Optical Transceiver

(That is, the box with the lenses in it...)

Why a two lens enclosure?

You might be reading this and ask why this (and the other) enclosure(s) have two lenses instead of just one?

So, why not just a single lens?

One of the biggest advantages of having a two lenses is that one can be dedicated for receive and the other for transmit.  In doing this, full-duplex operation (e.g. being able to transmit and receive at the same time) is possible which is not only very helpful during operation, but it aids in final peaking and alignment once the system is set up as the lightbeam path itself can be used to relay information about the signal quality.

Another advantage is that nothing needs to be done to the box itself when switching between receive and transmit as there are no moving parts.

What about a single-lens box, then, which could be smaller, lighter, and possibly cheaper?  It comes down to your performance needs and mechanical skill in construction, really.  There are two ways to use a single lens for both receive and transmit:
  • Use the same device for both emitting and detecting.  LEDs have the property that they can be used to detect light at the same wavelength that they emit, so it's possible for it to do double-duty.  The problem with this is that LEDs really don't make very good detectors owing to noise, leakage and capacitance but if your goal is to just be able to go several 10's of kilometers instead of hundreds of kilometers, this approach may be good enough!
  • Switch between a transmit and receive device.  This may be done by either physically moving the transmit and receive modules to the focus of the lens by mounting them on a slide or swing arm or by interposing a mirror in the optical path and moving it to select which device is at the focus of the lens.  Yet another approach taken by some folks in the UK was to mount the detector and emitter side-by-side equidistant from the center and simply rotate the entire TX/RX module to put either the emitter or detector at the focus of the lens.  This approach requires that  precision and repeatability be maintained as both the axial (left, right, up and down) and focus (in and out) position of both the emitter and detector be maintained to under a millimeter.  The other issue is that in order to go between receive and transmit it will be necessary to move things back-and-forth.  There is a risk when this is done manually as it must not only be done carefully as to avoid disturbing the aiming of the system on, say, a tripod, but it also must be done so that the mechanical elements are set at the transmit and receive positions securely each time the change is made.
Obviously, I have a bias toward a dual lens system when it comes to my recommendation as it offers the ultimate in simplicity of operation and performance, but depending on your needs and mechanical skills, you can make the appropriate choice for your system!

In order to construct an optical ("lightbeam") transceiver, this enclosure was designed to mount two Fresnel lenses, each being 318mm x 250mm (approx 12-1/2" x 9-7/8") with a 330mm (13") focal length and are about 2mm thick.  These lenses were obtained from Surplus Shed and were part number L3606:  As of 12/2007, they were no longer shown in their online catalog.

A "truncated" pyramid design was suggested by the Chris, VK3AML, and was selected as it is somewhat more compact and uses less material to construct - not to mention being somewhat stronger than a cubical enclosure, owing to the triangular construction.  Note that only the top/bottom panels are sloped and not the sides as, at the time, I didn't feel comfortable in trying to take into account so many compound angles..

The main enclosure body:

The body of the enclosure was constructed using "5.2mm Hardwood Plywood" - a 4' x 8' sheet of which was obtained at Lowes for about $12.  These sheets are 3-ply - not counting the two very thin exterior veneers - with the "finished" veneer (being "A" grade) being thin enough that it cannot take much sanding at all.  The obverse veneer is somewhat thicker and is of "C" grade.  It is worth mentioning that sheets of plywood this large and thin are not particularly flat - something to be considered during cutting and construction as this uneven-ness needs to be accommodated during construction:  The slots made from scrap wood seen in the photos help force the pieces into alignment.

For the lens mounts - and in a few other places - some 1"x2" poplar strips (actual dimensions are 3/4" x 1-1/2") - were used.  Poplar was chosen because it is relatively inexpensive, quite light in weight, and more durable than pine.

Because the plywood is fairly lightweight and not extremely strong, the back panel and the front cover were made by laminating two pieces of plywood together:  This was done by liberally applying yellow wood glue, tacking the two pieces together with small brads, and then clamping then between two sheets of heavier plywood in order to keep them flat.  After the glue has set, the brads are then removed and the holes filled in with wood putty.

After gluing, the back panel was cut down to size as necessary and the holes were cut using a 4" hole saw.  As can be seen from the pictures, scraps strips of plywood were used as guides to align the center divider as well as the sloping top and bottom sides and were secured using yellow wood glue and stapled into place.  These strips not only provide a guide for alignment and help straighten the material, but they also add to the thickness of the material at mating edges, providing more surface area for gluing and the application of brads.

Chris, VK3AML, pointed out that it is best to orient the plywood such that the end-grain of the middle ply is exposed on those edges into which brads and nails will be inserted.  For example, the end grain of the middle ply of the center divider panel is exposed along the rear panel.  Because nails need to be applied from all sides, the use of the "reinforcing strips" made from scrap plywood minimized this problem and increased strength.
Figure 1
Views of the enclosure before the sides were attached.
Click on either image for a larger view.
Side view of enclosure before side was
                    view of enclosure before side attached

In looking at the pictures carefully, one can see that the three pieces of the "1x2" poplar are used for not only mounting the lenses, but as the front support for the enclosure in the form of a top and bottom lens rail and a center divider rail.  In these pieces, 1/4" wide, 1/4" deep slots were cut using a router into which the lenses slide.  The center divider has three such slots:  The two slots on either side to accommodate the lenses, but also a slot facing backwards into which the plywood center divider is glued. Although the 1/4" (6.4mm) slot is a bit wide for the plywood's width (which is just under 5mm) the liberal application of wood glue makes for a very strong joint:  Ideally, the slot's width would be cut to provide a snug fit for the plywood using an adjustable dado on a table saw (or by using a reverse cut using a smaller router bit) - but I didn't have either of these at the time of construction.

