For a printer-friendly PDF version of this article, please click here.
What is Selective-Yellow Light?
It's what happens when you subtract blue from the output of a lamp producing white light. But first, what means "white light"? Under US Federal Motor Vehicle Safety Standard number 108 and Canadian Motor Vehicle Standards 108 and 108.1, headlamps as originally installed on motor vehicles (and as installed by anyone other than the vehicle owner) must produce white light. The relevant SAE (and identical ECE) color standards define "white" light as a rather large range within the CIE 1931 colorspace. That's why both brownish sealed beams and bluish HID headlamps are considered "white". It's also why "blue ion" or "crystal blue" bulbs with blue-pass dichroic filters sold to poseurs who want to try to pretend they have HIDs are not considered "white". The light can tend towards a yellow tint to a certain degree and still qualify as acceptable "white" light.
In 1936, the French for tactical reasons wanted a way to identify the registration nationality of vehicles at night. However, they did not want to reduce roadway safety, and wanted in fact to improve it if possible. So, they figured to remove the blue from the output spectrum of their vehicles' front lamps. Some technical papers out of France on the subject can be had here and here. White light with the blue component subtracted is known as "selective yellow" light. It is a pure yellow color with little or no orange component—hence the French yellow headlamps. Yellow lamps have consistently over the years been subjectively ranked as better in poor weather and lower in glare than white ones, but is the effect real? Or is it just a subjective impression?
One problem with this conclusion as drawn from the French experience with selective-yellow headlamps in France is that when the question was being considered, the lamps that were being compared with white lamps reduced the absolute intensity of the beam by about 12 percent. This fact may have had a part in reducing the glare. Because the requirement for yellow light no longer exists (though it remains allowed in many countries) we probably will never know the vagaries of the answer to this question.
What explains the persistent subjective preference amongst experienced poor-weather drivers for yellow fog lamps, despite decades of white fog lamp prevalence? Selective yellow light can improve a driver's ability to see in fog or rain or snow, but not because it 'penetrates fog better' or 'reflects less off droplets' as is commonly thought. That effect is known as Rayleigh Scattering, and is why the sky appears blue. However, it occurs only when the droplet size is equal or smaller than the wavelength of the light, which is certainly not the case with ordinary fog, rain or snow. Roadway Fog droplets are several orders of magnitude larger than visible light wavelengths, so there's no Rayleigh Scattering.
So, why do yellow fog lamps seem to work better? It's because of the way the human eye interacts with different colors of light. Blue and violet are very difficult for the human optical system to process correctly. They are the shortest visible wavelengths and tend to focus in front of our eyes' retinae, rather than upon it. To demonstrate this to yourself, find a dark blue store front sign or something else that's a dark, pure blue against a dark background in the absence of white light—from any appreciable distance, it's almost impossible for your eyes to see the blue lighted object as a sharply defined form;the edges blur significantly. Deep blue runway lights exhibit the same effect; check it out the next time you land at night.
Blue also is a very difficult color of light to look at; it stimulates the reaction we call glare. Within the range of allowable white light, bluer headlamps have been shown to be 46% more glaring than yellower ones for a given intensity of light — see studies here and here. So, it seems culling the blue out of the spectrum lightens the optical workload and reduces glare. For a more detailed examination of this effect with respect to driving in foul weather, see Bullough & Rea's study on the topic.
So, what's the best method of getting selective-yellow light? Until the mid 1990s, headlamps in France were required to produce yellow light. This was accomplished in one of several ways: With a headlamp lens made out of yellow glass, with a yellow glass balloon in front of the bulb either as part of the bulb or as part of the lamp unit, or, later, with a yellow-pass dichroic filter coating on a lamp's lens, reflector, condensor or on halogen bulbs themselves.
Cadmium glass was used to make the old French-market Selective Yellow bulbs; now that Cadmium's been more or less banned from auto parts for environmental reasons, the best remaining option is an absorption (non-dichroic) filters applied to one of the optical elements—the lens or reflector.
The blue-appearing lenses in many Asian-made fog lamps ("ion crystal", "gold irridium", and other such whimsical marketing names) are coated with a multilayer dichroic interference coating which passes selective-yellow light on axis, i.e., straight ahead. However, these dichroic filters don't absorb/block the blue light, they simply diffract it so it leaves the lamp off axis. So these lamps tend to glow blue when viewed off-axis, and in extreme cases there can be objectionable blue haze outside the brightest areas of the beam. The irridescence of these coatings causes or aggravates secondary-reflection problems where none would exist absent the coating. With the mirrorlike dichroic coating reflecting images of the glowing filament, light goes where it doesn't belong.
There are generally no more yellow bulbs available in quality worth buying. The market for them in Europe pretty much dried up once France stopped requiring yellow light; they're now regarded in Europe primarily as a retro or styling item, which no longer justifies first-line manufacture, so most of the yellow bulbs now on the market are of iffy quality (though some of them are hideously expensive!). Selective-yellow HID headlight bulbs (D2R, D2S) have been marked by Philips in Japan and other parts of Asia for quite a few years—Philips part numbers are 85122YX and 85126YX—these are now showing up in Europe as well.
