A two-terminal solid-state Headlight Modulator
Written around 1994
Disclaimer (don't let this bother you, I just feel that I need to warn some people)
This disclaimer may seem a little harsh. I don't want to discourage anyone, but some people have tried to hold me legally liable for their tickets received when using a headlight modulator. One person had used a commercially made one, and then tried to get me involved. I don't know what happened with that, I never heard back.
I designed and built this headlight modulator after participating in a thread where some so-called "electrical engineers" proclaimed that a solid state headlight modulator *had* to have more than two leads exiting the package. I won't bore you with their incorrect reasoning, but suffice to say that I designed this circuit to demonstrate that it was possible.
Due to the way this project was conceived, I was primarily interested in function, not federal and state laws. I do not claim to be conversant in vehicle law. Or interested, for that matter.
There is a very real chance that if you build and use this device, you will be violating the the law. Two points that have been discussed are the required amount of 'on' time and the possible need for a photo cell (or some other mechanism) that automatically switches the light from modulated to full on when it gets dark out.
The first requirement, if it is true, is to my mind there to satisfy small minded bureaucrats. The second one is required (IMHNSHO) due to the prevailing feeling in this country that people can't think for themselves. In this case, that they are dumb enough to ride around at night with a flashing headlight.
Now, I've seen the existence of these laws debated on several motorcycle mailing lists and rec.moto. I can't remember any conclusive evidence being supplied either way, but I'm fairly sure the second law really does exists. For reasons described in the next paragraph, I have no interest in researching the matter, or debating it.
Despite my thoughts about the stupidity of these possible laws, if I was manufacturing these devices, I'd have to learn and follow all applicable laws. But I'm not manufacturing headlight modulators. Any home business is too much of a legal morass these days. Vehicle accessories are probably one of the worst. So don't even ask if I'll make 20 of these for your club. I know you'll pay, yes I like money. But I don't want the legal hassles involved.
What it boils down to is that I'm offering these plans for those who are interested. You may use them as you see fit, taking any and all responsibility for the results.
OK, here's the actual post about the modulator.
It took me a while, but I got around to building the two lead headlight-modulator we talked about last month. Here is the circuit I came up with. It only has two leads, and is inserted in the ground lead of the headlight bulb. Terminal 'A' goes to the bulb, terminal 'B' goes to chassis ground.
Naturally you could put this in the positive lead of a lamp as well. Just remember that you'll likely only make either the high or the low beam blink that way. Which might be what you want.
C1 1000 uF 20V Aluminum Electrolytic
C2 .1 uF Ceramic Disk
C3 .1 uF Mylar (Ceramic Disk should be OK)
C4 .1 uF Ceramic Disk
C5 .1 uF Ceramic Disk
C6 2 uF 20V Tantalum Slug (Aluminum Electrolytic should be OK)
C7 .1 uF Ceramic Disk
R1 680 Ohm 1/4 W
R2 100 Ohm 1/4 W
R3 2.2 K Ohm 1/4 W
R4 22 K Ohm 1/4 W
R5 4.7 K Ohm 1/4 W
R6 47 K Ohm 1/4 W
S1 SPST 10 A Switch (to disable blink)
U1 555 Timer IC
U2 555 Timer IC
Q2 interrupts the power to the bulb by breaking the ground return. Q2 is used in switching-mode. It is either on or off, nothing in between. This minimizes heating. On my breadboarded prototype, Q2's heatsink is a 2 by 3 inch piece of aluminum. After running for several hours (with a 45 W halogen bulb), it is only slightly warm. Q1 has no heatsink, and is running at ambient temperature.
Power for the circuit is supplied by D1 and C1. When the light is blinking, chopped DC is present on the collector of Q2. This is rectified, filtered, and supplied to the rest of the circuit. It's far from 'clean', but 555 timers aren't too picky.
When oscillating, U1 provides base drive to Q1 at about 300 Hz. The duty cycle is near 50%, this provides the dim portion of the blink cycle.
U2 oscillates at the visible blink rate. This is about 5Hz, once again at about a 50% duty cycle. When pin 3 (the output) of U2 is low, diode D2 conducts, keeping pin 6 of U1 low, which inhibits oscillation and leaves pin 3 (the output) high. In this state, Q1 and Q2 are turned on and the bulb is lit at (very near) full intensity. When pin 3 of U2 is high, D2 is reverse biased and U1 may operate normally. In this condition, the light is pulsed on and off 300 times a second and appears to be dimmer. Since U2 runs at a visible rate, this make the bulb appear to blink between dim and bright.
When we were first talking about this, there was some question as to whether blinking a halogen light would cool the gases inside it to the point where the halogen cycle would not occur, thereby clouding the inside of the glass with a tungsten film, dimming the bulb, and shortening the bulb life. An engineer at Delco who reads this list asked one of our lighting engineers about this. He said not to worry, it would probably be hot enough, and if it wasn't, then it would be too cool for much tungsten to 'boil off' (he'd probably gag at my terminology). He also said that a halogen bulb would be an excellent choice for a modulated headlight, since the constant re-deposition of tungsten on the filament would prevent the change in crystalline structure that leads to filament failure when a bulb is turned on and off. Apparently this is a far bigger factor in bulb life than actual 'thermal shock' that we all assume to be the culprit.