The top and bottom lens rails are identical to each other and they also have slots cut into them - but the lengthwise slots are not continuous across the entire front.  Although it cannot be seen from the picture, the lens rail slots stop in the middle (at the center divider) and this was done to prevent the leakage of light from one lens (the "transmit" lens) into the other (the "receive" lens.)  As can be seen from the pictures, some "grabber" screws were used (along with yellow wood glue) to fasten the top and bottom lens rails to the center divider rail, resulting in a surprisingly strong "H"-shaped frame.  Of course, before gluing, one should check (and adjust) the fit of the lenses into the rails.

From the picture on the left, one can see that the lens rails have another slot  facing backwards to receive the sloping top/bottom panels of the enclosure.  Ideally, these would have been cut to match the 26 degree angle of the sloping sides - and I would have done this if I had a table saw and a dado, but I used, instead, a router to cut a 1/4" wide by 1/4" deep slot.  Using a sharp wood chisel, I then removed the edge on one side of this slot to allow the sloping top and bottom panels to be angled while still resting at the bottom of the slot.

Once the top, bottom, center divider panel, and rear panel plywood pieces were "dry fit" together (with the "front" facing down) with the poplar lens rail (which was now an "H" frame consisting of the three pieces screwed and glued together) to verify proper fit - and to make sure that the lenses would fit properly - and yellow wood glue was applied liberally to the "rear" slot of the front lens rail and the edges of the center divider panel, the rear panel was dry-fit into place, and about 100 pounds (approx. 45kg) of weight was set on the rear panel.  Alignment of all of the panels was immediately re-checked, and the glue was allowed to cure.

After the glue holding the front panel into place cured, the rear panel was lifted off, glue applied to the mating surfaces, the rear panel re-set into position and then small brads were used to tack everything into place while continually checking to make sure that all panels were in alignment.  After tacking, the enclosure was placed face up (with the rear panel on the ground) and a piece of plywood was put across the front and the 100 pounds was again used to compress everything together while the glue cured.
Figure 2
Top left:
  The attachment of the "tab" on the side panel to the rear panel.  Top right:  The attachment of the side panel to the lens rails.  Bottom left:  View of the enclosure with the side panels attached.  Bottom right:  One of the side lens rails, showing the attached alignment tab and the rabbet used to hold the side panel in position.
Click on an image for a larger view.
Close up of side-panel attachment to rear
Close up
                    of side panel attachment to lens rail
                    view of enclosure before side attached Close up
                    detail of the side lens rails

The side panels were fabricated with a "tab" protruding from the rear that allowed attachment of the side panel to the rear mounting panel.  As can be seen from the pictures, scrap strips of plywood were attached using staples and glue to the outside edges of the top and bottom panels to double their thickness near the edge to provide additional surface for gluing and nailing.  Glue was applied to the edges of the side panels and they were then tacked into place with small brads and the glue was allowed to cure.  After curing, a bead of black RTV was run along all inside edges - especially the center divider.  This RTV provides not only additional structural strength, being black it also provides an effective barrier to light that might leak from the transmit side of the enclosure to the receive side.

Along the sides of the front of the enclosure are the side lens rails.  These have a 1/4" x 1/4" slot cut in them to match the top, bottom, and center lens rails, plus there's a rabbet cut along the rear edge to accommodate the thickness of the paneling of the side panel:  This detail can be seen in the lower-right picture in figure 2.  In the center of each side lens rail is a scrap piece of paneling that is tacked and glued to the side lens rail over the rabbet:  When installed, the side panel is inserted into the slot formed by the rabbet and this piece (the "alignment tab") and this provides rigidity to the side panel.  Note that the lens rails are to be removable to allow installation/removal of the lenses as necessary.

Optical receiver and transmitter mounting

In future enclosures, I did not use the same mounting scheme for the emitter and detector modules - see below.
This enclosure was constructed to a depth somewhat shorter than the focal length of the lens to allow precise focusing of the optics to be accomplished.  In order to do this, some sort of fitting is required on which the emitter and detector can be mounted:  I chose to use 3" ABS pipe hardware for this.

As can be seen from the picture, the receptacle for mounting the emitter and detector is constructed from a 3" ABS pipe coupler.  As it turns out, this coupler has an outside diameter of just under 4 inches, hence the holes in the rear panel.  In order to securely mount these, I used some 2" angle brackets.

In reviewing the geometry of the enclosure and the relationship of the lens, the light rays from the lens, and the size of the pipe coupler, I realized that the inside diameter of the coupler was small enough that it would block some of the light from the "long" dimension of the rectangular Fresnel lens and to minimize this blockage I cut slots in the pipe coupler.  Because of the limited room on the back panel to which the angle brackets could be mounted, the angle brackets were not mounted at 90 degree intervals - but rather, they were mounted as shown in the picture.  Using 6-32 hardware, a pair of these couplers was mounted to the back panel and then black RTV was applied around them to provide lightproofing as well as to increase rigidity.

After the unit was assembled, I noted that some slight blocking of light was still occurring around the edges of the Fresnel lens.  To remedy this, I removed the mounting tubes and cut them back even further, as shown in the pictures.
After the RTV was cured, the side lens rails were installed, the front surfaces were covered with masking tape and the interior surfaces of the enclosure were sprayed with several coats of flat black paint in order to minimize the effects of stray off-angle light.

Figure 3
Upper left:
  Close up of the 3" ABS coupler with the sides slotted and angle brackets attached.  Upper right:  The coupler mounted in position.  Middle left and Middle right:  After the mounting tubes have been cut down further to minimize light blockage.  Bottom left:  The inside of the enclosure, with couplers installed and having been RTV'd into place and with the interior surface of the enclosure painted with flat black paint.  Bottom right:  This is what it looks like when there is no picture.
Click on an image for a larger view.
Side view of enclosure before side was
                    view of enclosure before side attached
                    version of the mounting tubes, after further cutting
                    down of the sides
                    version of the mounting tubes, after further cutting
                    down of the sides
                    inside the enclosure after mounting couplers and
                    painting the inside black
Front cover/Support base

In designing the enclosure I decided to make the front panel of the enclosure perform a double-duty:  It should not only protect the lenses both while in transit - to prevent them from being scratched, but also to to keep direct sunlight out to avoid accidentally setting fire to anything with the focused light, but it could be used as a firm attachment point for a tripod or as as platform to be set on the ground or a table.  Because of this, the front panel is mostly made from two pieces of plywood laminated together.  Note, however, that the panel is not entirely double-thickness:  It is only a single thickness near the hinges as a double-thickness would interfere with the bottom lens rail as it was folded backwards. 