If you want selective yellow lights for whatever reason, applying a coating to an optical element is a more permanent, optically cleaner option that eliminates the need to find and get special bulbs. Good results have been obtained by removing the lamps, cleaning the lenses thoroughly and making sure they're warm, then spraying them with several wet-but-not-drippy coats of Dupli-Color Metalcast yellow or the equivalent product from Krylon. This is a transparent yellow paint product of the correct hue, with good adhesion and durability. Let each coat "flash off" (dry most of the way) before applying the next, and use thin coats so you don't get drips and sags in the wet paint. With each successive coat, the yellow tint will grow deeper. Make it about 2 shades deeper than you think looks right, and it'll turn out well in the end. Of course, the coating needs to be permitted to dry and harden completely before you take the fog lamps out on the road, otherwise dust and grit will become embedded in the still-tacky surface. In the case of lamps with removable lenses, by coating the interior surface of the lens obviously answers questions of coating durability against pitting and scratching. Results of conversion can be seen here.
What about light loss due to filtration?
There is no substantial loss of light from a properly-applied absorption-type selective yellow filter placed in the optical system (reflector, bulb, lens, or intermediary filter element) of a lamp, because the blue being filtered out is only a very small part of the bulb's total output. The visible spectrum consists of all the colors of the rainbow: Red, orange, yellow, green, blue, and indigo + violet. Glowing filaments produce a whole lot of light in the red-orange-yellow-green wavelengths, and only very little light in the blue-violet wavelengths. To put very rough numbers on the matter, a 9006 bulb produces 1000 lumens, of which approximately 250 are red, 250 are orange, 250 are yellow, 175 are green, 50 are blue and 25 are violet.
Now, suppose we want to add a filter to the glass that makes the light look bluer/colder. How does it do that? Well, there's no such thing as a filter that adds light into the beam passing through it; filters can only suppress light, not add it. So if we can't add green-blue-violet light, then the only way to get the light to look colder is to suppress green-blue-violet's opposites, which are red-orange-yellow. If we want the light to look, let's say, 20% colder, we suppress red-orange-yellow by 20%. Looking up above, we see that we've got a total of 750 lumens' worth of red, orange and yellow. So, cutting this by 20% leaves 600 lumens, plus essentially all of the bulb's original green-blue-violet output of 250 lumens, so we've now got a bulb that produces light that looks 20% colder and produces 850 lumens.
850 lumens happens to be the minimum legal output for a 9006. Unless we're craven marketeers who don't care about compliance or performance, we can't produce a bulb that produces only the bare minimum of light, because 50% of production will be 849 lumens or less. So, we have to put in a high-luminance filament to try to counteract some of the filtering losses. BUT we still have to come in under the max-allowable-wattage spec in DOT or ECE regulations.
So, let's say we build our 9006 with a super-duper filament that produces 1200 lumens. That's too much for a 9006, but we're going to take away some of those lumens with our colored filter (blue glass). This 1200-lumen filament produces, let's say, 300 lumens red, 300 lumens orange, 300 lumens yellow, 210 lumens green, 60 lumens blue and 30 lumens violet. Now we put that same blue glass over it, which suppresses red-orange-yellow by 20%. Now we've got 720 lumens' worth of red-orange-yellow after filtration, plus 300 lumens' worth of green-blue-violet. That gives us a 910-lumen bulb, which is enough above the 850-lumen legal "floor" that we can run the bulb and even if some filaments only produce 1150 lumens instead of 1200, we're still legally OK. Of course, we still only have 910 lumens instead of 1000, and our 1200-lumen filament is going to have a significantly shorter life than a 1000-lumen filament, but we've got our colder/bluer light appearance in a legal bulb.
I bet by now you see why filtering for yellow does not significantly reduce light output: Take our 1000-lumen 9006 as broken down by colour output above. No such thing as a filter that adds extra yellow light, so we have to get our yellow by suppressing blue-violet (the particular yellow in question, selective yellow, contains all the green found in white light. If we took out green, we'd have a turn signal type of amber-orange light.) OK, then, let's cut blue-violet by 80%. That means we've got our 925 lumens' worth of red-orange-yellow-green, plus 15 lumens' worth of blue-violet (after filtration). Total: 940 lumens. MUCH smaller loss! OK, so we put in a very slightly better filament, say one that produces 1060 lumens, and now we've got 980 lumens' worth of red-orange-yellow-green, plus 16 lumens' worth of blue-violet (after filtration) for a total of 996 lumens, which is for all intents and purposes identical to our original 1000-lumen uncoloured bulb—for context, the dimmest allowable parking lamp bulb produces 30 lumens.
Daniel Stern Lighting (Daniel J. Stern, Proprietor)
Copyright ©2009 Daniel J. Stern. Latest revisions 10/12. No part of this text may be reproduced without express permission of author. Permission to quote is granted for the purposes of communication with the author or general discussion.