| Lynx MR16 |
 |
 |
 |
 While the heart of the Lynx DMX SSR4 is the microprocessor, the Microchip 18F1220, the heart of the Lynx MR16 is the Texas Instruments TLC5940 – despite the fact that the MR16 also has a 18F1220.The reason is that the MR16 controls 16 channels of DMX, whereas the SSR4 control four channels – the extra horsepower for the 12 additional channels has to come from somewhere. Designer Robert Jordan (“RJ” of DIY Light Animations) moved most of the control of the intended illumination devices – light-emitting diodes, or LEDs – off of the 18F1220 and onto the TLC5940, a 16-channel, constant-current pulse-width modulation (PWM) LED driver. This means that the 18F1220 is used mostly for data communications and DMX translation.  The TLC5940, in turn, handles the on-and-off of the LEDs in addition to the dimming (PWM is a fancy way of saying “dimming”). In order for the MR16 to handle a large number of LEDs, though, the TLC5940 doesn’t directly control the lamps; it in turn drives 16 MOSFETs, which are transistors. This n-type device is rated at 100 volts and 15 amps, so there’s plenty of room for current (RJ says each channel is rated at 2½ amps per channel).RJ designed the Lynx MR16 to control – and this is where the nomenclature gets confusing – MR16 LED lamps. Originally designed as halogen multifaceted reflector lamps, with the advent of highly efficient LEDs, some manufacturers have replaced the single-element halogen bulb with multiple LEDs in the same form-factor and connector pin as the original lamp (of course, using far less current). The Lynx MR16 requires a lot of good-quality direct current; though a number of light masters buy purpose-dedicated power supplies, a few of us have modified a used computer power supply for the project. I actually found a new ATX supply on Craigslist for $10 in the fall of 2009 and wired it up similar to what many have done as a bench supply (though I leave out the secondary switch, add a fuse to the 12-volt lines and use a four-terminal barrier strip as a connector rather than the binding posts). I mounted the power supply and the MR16 board onto a piece of plywood and then mounted both to the garage wall.  While RJ doesn’t like his devices to be used beyond their intended design, he has bent the rules with the Lynx MR16, as he has favorably mentioned their use as controllers of various devices, including water solenoids.Here at PacificaLights, we’re using the Lynx MR16 to control LEDs, just not bundled in MR16 form-factors. We have built seven frames to go onto the front-porch bannisters and the windows, which rings them with red, green and blue LEDs. Why not just use strings of commercial lights? Why put the lights on a frame and why build your own circuits? I had expended a great deal of energy in 2008 trying to install commercial strings of mini-lights neat and straight: I purchased a commercial product that was supposed to make ringing doors and windows with straight lights a breeze; it didn’t work with the brand of mini-lights I’d bought in the after-holiday sales of 2007. I then tried to staple commercial mini-lights to a wood frame but had trouble – both with keeping the lights straight and not destroying the strings while installing them. This desire for straight lights was coupled with further 2008 failures: one half of one second-story window was out for most of the light season because while modern mini-lights are supposed to fail one bulb at a time, they can also take out either half or a whole string. I ended up restringing the bannister midway through the season to compensate for another dead string. Commercial strings use series circuitry and series can take out multiple lights. There’s another issue with using commercial strings: they rarely fit around your bannister or your window or your door accurately. A light master usually has a few too many bulbs (or a few too few), leading to the shortening of strings or “blacking out” of bulbs. Yeach.  These frames turned out to be a much more extensive construction project than originally envisioned. I had assumed (uh-oh) that I would get bulk LEDs that had the same type of two-inch leads on them that one gets when purchasing smaller quantities of such bulbs. When the Hong Kong reseller’s bags-o-bulbs arrived here in October 2009, it turned out that these came with quarter-inch leads. Which in turn meant that each one of the thousands of LEDs I planned to install would have to have more wire soldered onto each lead.And that in turn became a much more time-consuming process than I originally had thought: the first ones took me as much as 10 minutes per LED to cut the wire, strip the wire, twist it onto the lead and solder it. I elected to use 22-gauge copper wire; I had overbought Cat5 cable in 2008 and had a spare 500 feet of it lying around; further, by assuming that orange insulation indicated a red LED, I was able to color-code the LEDs with the individual Cat5 pairs (I interspersed the browns so that every LED had at least one “true” color). By the time I was doing my 300th LED, I had gotten that down to a little more than three minutes per (I was able to do 36 LEDs in under two hours per session).  Matters just got worse: while building the frame itself wasn’t too bad (about an hour to measure out the 1x3 lumber, mitre-cut the corners and deck-screw them together), laying out the spots for each LED and drilling the holes (two 1/16-inch holes per lamp) ended up being about twice that length of time (then the frame needed to be sanded and painted – flat white all around and gloss white on the front; add another couple of hours).But the real pain came in wiring the back of the frame: I used a parallel-series type circuit, so the LEDs were clustered together in series-circuits of three lamps each; those in turn required a resistor to be installed into the mix (while I don’t like series circuits, these are limited to three bulbs each, so only three bulbs might die at a time, much better than a commercial string taking out 30-50 lamps). The first few frames I wired loosely, just hand-twisting pairs together. I then went back, soldered those wire-twists and dipped them in silicon sealant. This left a lot of loose wires on the back of the frames and while I didn’t change the ones I’d built, I developed a new method by my third frame: I drive little brads into the wood, wrap a length of one wire around the brad and then a length of the connecting wire around the same brad. It’s like the old wire-wrap process used to build circuit boards in the 1950s, 1960s and 1970s. It was slow, tedious handwork (not my specialty) and while theoretically I should have been able to pound out all the wiring in eight hours for a larger frame, it really ended up taking closer to a dozen hours (and spread out over many days as I just couldn’t sit and do it for longer than 45 minutes or an hour at a time). My little calculator says that the first bannister frame took 22½ hours to build. Hence, only one bannister frame was displayed in 2009 and it didn’t get installed until after Christmas. My plan had been to build one frame a month in 2010, meaning I would be done sometime in the summer; like all other aspects of this particular project, that didn’t happen either. I was still wiring frames in early December but the frames were finally installed by Dec. 5. Grand totals: There are 1488 LEDs on the windows and bannisters – 496 red, green and blue – grouped into 21 different DMX channels. There were at least 3000 solder joints and a similar number of connection points. By my estimate, it took about 170 man-hours to construct all the frames. Then the bad news: I had noticed midway through the frame construction that the green LEDs were failing at a higher rate than the other colors but I didn't take a precautions to prevent there from being a wholesale green LED crisis. By the second night of testing the installed frames in 2010, it was clear that there had been many, many bad green LEDs, so there are many bald green patches on the frames. There are also some blue bald patches – nowhere near as many as the green – and there are a couple of smaller circuits that died and took out greens, blues and reds. There was a further problem: I bought the brightest LEDs I could find and the result was that watching the show actually hurt your eyes (nevermind the across-the-street neighbors who were blinded by the light). During 2011 I came up with a plan to place covers over all the LEDs that face the street – I used plastic caps that were hot-glued onto the LEDs. Additionally, I burned in about 100 green LEDs for about 100 hours each so that I could get them to fail. I spent another dozen hours or so in November 2011 replacing dead green LEDs and hot-gluing on plastic caps. Such are the joys of a Christmas light master. There’s one last interesting story regarding my display: the electronic sign. |