The front panel pivots on four 3 inch "Bi-fold Door hinges":  These are "split" hinges designed to be flush-mounted and as such, their mechanism's range of motion is limited only by what is attached to them which, in theory, means that they have nearly a 360 degree range of motion.  Note that the "inside" portion of the hinge is mounted to the lens rail, while the "outside" portion of the hinge is connected to to the cover.  This was done because the poplar lens rail is quite strong, while the cover plywood - being only a single thickness - is quite weak by comparison.  By mounting the cover to the "outside" hinge segment, the mounting screws are spaced farther apart and will better-distribute load across the single-ply panel.

Although it is difficult to tell from the pictures, the hinges are mounted offset from the lens rail by the thickness of the plywood.  When the panel is closed, protecting the lenses, the two metal halves of the hinges lie side-by-side, but when the panel is flipped backwards, the hinges need to be offset from the lens rail slightly to accommodate the fact that the thickness of the plywood is between the hinge and the lens rail.

While wood screws were used to mount the hinges to the lens rail, 6-32 screws and nuts were used to bolt the cover to the hinges as there is too little material to accommodate a wood screw.  In the picture, one can see that countersinks were drilled into the lens rail to accommodate the thickness of the 6-32 nut and the length of the screw:  If this had not been done, the cover could not be folded backwards.  It should be noted that upon final assembly (after finishing) 6-32 flat washers were added between the nut and the wood to spread the stress of the screws on the wood and blue "Locktite" (tm) was put on the threads to keep the nuts from loosening.

In order to provide a "stop" for the cover when folded backwards under the enclosure, another piece of poplar was cut and a 1/4" wide slot was cut into it - the depth of the slot being calculated such that when the bottom cover was folded over, it's surface was at a right angle to the lens plane and this piece was then glued to the bottom edge of the rear panel.  Now, if I had planned ahead sufficiently, I could have simply made the rear panel large enough (or offset it) so that the bottom edge would have performed this function and I could have avoided the need for this extra piece of wood!

Once all of the pieces were checked for proper fit and attached, all of those pieces that could be removed or disassembled were removed and all exterior surfaces (plus the interior surfaces that were not already painted black) were finished - first, with a Minwax (tm) red oak finish, and then with two coats of semi-gloss polyurethane finish.  After these finishes had dried, the enclosure was re-assembled.

In order to provide a firm tripod mount a piece of 1/8" thick aluminum, 130mm x 200mm was cut.  In the center of it in a line I drilled and tapped seven 1/4-20 holes to accommodate a standard tripod screw.  These holes were in a straight line and spaced 1/2" apart for several reasons:

The center of gravity of the enclosure (with the lenses mounted) was determined empirically and the position of the plate was selected so that the center hole in the plate was at that location.  The position of the plate and its holes (both the 1/4-20 tripod mounts and the four corner mounting screws) were marked, 3/8" diameter countersinks were drilled into the cover at the position of each of the seven 1/4-20 mounting holes, and four more holes were drilled for the 6-32 screws used to mount the aluminum plate to the cover.

Comment about a tripod mount:

Figure 4
Top left:
  The enclosure with the front cover folded backwards.  Top right:  An "inside" view of the hinges, showing the countersunk holes that accommodate the mounting screws, and also the need for a single thickness of wood near the hinges in order to allow full movement of the hinges.  Bottom left:  Close-up view of the hinge.  Bottom right:  The front cover, folded completely back and meeting the rail on the rear panel.
Click on an image for a larger view.
Close up of
                    side-panel attachment to rear panel
Close up
                    of inside of hinge point
Close up
                    of the hinge mounting
Extension of rear panel as a stop for the
Mounting the lenses

Due to a miscalculation, the width of the front panel lens rails (inside the grooves) was inadvertently made about 1/4" wider (on each side) than the 250mm width of the lens - and this meant that the lens was likely to slide around a bit, but it also meant that if the lens was slid to one side or another, it would not engage the opposite lens rail slot simultaneously.  To fix this, some 1/4" wide shims were cut from some 0.062" (1/16") printed circuit board material, placed in the slot in the center lens rail, and secured in place with black RTV.  Only three shims (each about an inch long) were used:  One in each corner where they can provide additional light blockage between the two lenses, and one in the middle.

As mentioned before, the thickness of the Fresnel lens itself is about 2mm.  What is also important to note is that the lens is rather fragile - especially the "grooved" side, which must face outwards.  Were these grooves not protected they could be easily damaged by abrasion or worse, they could accumulate dust and dirt:  A protective sheet is required.

I found at Lowes a 18" x 24" piece of 0.08" (approx. 5/64" or 2mm) thick Plexiglas that I cut into two 250mm x 318mm pieces to precisely match the size of the Fresnel lenses.  This Plexiglas was cut by first scoring the plastic on both sides and then snapping it along the score line with a crude brake made by clamping the Plexiglas between two pieces of wood and using a third piece of wood to snap it.  If a fine-toothed bandsaw or table saw is available, it can be easily cut with that.

