide attached to their long edge to be in contact with the cam ring edge. The eight half diamonds centered the cam ring. The "full" diamonds had 1/8" UHMW PE attached to their surface by SpeedTape which I obtained (both UHMW PE and SpeedTape) at Woodcraft. I oversized the plastic and attached it to the wood diamond, then used a router flush bit to trim the plastic to the diamond shape. I used a 3/8" Forstner bit to drill a matching hole in the plastic. See the following pictures:


The 8 leaves required a considerable amount of work. First a 3/8" deep 1/8" wide slot was routed in each of the long sides. I should mention here that all this work was new to me. I usually tried a process on some scrap wood first to see how to do it. Then I might start with #1 leaf, skip two to do #4, skip two to do #7, skip two to do #2, and so forth until I did all 8 leaves. The next process I would start at #2, skip two, and you get the idea. I did this so one leaf did not get all my lack of experience and another benefit from me doing everything for the eighth time! Call it randomizing my experience.

After routing the slots for the splines I then inserted the metal inserts. To drive the inserts I cut a 5/16" bolt at both ends. I cut off the head to leave a shaft to chuck into a drill press and I shortened the thread portion to hold a winged knob, chrome USS washer and one inverted insert (slot down such that a large flat blade screwdriver could tighten/loosen the assembly). I unplugged the drill press and removed the v-belt. I chucked the tool with insert (and a dab of wood glue on the insert) into the drill press. From the backside of the drill press I lowered the quill with my left hand and screwed the insert into the leaf with my right hand until the tool bottomed out. Holding the quill steady, I unchucked the tool. I took the tool and leaf and with the large blade screwdriver holding the insert from the underside of the leaf, I unscrewed the tool from the insert. This whole procedure kept the insert as perpendicular to the leaf as possible. The wood glue acted like a lubricant when wet but bound the insert into the wood when dry. An excellent machinist and friend, Richard Jordan, made the 5/16" diameter axles for me. They were 3/4" in length with 1/2" threaded for the insert and 1/4" unthreaded for the needle cap bearing. The inserts, axles, and needle cap bearings worked great, rolling, not sliding in the cam grooves. Here are some pictures:



I then cut 3/4" wide, 1/8" thick UHMW PE splines and used 3 1/2" x 18 Ga. brass escutcheon pins to fasten the splines to the slot in the shorter of the two slotted sides. My first design had the leaf outline symmetric but that resulted in a 48" side-to-side octogon. It wouldn't fit in over the window, hitting the sloped ceiling. With the leaf outine asymmetric the octogon is 42" side-to-side. Then to fasten the splines I drilled a 1/32" pilot hole, hammered the pin leaving the pin 1/16" proud. I took a 5/32" copper tube, sharpened it, chucked it in a hand drill and rotated over the proud pin head. This cut the wood fibers around the pin. I then used a nail set to hammer the pin head flush with the surface without splitting any wood fibers. There were 3 pins per leaf, 4" apart, pinned from the side with the runner. The runners were attached to each leaf with four brass #6 x 3/4" wood screws.


Assembling Everything Together

  1. Lay horizontal the back plate with the 3" flat socket head screws in place.
  2. Add the diamonds, half diamonds and 1/16" thick 5/16" USS washers over the screws.
  3. Add the 8 leaves in the full extended or open position; make sure that the splines engage with adjacent leaves.
  4. Lay on the cam ring such that the handles are in the open postion.
  5. Add the sides with their tongues into their grooves in the back plate.
  6. Lay on the cover plate, fitting the holes over the screws and side tongues into their grooves on the cover plate. Add brass washers and cap nuts. Tighten but not too tight.

Here are some assembly pictures:




(7.) Tilt the iris vertical and check for ease of opening and closing.

Here is the finished iris opening and closing.

Flash Video (flv) of Looking through the Iris

Hanging the Iris in Front of the Round Window

The iris is a 42" side-to-side octogon, 2-9/16" thick (excluding brass washers/cap nuts), and weighs approximately 47.5 lbs. The iris rests on a horizontal 2x2 at the bottom, with the 2x2 screwed into studs. Three Z-brackets (no. 47) hold the iris against the wall. The brackets are screwed into studs (with 3/8" diameter, 4" long hanger bolts) or the horizontal 2x2 (with a 3/8" diameter, 3" long hanger bolt) and act as clips on the iris. One can see the roof line is about 40° (10:12 rise:run) and not 45° (12:12 rise:run). Here are some pictures of the in-place iris:


Flash Video (flv) of Iris In-Place

Thank You and Postscript

Thank you all who have helped me on this project. I couldn't have done it without you.

I think the most important action was the CNC routing of the parts. This allowed a consistent accurate duplication of 8 identical paths/parts/?. The paths are easy to plan on a computer but would be difficult if not impossible to do by hand, especially for this woodworker (i.e, me.)

I have put a lot of detail in this blog so that if a reader is interested to build something similar, then they have a starting point. (Also I can go back when I need the information.) I have learned a lot of different things doing this project and I find that, in itself, a very rewarding bonus. I really encourage everyone to challenge themselves, I know you will be successful.

Here's a mechanical iris for an octogon opening.


Here is a pair of irises each with a transparent leaf. One for a circular opening and one for an octogon opening. The cam ring rotates 90° to open and close.
Opening Shape
Circle Octogon