Search all of my sites with Google

Thursday, May 8, 2008

Red Light District?

Not exactly....but the red LumiLEDs for the brake and rear signal lights showed up today, so I did a bit of work on installing some of them, as well as adding part of the front turn signals/marker lights. No pics today, just a dorkily-narrated (by me) video of the semi-completed bike lighting in a completely dark room. (Ignore the background voice; someone else kept talking while I was trying to record this).



Keep in mind the camera apparently is trying to auto-contrast or something, so all you can really see are the lights themselves--the room, which to the eye was very brightly lit by the bike, appears very dark--almost no reflection of bike lighting is shown in the camera video. I guess I should try just pointing it at the walls without any bike lights in-frame, and it might capture the effect (I just thought of that now, cuz I'm a dork).

Sunday, May 4, 2008

Let There Be Light

These are the amber LumiLEDs I've wanted for a while; the exact same type you see in modern automobile turn signals and traffic lights. They're VERY bright for their size and power input (70mA @ 2VDC). To know how bright, just go look for cars that have turn signals that switch on and off instantly (LED), rather than fading in and out (incandescent bulbs). That's what these are, and how bright they are.

Courtesy of my sister Kitty, I've got 180 of them to experiment with, which is good, because I can use a lot of them for making my bikes more visible, both in turn signals and side marker lights. What you will see in this post is the first experiment with them to see how well they work in small amounts (4 and 6 LED strips). I expect to have to use 10 of them at a time for the final version, to get good daylight visiblity, but for the moment I'm happy if they work at night (when people really can't see hand-signals very well, assuming they even are looking, and assuming they even know what handsignals are--too many pedestrians, cyclists, motorbikers, and other motorists do not seem to recognize handsignals at all, probably because they're no longer on driving tests, AFAIK (at least here in AZ).

I've got red ones on the way, purchased partly with my birthday money from another sister (Cara). Expected in the next two weeks or so, they'll replace the standard red LEDs you see in the partly-assembled rear lighting unit below:

This was my scanner-light taillight, which I decided was a nice durable housing. I simply pulled out the cold-cathode fluorescent tube and it's driving electronics from the housing (saving them for some future project I haven't imagined yet), and used some translucent silicone caulk to glue them in a line where the bulb used to be, pointing *away* from the light opening, so that they will use the same diffuser/reflector the bulb did, and instead of being bright red dots they'll be a big red rectangle, more visible and easier on the eyes.

They're unfortunately not all the same kind or color of red LEDs, so some are brighter and some are redder or more orange, as they came from various high-brightness lighting items. The 7 LEDs comprising the brake light (one center, 3 each outside end) came from a bike seat with integrated taillight I found crushed in a crosswalk one day (moment of silence...); it's electronics were destroyed but the LEDs were all good, and pretty bright. One of the LEDs is a lot dimmer than the others, which is funny because it was an unopened Radio Shack Super Ultra High Brightness 5000MCD LED from the 80's, which I paid at that time (according to its faded tag) $4.99 for the single LED. Wow. Now, I guess 5000MCD is a lot even now, but compared to the others that are probably 8000 and 10000MCD (1000MCD is roughly one candlepower, or the approximate brightness of a candle flame, simplistically), it seems quite dim. But I needed something that would drop about 2v like the others do to complete a pair of evenly divided serially-wired strings, and this was the only one that did, out of all the reds I had (none of the others are anywhere near as bright, and I think some of them may go back to the 70s, given their style and almost invisible red glow in even normal room lighting. Next to modern ones, I can't even tell they're on!

This shows you the partly-assembled diffusers for the lighting unit. I *was* going to just put the milk-plexiglass back on, but then I realized when I held it up to the 4x40w overhead fluorescents in the kitchen that almost none of the light was coming thru it. What a waste! So I opted for less diffusion and more usable light, and dug thru a drawer of old lids (I *knew* I saved them for something!), and found a few translucent flexible ones, probably from cottage cheese containers. I cut out the biggest rectangles I could, then trimmed them to fit one pair across the main bar opening, and siliconed them in place just as I had the LEDs. To cure a little faster, I put it out in the sunlight, which that day was making the air over 90F, even with the decent winds (had to put a brick on the end of the plastic to keep it from blowing around). Silicone was used because I know of no common solvent or glue that will dissolve or bond to this kind of plastic (HDPE or PE, I'd guess--it isn't marked), and it's easier to remove in case of mistakes or changes in design than epoxies. Less dangerous, and probably less unfriendly to the environment too.