Because the combined thickness of the Fresnel lens and the Plexiglas was only about 0.16" (5/32" or 4mm) additional thickness was required to fully fill up the 1/4 lens rail.  This was accomplished by solvent-gluing 8 small shims cut from scraps of Plexiglas to the edge of the protector using "free-flowing" solvent-welding cement, "model glue", a thin solution of acetone or Methyl-Ethyl-Ketone with some dissolved Plexiglas or even some clear RTV (silicone) adhesive:  One small, square piece was used in each corner and a slightly larger strip at the midpoint on each side.

The combined thickness of the protective cover plus these shims is about 0.24" (6mm) - a very close match to the 1/4" (6.4mm) lens rail width and when slid into place, the protector and the Fresnel fit fairly snugly.  By providing a gap between the protector and the Fresnel lens there's a reduced possibility of damage to both the pieces because they cannot come into contact with each other and scratch or ablate their surfaces and they can move around freely as their relative positions shift slightly with temperature and while being transported.  Because of the support of the lens and its protective cover by the lens rails on all four sides, there has been no tendency of either the lens or the protective cover to bow or warp on any of the units that I have built!.

Once the lens and protective cover are slid into the lens rails the side lens rails are screwed into place:  The rubber padding prevents side-to-side movement of the lens assembly while still allowing for differing coefficients of expansion of the plastic and wood pieces.  In addition to the use of the foam rubber to prevent the lens from sliding side-to-side, a small piece of foam was added at the top of the lens to prevent vertical movement:  A 1/2" hole was drilled into the center of each top lens rail and a 3/8" dowel with a piece of the same foam rubber was inserted into the hole, with the foam holding both the lens and cover plate in place, vertically.  Atop the dowel was a small spring (to maintain compression after the foam had conformed to its final shape, holding the lens and cover in place) and a small piece of plywood was screwed into place to hold everything else in place.
Figure 5
Top left:
  The side lens rail with a piece of rubber foam installed.  Top right: A shim on the corner of the Plexiglas protector.  Bottom left: Near edge-on view of protector lens and the Fresnel.  Bottom right:  A front-on view of the enclosure with light being emitted.
Click on an image for a larger view
Close up of foam spacer in side lens rail
Close up
                    of shim piece attached to protective lens
Close up of the lens with the plexiglass cover
                    sheet with spacer Front view of enclosure with light source

Electronic assembly mounts:

The mounts for the electronics contain three pieces:

The mounting tube:

This consists of a short piece of 3" ABS tubing of the sort typically used for wastewater pipes.  The type that I used is the "foam" type, so-called because it is infused with tiny air bubbles during manufacture - a process that makes it far lighter in weight and easier to work than solid ABS.

Because this ABS tubing is intended to provide a very snug friction fit with its couplings but not easily be removed, it is necessary to reduce the outside diameter of the tubing slightly in order for it to be easily slid in and out of the receptacle tube on the enclosure.  This diameter reduction was accomplished by clamping a piece of the tubing (already cut to length) in a vise, sanding it with 60 grit paper, rotating the tube 90 degrees, and then sanding some more in overlapping patterns.

Because it is foam-based this tubing acquires a velvety texture when sanded, causing quite a bit more friction that would normally occur when inserted into the receptacle tube.  To remedy this, I wiped the tube very quickly (using a rag) with lacquer thinner:  The solvent nature of the lacquer thinner lightly dissolved the fuzzy surface, reforming it as a solid, smooth surface.  When wiping the tubing with the thinner, one must move quickly, or the rag will stick to the ABS, leaving a mark and deforming the surface.

Later, when additional emitter and detector units were constructed, small slits were cut into the tubing with a fine-toothed "jeweler's saw" instead of reducing the diameter of the ABS tubing.  This small kerf allowed a snug fit into the receptacle by allowing the diameter of the pipe itself to compress slightly as it is inserted.
As can be inferred from the picture Figure 6 four screws were used hold the piece of tubing the the circuit board material.  The center of the square circuit board material was located by drawing an "X" on the board, and using a ruler, the tube was centered precisely.  At that point, the inside and outside of the tube was traced using a pencil and marks were placed on the tube, along with a corresponding mark on the circuit board material so that the orientation of the tube could be repeated.  After this was done the precise location for the mounting holes were located by marking midpoint between the inside diameter circle and the outside diameter circle where the line for the "X" crossed the circles.

Holes were drilled to accommodate the 1/2" long #6 wood screws and the opposite side of the material was counter bored so that the heads of the screws would be flush with the board's surface.  After drilling in the circuit board, the tube was aligned with the holes, the positions marked on the edge of the tube, and then 1/16" pilot holes were drilled in the edge of the tube.

After verifying a good fit, the tube was removed, the edge coated with black RTV and then the tube was reattached with screws and the RTV that squeezed out was smoothed  to form a nice fillet between the tube and the PC board material using a wet finger.  Again, this RTV provides the majority of the physical strength of the attachment between the tube and the board material as well as providing the very important function of providing a light-tight seal between the board and the tube.  After the RTV cured, the inside of the assembly was painted with flat black paint to minimize stray reflections.

The emitter mount:

When operated anywhere near their maximum ratings, the Luxeon Emitters must be mounted on a good heat sink with low thermal resistance from the LED's slug (the aluminum substrate) to the heat sink itself.  When I first went looking for Luxeon devices, my source had only the raw emitters (instead of the Luxeon Stars) in stock:  The Luxeon Stars are simply an emitter that has been already mounted to a small star-shaped aluminum heat-spreading plate.

