Reflector Tech
From Fab Lab wiki
Contents |
Basic Science, Simple Solutions
Reflector Shape
FabFi reflectors use the properties of parabolas to focus the RF energy of the WRT54GL router into a narrow beam that can go long distances. (see the Tech Summary for an explanation of how this works). Though the reflectors we use are designed in CAD program and built using computer-driven machining tools, parabolas can also be constructed mechanically using a square and some string, meaning that a reflector could theoretically be designed in the field using whatever might be available.
For the FabLabbers fortunate enough to have access to CAD.py, the following function can be used to create a generalized paraboloid:
def paraboloid(x0,y0,a,b,h): part = "((((((X-x0)/a)**2) + (((Y-y0)/b)**2) - Z) <= 0) & (Z <= h))" part = replace(part,'x0',str(x0)) part = replace(part,'y0',str(y0)) part = replace(part,'a',str(a)) part = replace(part,'b',str(b)) part = replace(part,'h',str(h)) return part
Reflector Materials
The crucial component of any reflector is the surface of it's parabolic face. This material must be a conductive metallic mesh with a hole size smaller than 1/10 the wavelength of the radio frequency you are trying to reflect. The 1/10 wavelength number is an approximation that is generally considered to create an RF reflective surface of sufficient quality. Decreasing the mesh size generally increases the reflectivity of the surface (ex: window screen makes a better reflector than chicken wire). Given, that smaller meshes generally make better reflectors, our empirical evidence suggests that a solid conductive reflector (the limit of decreasing mesh size)does not make a good reflector. We hypothesize that the rounded-ness of the wires making up a true mesh compensates for imperfections in the surface of the parabola in a way that a solid conductor cannot.
Provided that the parabola surface is made of a conductive wire mesh, the rest of the reflector can be made from any material. We have used sign-board, plywood and acrylic with various degrees of success.
Reflector Construction
Because the wire mesh seems to compensate for slight imperfections in the shape of the reflector's parabolic face, there is no need to construct a one-piece solid reflector surface. Instead, we create our reflectors by assembling multiple 2D parabolic members into a grid shape that approximates a true parabola. This method of construction is ideal for the slotted, press-fit designs that can be easily created with the FabLab's ShopBot or Laser Cutter.
Slotted, press-fit construction has the advantage of straightforward, often tool-free, assembly, but has the limitation that the slots in each design must be modified based on the thickness of the material used. Some materials, especially acrylic, are very sensitive to slots that are too tight (even by .005") and will crack if forced together.
See the reflectors below for examples and design files.
Reflector Designs
NOTE: We're having some trouble getting directory browsing to work, so the links in this section are broken. For the most current design files go to The FabFi download site
All of the following designs are, or have been deployed in FabFi systems. A short description, design files any available specs are listed for each. Check the slot size of each design with a small piece of your material before cutting an entire reflector.
Note: Gain numbers listed below include a (somewhat arbitrarily added) 1dB internal cable loss in the wrt54gl.
Large Reflector (4-foot)
This is the highest-gain version we have designed. A pair can be cut from 5 half-sheets of 4'x8' material. Our deployed version uses sign-board that is nominally .5" thick. The reflector dish measures approximately 4' square.
We have tried two different materials as the dish surface, multi-layer chicken wire (layered with holes partially overlapping to decrease effective hole size) and window screen. Though the two surface materials have not been pitted head-to-head, the two materials gave similar gain results. Chicken wire tests yielded 15-15.5dBm gain, while the window screen yielded 14dBm. The tests cannot be directly compared, however because the window-screen test was not line of sight (some trees were in the way) and the fresnel zone was likely encroached upon. The design files for the pictured version can be found here.
As of 9/1/2009, this design also has a version with a built-in acrylic box to protect the router from the weather.
Small Reflector (2-foot)
The small reflector is currently the most used in FabFi. It measures 2' tall by about 16" wide, and has been implemented with both the multi-layer chicken wire of the large reflector and a knitted mesh, with a signboard (.5") frame. A pair of these can be cut from one half-sheet of 4'x8' material.
Gain testing with the knitted mesh version, yielded widely varying results based on installation. Our clearest line of sight with proper alignment wielded a gain of 10dBm. Other installations gave gains of 6dBm and 3dBm. In both of these later cases there was a significant amount of slop in the pointing and the fresnel zone. There was no clear relationship between the distance of the test and the results. The only gain measurement available for the chicken-wire version is from the Public Hospital downlink, at 8.5dBm. This installation had a lot of slop in the pointing, but clear line of sight. Pictured at left is the small reflector from the Public Hospital downlink (chicken-wire).
Also shown is a graph of the radial gain profile of the reflector generated using a version with the knitted mesh.
There are two frame versions for this reflector, one with a carry handle and one without. Find the design files for the carry handle version here and the original version here.
As of 9/1/2009, this design also has a version with a built-in acrylic box to protect the router from the weather.
X-Small Reflector (18-inch) -- "The One That Fits In The Laser Cutter"
This reflector was designed specifically for the Epilog laser cutter which has a 24" x 12" bed size. It was designed completely in Jalalabad, Afghanistan! Measuring approximately 15"x18" this design can be cut from four sheets of 24" x 12" material. Slot size varies a lot with common acrylic, but the file are designed around .22" acrylic (see note for CAD.py users at end.)
Using the same window screen material as used in the large reflector we observed a gain of roughly 5dBm with this reflector in the Bagrami link. The image at left is an early version of the design. The files linked below have been updated to more comfortably accommodate the wrt54gl. (in the new version, the wrt54gl mounts with the display panel of the router facing the reflector).
Find the design files here. the .png files are at 150ppi resolution and can be opened directly in CAD.py. The other files are 600ppi.
Note for CAD.py users: CAD.py by default interprets the reflector pieces in the .png files as waste. In other words, it cuts the toolpath inside the pieces. This is by design. Because the acrylic used in most fab labs varies so widely in thickness from piece to piece, the slot sizes need to be modified frequently. To increase the slot size, simply increase the tool size by the corresponding amount. Yeah, it's a dirty hack, but it's quick.
Also, we have found that using the Epilog with CAD.py requires the Y min and max be set to -12 and 12, respectively (instead of 0 to 12) for the files to cut correctly. This may or may not vary from setup to setup.
Wtgphoben 17:29, 24 March 2009 (UTC)