You can see the lid remnants above the diffuser frame. Those will be cut up further for more pieces on the ends, and then in front for the other signals there.


Here's some LEDs raw in a little pile waiting to be glued into strips, like you see on the left. I used the perfboard to hold them lined up, then added tiny drops of superglue between each one and the next, to form strips of 4 for the side markers, and 6 (wired in 2 sets of 3) for the rear-facing turn signals. Once dry, I then painted them from the back with flat white Pactra model paint (which oddly enough was still good after sitting in a storage box for almost 20 years). The white paint makes a huge difference to how much light is captured and reflected back out from the LEDs--without it, all you really get is a round dot of very bright light at the center of each otherwise-dark square LED. With it, you get a round dot of very bright light in the center of a gently-lit square, which makes the whole unit look larger, and more eyecatching (and also less hurtful to the eye, probably due to lower contrast between areas? Maybe overall a bit brighter, making the pupil contract more? I am not sure).


These two pics are in normal room lighting, which is the kitchen in this case, with 4x 40w bare fluorescent tubes directly overhead, in a white room with canary yellow cabinet doors. Plus the camera flash. Even with all that, the LEDs are pretty bright. You can see the current readings on the meter (in mA) for each set. It's higher for the longer strip because it's wired as 2 parallel strips of 3 series-wired LEDs, but it's using the same resistor as the shorter strip, due to the way I am switching them. I had a bit of experimentation to do to get the current reading on the longer strips to be below the 70mA max spec on these LumiLEDs (it's almost 80mA in the pic here); calculations had shown it should be 69mA but the voltage drops aren't exactly 2.00v, they're a bit lower, and I couldn't quite make the exact value I needed out of the resistors I had available (at least, not without a lot of series/paralleling I didn't have the space for).

It's a simple circuit, with a 12k resistor input (from the turn signal generator up front) to the base of a 2N2222. The emitter of the 2N2222 is wired to the top of a pair of 190ohm (originally, it's something else now, but can't remember the value) resistors in parallel, to ground. Those resistors are creating the current-limit for the LEDs, to keep them from being damaged at whatever the highest possible system voltage is. The collector is wired to the cathode of the rear-facing 2x3 strip. The 4 LED series-wired strip's cathode is wired to the emitter as well. Power (12VDC from the lighting system battery) is directly connected to the anodes of both strips.

That means that the 4LED strip is always on, except when the transistor is turned on by the turn signal square wave input (about 1Hz). When the transistor is on, the voltage drop across it (.7v) plus the LEDs in it's collector (about 6v) is less than the drop across the 4LED strip (about 8v), so they don't get enough current to light up anymore, as most of the current is now going thru the 2x3LED strip instead, creating the turn signal. This makes an alternating light between the corner marker and the rear, giving anyone around me in any direction that should matter an indication I'm turning or changing lanes. The same circuit will be used in front.

The resistor values I used are only because I happened to have bags of those values around from a box of stuff someone gave me perhaps 15 years ago or more. I probably have better values on various scrapped boards, but haven't checked them yet, as it works as-is. (though I still need to find the optimal resistors for the current limiter, it's in safe limits with the values I actually used, rather than the 190ohm pair (95ohm actual). Also, the ones I have are all 1/8watt, which helps because they're small, but they get fairly warm in-use. Not hot, just warm. :-)

This is the best pic I could get of my attempt to use one of Maxim's great automotive-lighting chips; this one is intended for use in rear cluster lamps (RCL) like mine to keep the current constant on the LED strings regardless of input voltage, within the 5-36v limits of the chip itself. It can handle 3 separately controlled strings of LEDs, up to 350mA per string. There are other chips Maxim has that do less current, or only one LED, or have other features (including some that have uC interfaces). It uses a simple single-resistor current sense, and the chip itself dissipates the power that's not used in the LEDs. That means it needs to be mounted with it's bottom thermal pad on a thermal or groundplane on a PCB, but since I don't yet have the ability to make SMT PCBs I had to go for some perfboard, drilled out under the pad location, and soldered wire to the pad to spread out as a heatsink underneath, which you can't see here (it's a mess, so you're not missing anything).