Figure 6
Top left:  Raw pieces of the mounting tube that connects to the enclosure to provide focusing.  Top right:  The emitter mounting assembly.  Upper-center left:  Luxeon emitter epoxied to the heat sink.  Upper-center right:  Close up view of the Luxeon emitter on the heat sink.  Lower-center left:  The PCX lens mounted and spaced above the emitter.  Lower-center right:  A rear view of the lens mount.  Bottom-left:  Emitter and lens assembly on mounting tube.  Bottom-right:  The "business end" of the mounting tube, showing the emitter (and connecting wire) through the PCX lens.  Note:  The emitter is slightly "off center" from the mounting tube owing to a slight miscalculation of the locations of the positions of the mounting tubes' mounts on the enclosure.
Click on an image for a larger view.
View of some of the pieces of the electronic
                    assembly mount
                    view of enclosure before side attached
High-power LED mounted on heat sink
Close up
                    of high power LED mounted on heat sink
Plano-Convex lens mounted in front of LED
Rear view of mounted lens
                    and lens assembly mounted in position
                    end of emitter and lens assembly
In doing some research, I determined that a 2-part metal-filled epoxy adhesive commonly available in the U.S. called "J.B. Weld" (tm) also had excellent thermal conducting properties and was used widely amongst the computer gamers to attach large heat sinks to their overclocked processors.  In past experimentations I had determined that this epoxy did, in fact, provide both excellent thermal conductivity as well as adhesion to the surface.  Because this epoxy is also rated for high temperatures (it is intended to repair engine parts) there was little concern that the heat of the Luxeon could compromise the bond.  J.B. Weld can be easily found at many auto parts stores and home improvement centers in the U.S.

Rummaging around the junk box I found some small '486-type CPU heat sinks that were large enough to adequately dissipate the LED's heat.  Locating the center of the heat sink (using the "X" method described above) I lightly scribed a circle around the center to provide a reference point.  I then straightened the LED's leads, degreased both the heat sink and the LED's metal slug using denatured alcohol and placed a small dab of epoxy in the center of the heat sink.  Placing the LED carefully in the center of the scribed circle I clamped the LED and heat sink in a vise overnight, using a "Pink Pearl" pencil eraser to protect the LED's body and to provide a springy, yet firm, compression surface.  Leaving it in the vise until the next evening, I had a nicely-bonded LED as seen in the picture.

It should be noted that the raw Luxeon Emitter has 10-15% lower thermal resistance to its heat sink than a Luxeon Star, so given equal heat sinks the emitting die of a raw Luxeon Emitter mounted in this way will run cooler than a Luxeon Star, potentially yielding more light output for a given set of operational conditions as light output for these emitters drops with increasing temperature.  I have had the occasion to remove a damaged Luxeon Emitter (after accidentally fusing a bond wire) from a heat sink attached with the described method and found that it took quite a bit of force or a lot of heat to do so.  I also noted that the epoxy layer between the heat sink and emitter's slug was extremely thin - mostly having been extruded by being clamped during curing:  The thin-ness of this layer helps contribute to the efficient heat transfer from the slug to the heat sink!

For more general information about the Luxeon emitter, see this page.

Comments on mounting the emitter and detector modules:

As can be seen from the pictures in Figure 3 the emitter and detector modules were mounted by using short pieces of ABS pipe along with ABS pipe fittings.  To mount the modules, these are simply "plugged in" and the friction is used to hold them into place.

While this system works, the later enclosures use a different scheme:  Flush-mounting of the emitter/detector modules to the rear panel as described on the foldable enclosure pageThe reason for this change is that the flush-mount system is actually simpler, consisting of just a hole in the box and the four tee-nuts used to secure the electronics to the enclosure.  When the units are bolted to the back of the other enclosure, there is no uncertainty with regards to making sure that they are being pushed in all of the way (which would affect focus) or axial alignment caused by not inserting the module the right-way up!

Since I didn't measure perfectly when making this enclosure, the emitter and detector aren't in the exact center of the hole which means that if they are rotated when plugged in, they won't be exactly centered at the focus of the lens.  Because of this, these modules all have "alignment" marks that the user must observe when the emitter/detector units are plugged in.

Matching the emitter and the lens:

It was originally thought that the percentage of light lost due to the mismatch of the radiation angle of the Luxeon (the Lambertian pattern) and the subtended angle of the lens at the focal point was not likely to be significant.  In subsequent geometrical calculation as well as actual testing testing it was noted that more than 50% (more like 75-80%, as it turned out) of the luminous flux was lost due to this mismatch!  I was able to locate some "strong" PCX (Plano-ConveX) lenses (48.5mm diameter glass lenses with a 51mm focal length) from American Science and Surplus (P/N 67956, for $4.50 each) that worked very nicely as a "secondary" lens.

More info about the necessity of a secondary lens in certain instances may be found at this siteScroll down to the middle of the page and look for the diagram labeled "Enlarging the Effective Source Size with a Secondary Lens."

As can be seen from the pictures, the lens was mounted in a piece of 0.062" glass-epoxy circuit board material.  A hole, slightly larger than the lens, was cut, using a hole saw, in the material.  The circuit board material was clamped to a flat piece of wood (using clothespins) and the lens centered in the hole - and a piece of a polyethylene bag was placed between the lens/circuit board and the piece of wood as a release agent:  Epoxy does not stick to polyethylene!  At this point, J.B. Weld epoxy was used to fill the gap between the circuit board and the perimeter of the lens - and because the edge of the lens is slightly beveled, there is a fairly large "capture" area for the epoxy.  After it had cured, the polyethylene was peeled away from the back of the assembly and the slight amount of epoxy that had gotten into the visual portion of the lens was carefully scraped away with a sharp knife.

To determine the precise focus I marked a spacing of 330mm (the focal length of the lens) and marked the large dimension of the lens (318mm) on a piece of paper.  By doing this, I was able to determine what distance the secondary lens had to be from the emitter to cast a circle of light that was about 318mm in diameter.

Because these Fresnel Lenses are rectangular rather than square (318mm x 250mm) some light would be lost due to overspilling on the 250mm side:  I set the emitter-lens distance to create a circle that was a little under 318mm diameter when the emitter-lens assembly was at the focal length distance of 330mm.  In other words, when illuminated a rectangular lens, it is best to "just touch" the edges of the long dimension with the LED's circle of light while allowing light to spill over on the edges of the narrow dimension, a configuration the results in somewhat darkened corners.  Both emperical testing and later ray-trace analysis has shown that this results in very close (within a few percent) of optimal illumination and source-size magnification and subsequent far-field flux.

