In the process of retrofitting lighting into new and existing designs, I come across particularly problematic situations that no products on the market seem to exist. The most recent was a single head pedant I needed a light source with a light pattern somewhere between an old R20 spot and a PAR20 Flood, at around the R20 flood light output. The fixture was designed around the R20 and its soft beam.
I tried a couple commercial R20 LED products, and found them all to be too harsh and splashy, and too narrow in distribution. So, I figured, how hard can this be? I was impatient and tired of buying stuff that don’t work…
Igor! gather me up some parts!
With a heat sink from an older project salvaged from a spare parts bin, a PAR20 cadaver, a tortured to its death screw-base CFL lamp, a fresh 4W 120VAC Acriche LED and my precious collection of epoxies (glue is what holds the world together) – I headed to the lab. The surgery started with cutting the PAR lamp apart with a glass-cutting band saw to separate its optic head from the body. Next was to cut the socket off the plastic CFL ballast compartment, leaving its remains on the floor. After soldered the leads onto the LED, it was glued it to the heat sink with thermal epoxy, then joined to the PAR 20 optic with standard 5 minute epoxy. Wires lead through the heat sink into the socket, grafted to the internal lead wires left from the CFL separation. I filled the socket with epoxy putty to stabilize it and keep the nervous system in its place, then pressed it onto the back of the heat sink. It was done… and not a drop of blood was spilled.
The whole project took less time than it takes to watch an episode of Top Gear on the tube. The end result actually surprised me. The beam pattern works like I needed, its smooth and soft at the edges, with a strong center punch. Overall light output in the fixture is better than the 50WR20, even though total light output is closer to a 25W R20. In use, the light on the table below the fixture is actually greater (21Fc at the center and 4 at the edges, vs. 11 at the center and 5 at the edges with the R20), and the heat is a lot less (50W down to 4W). For those who can’t sleep without the math, that’s a 92% saving in energy with no loss of usable light. Of course there is no UL label on this thing, and I have no idea what has happened to efficacy – don’t really care about either… the thing works (it’s ALIVE!) As a side note – Compared to a 3W LED MR16 I have from a very well known LED manufacturer, this gadget produces more light and a nicer beam pattern – and cost me millions of dollars less to develop. Igor, it’s time for a brew!
Concerned over heat sink effectiveness, I checked the temperature at the Ts point (a pad next to the LED on the PCB its mounted to) which is specified as max 100C, with 70C recommended. After two hours, the Ts temperature stabilized at 128F, or 54C, well under the 70C recommended by SSC. So far so good.
Not bad for a lunch break project.
Recommendation rating = 5 of 10. If you have the time and interest, experimenting with LEDs can be a satisfying experience. For others making ones own lamps is not a very a practical solution.
Actual product performance against manufacturer claims = 9 of 10. I’m convinced the product does 90% of what I wanted of it (maybe a little softer edged for this application.)