Still, it is my opinion that one test is worth a thousand expert opinions, so I intend to run the modulator on a halogen bulb whenever I'm out in my workshop for the next month or so. If there are no deposits on the bulb by Christmas, I'll assume that no damage is being done. If there are deposits, I'll get a new bulb of the same type and run it on steady DC for the same period of time and see what happens. Not rigorous, but probably good enough.
The bulb already has about 9 hours on it and looks fine.
BTW, I'm doing these tests on a no-name (Industrial Supply Company of New England "Made in Korea") automotive halogen replacement bulb. It is marked "H9004" and "HB1". It is a 65/45 W bulb, and I'm using the 45W filament so that the bulb is as cool as possible.
I've been very careful not to touch the glass, just as the package warns, but I just noticed that there is now a fingerprint etched into the glass. I guess the people at the bulb factory aren't as careful as they would like us to be. I hope the envelope doesn't fail during my test.
Well, I put several hundred hours on the bulb before I needed the space for other things. There was no noticeable darkening of the envelope, and the fingerprints caused no problem
Frequently asked questions
A note from years after most of this was written: Wow, I must have been bitter when I wrote these... They make some good points though, so I'll leave them. I had probably just been flamed when I wrote them though, so keep that in mind.
Here are some commonly asked questions. I get a lot of them on this project. Most of them are pretty cluefull, but a few have not been. :-)
1. Why didn't you use a MOSFET for the output stage?
Reason one: I didn't have one handy. I had a pile of 2N3055's. Since I was building this to prove a point, I wasn't interested in the most efficient possible circuit. I just wanted something that worked, had two and only two leads, and cost me as little as possible to build.
Reason two: I don't need to, Rik Steenwinkel has already done so. His circuit is also available on my web page. It looks good, but I haven't actually tried it. He seems pretty sharp, so I suspect it will work fine.
Reason three: It's not really clear that a suitable MOSFET exists (in the mid '90s, I don't know about the 2010s), let alone one that is cost effective (not that that would matter for a one-off). I haven't done circuit design professionally for about ten years (I've been doing programming mostly), so I'm not totally up to date on what's available, but I asked some of the engineers and designers around my place of employment about suitable MOSFETs. They felt that at the current levels a headlight requires, most current power MOSFETs would have a higher on voltage than a bipolar switching transistor, thereby defeating the whole purpose. One said, "Hell, go with the bipolar, it's a mature technology". Another said, "Someone's been reading NASA tech briefs, eh?". Another suggested using an IGBT (Insulated Gate Bipolar Transistor). This is really two transistors in one package, a MOSFET for low drive requirements, and a bipolar for output. So we're back to a bipolar transistor for the output. It's a mature technology. :-)
Several people have built headlight modulators with MOSFETs and told me about it. At least one reported needing a lot more heat sink than the '3055 needs. This led him to doubt that my findings are accurate, since everyone knows that MOSFETs are more efficient. Jeesh... BTW, most of the MOSFET headlight modulator builders are quite clueful, as are *most* of the people who've written me about this circuit. But since I get perhaps two notes a week on this, I get some that aren't. Thanks for all of the interest. Despite the grumpy tone here, I'm really pleased that so many people are interested.
Reason four: The 2N3055 works just fine. Despite dire warnings to the contrary, it is pretty efficient. It's always on or off, and with its small heatsink (2x3 inch piece of aluminum sheet), it's only slightly warm. What more do you want?
2. Why didn't you calculate the power dissipation of the output driver and then perform elaborate thermal calculations justifying the size, geometry and surface finish of the heat sink?
Reason: Pretty much "see above". Heat dissipation is not a major problem. I didn't expect it to be, and since I *do* have a bit of design experience (quite a bit of practical hands-on experience with bipolar switching transistors in power circuits), I wasn't really surprised. Basically, I went with empirical methods, and they worked fine. You'd be surprised how many things are designed that way. If it was going to be produced commercially, I would want to validate all of the assumptions mathematically, but heh, this is a hobby project, you know?
3. Why didn't you make this change to ease manufacturing or lower unit cost?
Hey, I only built one! The only thing I remember buying for the project was the halogen bulb I used for testing. How much more cost effective do you want? If I was going to manufacture them, I'd probably make some changes.
4. Hey, don't you know that lots of people have installed tree-scorching headlights that will bring this circuits to its knees in a cloud of acrid smoke? What were you thinking?
Reason: Once again people, "I designed this to prove a point, not as a product for sale to the public". It doesn't have to work in every bike, stock or modified. You're on your own building and using this. I don't mean to say that I'm not interested in hearing about a version you built, but I am tired of being flamed for my effort in designing this. I don't mind trying to help you troubleshoot a version either, but my experience shows that troubleshooting via email is usually not too effective. I'll try though.
5. Why don't you redesign the circuit to incorporate this suggestion?
Reason: I'm not a design service. I do electronics projects for fun and only do things that seem enjoyable at the time. If you make it seem interesting enough, maybe I'd try it. But probably not. I'm swamped in project ideas and really short on time. It would probably be better for you to try it yourself. If you really feel that you can do it better, why not do so? (No doubt that many people have had good ideas for improvements).