Knowing this distance (around 6 millimeters or so) I mounted the piece of circuit board material with the lens to the emitter heat sink, using standoffs to set the appropriate distance:  This assembly can be seen in Figure 6.


Initial lens focusing and alignment - the transmitter:

In order to maintain the best optical efficiency, it is necessary to obtain good alignment of the optical system:  This alignment not only requires proper focusing of the optical elements (both transmit and receive) but lateral (side-to-side) positioning to place the element in the point of best focus of the lens.

To do this, I used a carpenter's square and clamped to it a laser level as can be seen in Figure 7.  By holding the square against the front lens rail, which is precisely parallel to the plane of the lens, I could determine where the lens itself was aimed.  Note:  It is necessary to do this in both the horizontal and vertical planes:  The picture shows the laser being aligned in only one plane.

At the far end of my basement I made a paper target (see Figure 7) that contained three marks:  The one on the top that corresponded with the position of the laser as shown in the picture, another on the side that corresponded with the position of the laser when aligned to the box in the "horizontal" plane (note that the box is on its side in this picture - an arrangement more convenient at the time) and another mark that is measured to be in the center of the lens.

Just below the paper target (but not easily seen in the picture) I mounted a red LED below the center "transmit" cross hair target at precisely the same distance as the spacing between the center of the two lenses.  In this way, the receive system could be aligned by modulatint the LED with a tone, and then adjusting for proper alignment by noting the position of the loudest tone.
Figure 7
Top left:
  Using a carpenter square clamped to a laser level to determine the alignment of the lens.  Top right:  At the far end of the basement (about 32 feet away) is a target used for aligning the receive and transmit lenses.  The "alignment dot" of the laser is seen at the top of the target while the LED's image can be seen in the center of the target:  An LED used for generating a receiver alignment signal can just be seen below the paper target, attached to the wall.  The "orangeness" of the colors is due to the overloading of the camera's imager - even through the LED current was turned way down.  Upper-center left:  Front view of the enclosure.  A "wide angle" effect causes a distorted image of the wall behind the enclosure to be visible.  Upper-center right:  Side view of the enclosure showing the front cover being folded underneath, acting as a support.  Lower-center left:  Rear of the enclosure showing the orientation of the mounting tubes.  Lower-center right:  A dramatic illustration of the sun's rays being focused by the lens, instantly igniting a scrap piece of the plywood used to construct the box.  Note that the focus is safely behind the enclosure, as shown by the sharply-defined ray edges in the smoke.  This experiment in solar combustion was done on a winter day through a double-pane storm door.  Bottom left:  The elevation adjustment mechanism of the enclosure.  Bottom right:  The "alignment target" used for adjusting the receiver and transmitter.
Click on an image for a larger view.
Use of a square and laser level to align the
                    transmit optics
Alignment target used for setting up the
Front view of the enclosure
                    view of the enclosure
Rear view of the enclosure Dramatic
                    illustration of sunlight focused by the lens
Elevation adjustment mechanism of the
Alignment target used for adjusting the
                    receiver and transmitter

Using the laser and square, the enclosure was first aligned precisely onto the target.  At this point, the emitter assembly was moved around on the backside of the mounting tube so that the center of the LED's light fell precisely on the cross hairs.  As precise alignment was found, the emitter/lens assembly was soldered into place on the mounting tube assembly using short pieces of #12 wire.  After soldering, the alignment was re-checked with the laser/square to verify that the enclosure had not been bumped during adjustment.

Note:  Because the distance between the target and emitter was fairly short (only about 32 feet or 10 meters), parallax is still a major concern:  At much greater distances, the parallax will become irrelevant as the beamwidths of the transmit and receive parallel beams will merge.  Also, fine-tuning of focusing for longer paths is still likely to be required as the focus for these comparatively short distances is going to be slightly "off."

Through experiment, I determined that proper focus at "infinity" can be approximated thusly using a short-range (30 feet or 10 meter) test range:
With the above procedure, the focusing will be "close" to optimal, but it is still necessary to focus the spot on a target (such as the side of a house) that is at least several hundred yards/meters away, adjusting for the sharpest "square" of light.  Better yet, an electronic means of light measurement can be used - more on this below.

Although it is difficult to tell from the somewhat overexposed image in Figure 7, the paper target shows, in some detail, the top of the Luxeon's light-emitting element - along with the bond wire and some of the metallic connections - focused onto the target.
  In other words, when you adjust for best light output, you are trying to project a "picture" of the LED's emitter at the distant (receive) site!

Initial lens focusing and alignment - the receiver:

Once the emitter had been aligned it was much more convenient to "point" the enclosure at the target than to repeatedly check with the laser/square combination, so it is now practical to do a similar alignment for the receiver.

With the transmitter properly aligned (and shut off) the LED mounted below the transmit target was weakly modulated with a tone.  The center of focus was found simply by moving the detector around until highest amplitude of tone was noted, as measured using an oscilloscope or AC voltmeter.

Once the "center" had been found, the detector was temporarily held in place with small clamps (I used "binder clips" - those black metal spring-steel clips used for holding stacks of papers together) and the focus was adjusted for highest amplitude.  After this, the "center" was checked again, followed by a fine-tuning of the focusing.

This iterative process is necessary because as the proper focus is achieved, the "focus spot" becomes smaller and smaller and any slight offsets are going to be exaggerated as the detector is brought into sharper focus.  One method to verify that the focus is fairly close is to block different portions of the Fresnel lens:  If properly focused, one should be able to block any half of the lens and note that the signal drops by half as well:  If this isn't the case, that means that not all portions of the lens are being focused onto the detector.

As in the case of the emitter, the focus setting at this (relatively) short distance is not going to be optimized for longer distances:  This procedure just gets you "within the ballpark" when you go out onto the longer-distance test range!