The resistors are 1/8w types, so pretty small (just tan blurs in this image--I could not get the camera to focus on it no matter what mode it was in). The real problem is the chip is a 16pin small outline DIP, which means the pins are closer together than even the thickness of the yellowand black wire you se going to the board. I used wire-wrap style wire (very thin) and soldered from the perfboard copper strips to the pins, but it was VERY VERY hard to do, and I know I overheated the chip several times trying to do this. I'm surprised I didn't actually break it just doing this, but I did damage it, because it failed after only a few minutes of testing. Since I didn't have several more hours to try to replace the chip (I did say it was VERY VERY hard to do!), I went back to a circuit I'd designed with just resistors and a transistor to switch the signals on while switching the markers off, alternately.

The Maxim chip did work (before it died from the thermal abuse) to regulate the current (and therefore brightness), all the way down to about 6v (I used my rechargeable AA cells to simulate dropping voltage, removing one from the stack and checking current and guesstimating brightness, as I didn't have time to throw together a variable power source just to test with). At that point, it began getting noticeably dimmer and then went out, because I was testing a 3-LED strip, which is about 6v drop right there. One LED would likely have worked down to minimum chip voltage, but I will be using them in strings, mostly series but some series-parallel.

I'd prefer to use a PWM-type system (especially the buck/boost units) rather than linear, but they all require more components, and they all are more complex and larger to build than these Maxim chips require, which means they are less likely to be successful in my experiments. :-) If I could make SMT PCBs and do SMT work easily, it'd be another thing entirely, but currently I can't. I have yet to verify I can even make a good standard thru-hole type PCB, which has larger tolerances and clearances for pins, traces, and parts. :-) That'll come soon, once I get the magnets for the CFM conversion, and have a motor I actually need the controller for, so I then have to build the controller (which is designed, just not built and tested without the motor for it). :-)



The mostly-assembled rear lighting unit, in normal room lighting, with flash. Kinda tells you how bright the LEDs are that they and their reflected light still show up so well even with the area around them lit with the flash! First up is the side marker, which is always on when the lights are on. It's amber right now, but will be changed to red and covered with a diffuser made of more cottage cheese container lids *or* a bunch of translucent silicone; depends on which one works better in tests for seeing from bigger angles, in different street lighting conditions. Then the amber side strips will be moved to the front lighting unit, in the ends of the headlight bar.


Same thing with the rear-facing turn signals, which will remain amber for distinctiveness (as many, if not all, cars are made with these days). The leftmost pic has the diffuser in place over it, the other two don't. You can see the difference it makes in making it more apparent the light is on.



A couple shots of the assembled unit (without diffusers), with the side marker (left) and rear turn signal (right) lit, total darkness except for the LEDs.



This is the assembly next to the bike (which you can't see) on my carpet (which I wish you couldn't see), with the camera flash plus the room lighting on the left, and in total darkness except for the lighting unit itself on the right. Wierdly, the camera sees the red LEDs (in the long bar area) just about the same color as the amber LEDs on the ends. I guess it must be either light overload, or something in the automatic camera settings. (I'm not yet familiar enough with what few manual settings there are to get good pics using them, so I tend to use the fully automatic knob setting for quick stuff like this).


The bike in the dark with all the lights on (except brake and turn signals). The headlight is so bright it drowns out the camera's response to the reflected light on the walls and such, which to the eye were very well lit in red and amber.


Here you can see the turn signals on, along with the taillight. These are in total darkness except for the bike lights.

This is just a straight-on rear shot with side markers and taillight only, total darkness except for bike lights.


The bike's rear section, oblique, with the camera flash on and off, room is lit with a 15watt fluorescent striplight about 8 feet out of frame toward the front of the bike. The rear lights are just the side markers and taillight.



I didn't get any pics of the brake light in operation yet, because I didn't have anyone handy to pull the brake lever while I took a pic or vice-versa, and the shots I took with the camera timer either didn't go off right, or else the camera is already so overloaded by the taillight that it doesn't register any brighter with the brake light on, too. Or it's auto-compensating for the brightness. Well, the taillight and brakelight LEDs *are* just temporary till the week after next, when the red LumiLEDs should be here to replace them with.