Final lens focusing and alignment:

Once the transmit and receive adjustments had been "roughed on" using the indoor range, it was necessary to verify proper aiming and focus over a longer distance.  In this case, I set up a test range over a much larger distance - about 525 feet (160 meters) which is a much closer approximation to infinity than was obtainable with the indoor range:  For the sizes of lenses that I've used (up to about 430 millimeters on a side) this distance seems to be a pretty good approximation to "infinity" while not being so much distance that it becomes impractical to both see the alignment target (see below) from the optical transceiver's location and/or to walk between the two to set up/take down the gear!

To properly do this adjustment I had to construct the "alignment target" seen in the bottom-right image of Figure 7.  As you can see from the picture, this is constructed of a large piece of cardboard (a discarded box) and on it are black and white lines constituting crosshairs (which may be seen from a distance) spaced at the same distance as the centers of the receive and transmit lenses to allow compensation for parallax.

Also mounted on this target are two electronics circuits:  On the right is an "optical beacon."  This is simply a 4060 counter/oscillator connected to a 500 kHz ceramic resonator and to its divide-by-512 output is connected an LED with a potentiometer to adjust the LED current to yield a stable 976.5625 Hz (approximately) tone.  The LED current is adjusted to be just be visible to the naked eye at night at the 525 foot distance.

On the left of the target is an "audible light meter."  This is a simple circuit consisting of a Cadmium Sulfide (CdS) photocell connected in a 555-based oscillator circuit.  Wired thusly, the pitch of the resulting tone is roughly proportional to the conductance of the photocell - which, in turn, is more-or-less proportional to the luminous intensity.  Attached in front of the CdS cell is an "cellulosic annular optical view confinement device" (a portion of the cardboard tube from a roll of paper towels) to reduce the effects of stray light.  The output of the 555 is also coupled, via a capacitor and potentiometer, to a cable with a connector that plugs into a handie-talkie.  In this case I used my old FT-470 HT tuned to a 70cm frequency and set to low power, using a step attenuator to reduce the radiated power to a minimum from the small rubber duck antenna to the microwatt level - just enough to be easily copiable from 500+ feet away.  Using this configuration I was able to monitor the pitch of the tone and thus get a relative indication of luminous intensity on the target at the optical transmitter.  The use of an audio frequency counter or computer running a program such as Spectran or Spectrum Lab to visually display the frequency with a "waterfall display" allows precise measurement of the received frequency to facilitate peaking.

Using these two pieces of simple gear I was able to make certain that both the transmitter and receiver were pointed parallel to each other by making sure that the peak amplitude of the received signal coincided with the emitter's beam being peaked on its respective target.  I was also able, using Spectran, to verify the precise focus of the receiver by noting the axial and focus position of the detector and determining the maximum-recovered audio as well as verify that both the primary and secondary lenses of the emitter were optimally adjusted to provide maximum luminous intensity at the distant target.

It should be noted that the frequency-versus-light intensity indication from the above CdS/555 circuit is not linear, so the frequency alone cannot be used to determine the light intensity, but if you do have a means of measure the frequency and LED current you can make some meaningful readings if you are careful.  Because the optical output of the LEDs is proportional to the current - particularly when the LED is being operated at low currents - say, 10% of maximum - you can use this fact, along with precise frequency readings, to make reasonably accurate relative measurements.  For example, if you know that "X" milliamps produced a tone of 1000 Hz before an adjustment but it took "Y" milliamps after that adjustment, you can use that difference in current to closely approximate the efficiency change.

(There are "light-to-frequency" devices available - particularly those made by Taos Semiconductor - now "AMS" - that might be useful for such measurements.  An example of such a component is the TSL230.)

Pointing the photons in the right direction:

Out in the field, one has to contend with adjusting both the azimuth and elevation of the beam appropriately.  Given that one has a flat, stable surface such as a small, portable table on which the enclosure may be set, the azimuth is a bit easier to handle than the elevation in that one simply rotates the entire enclosure slightly to center the beam on the receive end of the path.  Alternately, if you have a suitable tripod, this may also be used to mount the transceiver.

Tweaking the elevation, however, is another matter.  This need was addressed in the manner shown in the bottom-left image in figure 7 by using a piece of threaded rod, along with the enclosure's cover, to provide a means of adjusting the elevation:  When being transported, the hinged cover shields the lens, protecting them not only from mechanical damage, but preventing their (accidental) exposure to sunlight.  Upon setup, the cover is folded underneath the enclosure and is used as part of its base.

The elevation adjustment mechanism is mounted to the main enclosure using 10-24 screws with wing (or "butterfly") nuts that screw into tee-nuts that are pressed into the wood on the opposite side.  In this way, the entire adjustment mechanism may be removed or, alternatively, the bottom portion may be detached from the front cover, allowing it to be folded back for transportation.

The bottom portion of the threaded rod screwed into a small piece of 1/2" aluminum rod into which some 1/4"-20 threads were tapped.  The bottom of this rod was turned down so that its outside diameter matched that of the inside of the bearing, threads were tapped into the bottom side (opposite the 1/4"-20 thread) and a flat-headed 6-32 screw was used to hold the aluminum piece to the bearing.  The bearing assembly was then held in place using some flat washers (with holes to provide clearance for the flat-headed 6-32 screw) and a spring.  With this mechanical assembly, the bearing/threaded rod is able to move about to accommodate the threaded rod as its angle changes from vertical as the elevation is adjusted to its extreme, yet the spring keeps the entire assembly from loosely moving around.

Also on the threaded rod are two stops using jam nuts:  The bottom jam nut prevents the elevation from being adjusted too far in that direction - something that would likely tear the tee nut out of its mount - while the upper jam nut prevents excess travel in the other direction.

You may notice that this assembly only allows "downward" adjustment of the elevation - and given any path with both ends at the same altitude, the two ends would always be pointed downwards.  Practically speaking, however, if one end is at significantly higher elevation than another - or if the enclosure is placed on a surface that isn't perfectly flat, it may be necessary to point the enclosure upwards slightly.  This can easily be accomplished by shimming the enclosure with a piece of wood, a book, or whatever might be handy:  Remember that it's not pointing the enclosure up or down that's particularly difficult - it's doing so in a precise and easily controllable manner that takes some care!

Even with a typical tripod, the use of this elevation screw is recommended as most photographic tripods have no means by which elevation can be adjusted other than by loosening the head and moving it up-and-down:  In doing this, it can be tricky to make precise, repeatable up-down movements without inadvertently affecting the azimuth, so having an "elevation" adjustment that is completely independent of the tripod is quite useful!  Some special-purpose tripods such as those used for telescope mounting, survey equipment or motion-picture cameras do have means of "smoothly" adjusting azimuth and/or elevation:  If you are lucky enough to have such a device and it can safely support your optical gear, by all means use it!

Spot Quality comparisons:

As was expected, the higher-quality optical acrylic Fresnel lens used in this enclosure produced better-quality "spots" than the inexpensive, vinyl "full-page magnifier" lenses in the "Cheap Enclosure."  During testing, I decided to do a direct comparison between the two.
Figure 8:
  Spot produced by the "first" (wooden) enclosure with the high-quality acrylic Fresnel lens.
Bottom:  Spot produced by the "second" (posterboard) enclosure with the vinyl "page magnifier" lens.
Both of these images have been converted to grayscale for easier comparison, and were taken using identical focal length and exposure settings and have been identically processed to show relative spot size, brightness, and beam containment.
Click on either image for a larger version.
Spot produced by the first (wooden) enclosure
Spot produced by the second (posterboard)

Ideally, one would have done everything possible to make the two tests equal, but there were some unavoidable differences that may affect the accuracy of the direct comparison of intensity:
In direct comparisons, it was determined that, given the same LED current, the "cheap enclosure" had about  38% of the luminous flux of the "first" enclosure.  This comparatively poor performance is due partially to the inferior optical quality of the flexible, vinyl lenses, but it mostly as to do with the fact that on this enclosure I didn't make too much of an attempt to optimize performance by selecting the best "secondary" lens for the job:  In other words, most of the light being emitted by the LED doesn't even get to the backside of the Fresnel lens in the first place!  Since this transceiver was intended to be "quick and dirty" and just "good enough" for initial testing, I've not felt the need to revisit it.  Of more relevance here is that this demonstrates how important it is that one carefully selects the proper secondary lens when one is constructing a high-performance optical transceiver!

Of more interest was the "quality" of the spots that the two boxes produced even when optimally focused.  As can be seen from Figure 8 the "main spot" (the brighter "square" portion) is almost identical in size, but the lower spot (from the low-quality, flexible page-magnifier Fresnel lenses) is not only dimmer, but more light is spread out beyond the main beam perimeter.  It should be pointed out that this effect is apparent not only from the pictures, but is arguably more visible when view with one's own eyes.  Although not visible in the picture, there is a sort of faint "X" pattern weakly emitted from the page-magnifier lenses that seemed to be totally absent from the higher-quality acrylic lenses.

Again, it should be noted that in the case of the "fuzziness" of the spot of the vinyl page-magnifier lens, this peripheral energy could not be removed by adjusting focus.  In the case of the images in Figure 8, the spot was projected onto the surface of a 33 foot (10 meter) diameter white satellite dish that was about 200 feet (60 meters) distant.  At this distance, the beams had "mostly" collimated, but were very slightly out of focus as compared to the normal test distance that I'd been using of about 525 feet (160 meters.)  The satellite dish was chosen because it was the only relatively large, flat, white surface that was available at a reasonable distance at the time of testing.  Note:  One can see the subreflector assembly in the bottom of these pictures, along with some lines from the sections of the main reflector as well.


Weight of the enclosure:
  • Without any electronics added, but with the lenses installed, the entire enclosure weighs around 12 pounds, or just under 5.5kg.  Even prior to adding the sides, it was sturdy enough to easily support over 220 pounds (100kg) of weight on its front or back.  (The top/bottom/side panels, being thinner, cannot support such weight.)
Safety concerns when in direct sunlight:
  • One advantage of the "mounting tubes" being used instead of setting the distance of the rear wall at the focus point is that there is that no portion of the enclosure is near the focus of the lens.  What this means is that if there is no optical transmitter or receiver installed, no damage can be done to the enclosure when it is exposed to direct sunlight.  Any light hitting any part of the enclosure is so far out of focus that it doesn't have the concentrated energy to cause more than slight heating.  If properly focused, there is a point - a few centimeters rearwards of the enclosure - where the rays converge and anything placed at that point will instantly burn!  Of course, if the emitter or detector is installed and the enclosure is pointed at the sun, the likely result is the immediate destruction of  one or both of the units!

Note:  Philips has phased out the Luxeon I, III, and V lines in favor of the lower-power Luxeon Rebel and similar devices.  Since I have not used those other devices, the techniques described here may not directly apply.  For the time being, however, the Luxeon III devices are still available from various sources in limited quantities.  Much higher-output devices such as the Luminus Phlatlight (tm) are available and have been used with this enclosure with great success.

Return to the KA7OEI Optical Communications Index page.
Keywords:  Lightbeam communications, light beam, lightbeam, laser beam, modulated light, optical communications, through-the-air optical communications, FSO communications, Free-Space Optical communications, LED communications, laser communications, LED, laser, light-emitting diode, lens, fresnel, fresnel lens, photodiode, photomultiplier, PMT, phototransistor, laser tube, laser diode, high power LED, luxeon, cree, phlatlight, lumileds, modulator, detector
Copyright, KA7OEI 2007-2015, last updated 20150811
Page count since 201005: