Why Most LED Headlight Upgrades Don’t Really Work: An Expert Explains

Plug-and-play LED replacements for halogen headlight bulbs are a popular car mod. LEDs often look brighter than incandescent lights, but “looks brighter” and “illuminates better” are not the same thing. I got a stern talking-to from an actual lighting expert about LED retrofits, and science says: Putting LEDs where halogen bulbs are supposed to be is generally not actually an upgrade.

After I blogged about different shapes of LED “bulbs” making some better than others, lighting expert Daniel Stern was kind enough to open a dialogue with me and answer some important questions about automotive lighting.

I’ll introduce him and pepper his wisdom into this call-and-response style explainer, laying out what the deal is with LED headlight retrofits in 2020.

What exactly do you mean by LED headlights?

LED tech has advanced beyond this 2015 model shown, but in many cases the housings still look like this. (Screenshot: Jalopnik)

In this blog, we’re talking about the kind of aftermarket LED (Light Emitting Diode) lights that fit directly into a car’s stock light housing — housings that were originally designed to work with incandescent halogen bulbs. These are sometimes called “LED replacements” or “LED bulbs.”

We’re not talking about cars that come with LEDs from the factory or self-contained LED lights like what you might use as fog lights or for off-roading. HID (High-Intensity Discharge) lights will not be discussed here, either.

Hasn’t this technology come a long way?

LED lighting’s been around for decades, and it’s crept into the automotive aftermarket in waves. Circa 2005 is the first time I remember seeing LED “bulb replacements” in auto parts stores, as in, the now-common kind made to fit in the same socket as standard filament bulbs for brake lights and turn signals.

They were primitive back then. All the diodes pointed in one direction, and light output was plainly poor. I remember buying one at the Route 128 AutoZone that I haunted in high school for my 50cc Honda Express scooter’s taillight, installing it, and immediately thinking it sucked.

Fast forward to 2015, at which point the accessibility of LED tech had expanded further. By then, several companies were selling LED headlight bulb replacements that you could direct-swap from halogen bulbs in a bunch of standard sizes. I tried some in a third-gen Toyota 4Runner and concluded that while the LED’d headlight was brighter to look at from the outside, a high-performance halogen bulb still threw better light down the road. (In retrospect, I recognise my testing methodology was ratchet, but I stand by the takeaway.)

Now it’s 2020 and there are countless companies selling LED bulb replacements. Bluechip name brands like Philips and Sylvania are in the mix. There are boutique high-end outfits like Diode Dynamics specialising exclusively in aftermarket LEDs. And of course, eBay and Amazon are chockers with cheapo off-brand options.

Unfortunately, for the most part, none of them really work correctly with few exceptions.

Why would anyone want LED headlights?

LEDs, when housed and aimed correctly, can translate minimal input power into a lot of light, which is what makes the technology appealing in general.

Everything else being equal, it seems like swapping power-hungry incandescent headlight bulbs for brighter lower-draw LEDs would be an upgrade on two fronts. Plus the “instant-on” effect and visual crispness of light that comes from LEDs is sharp and fresh. LEDs can give older cars modern styling.

In the simplest terms: LED headlights are easily installed and readily available things that can make cars look cooler. So, people get them.

So what’s wrong with that?

Most car headlights are a lot more than just a bulb in a socket. A cradle of reflectors is shaped and angled so that light emitted from the filament of an incandescent bulb will be thrown down the road in a way that maximises driver visibility without blinding oncoming traffic.

Most LEDs don’t emit light from the same space in the headlight housing as incandescents and from that point, they’re doomed to mediocrity.

What about LEDs that do emit from the right spot?

A few of the better-known companies selling and testing LED headlight replacements in 2020 have taken pains to mimic the positioning of incandescent bulbs to address this problem but it’s only a small part of the lighting equation.

In fact, I did a blog about this, pointing out why some LEDs seemed better than others when used as headlight retrofits. And that blog is why lighting expert Daniel Stern felt compelled to reach out to me explain how grossly underinformed I was.

Who’s Daniel Stern and why should I care what he thinks?

Stern is a professional lighting consultant with significant work and published papers on the subject under his belt. In his words, straight from his CV:

“I actively participate in technical standards development and research bodies, evaluate and critique relevant regulatory proposals, and have contributed materially to vehicle lighting regulations in several countries and territories. I participate in―and report on―the major international automotive lighting symposia and conferences, and have attended meetings of the United Nations international vehicle lighting and light signalling regulatory development working group at the invitation of its president.”

He’s also done extensive studying on LED headlight kits and retrofits, like the ones we’ve discussed on this site a few times. You can get his very detailed perspective on the matter from his own site, but since he specifically addressed some common questions about LED retrofits in his emails to me–politely I might add–I’d be remiss if I didn’t share those insights with you.

(That’s what you’re reading in this blog, in case you hadn’t worked that out.)

Why not just use an LED that is exactly the same size and shape as a regular bulb?

As Stern explains: “Longitudinal position of the light source (where the light source starts and ends, as measured from the base plane of the bulb) is only one critical aspect.” But it’s not the only thing that matters. “Others include shape, size, orientation, and luminance distribution. Getting one out of five right is better than zero out of five, but it’s still 20 per cent, a badly failing grade.”

“If we could wave a magic wand and come up with a cylindrical LED emitter of the same dimensions as a filament, with the necessary luminance and flux, then the incompatibility would vanish. That is not technically possible for the foreseeable future, so we have basically two-dimensional flat LEDs in place of a three-dimensional cylindrical filament.”

“There is significant space between the two back-to-back flat LEDs (there has to be, otherwise no material to carry away their heat), so now our light source is radically different from a filament in shape, size, position, and light distribution even if we’ve taken great care to put the emitters at exactly the same longitudinal position as the original filaments.”

And why does that matter again?

The problem is that light reflectors designed for halogen bulbs are inherently incompatible with the light output of LEDs.

Stern wrote: “…the near- and far-field light distribution is quite different to what the lamp’s optics were engineered for.” And as a result, the headlight’s beam pattern isn’t what it’s supposed to be, doesn’t line up with the way the vehicle’s engineered, and is all-around suboptimal.

Why do my lights have to keep the factory beam pattern?

Stern’s analogy: “I wear eyeglasses, and so does my next-door neighbour. It would be injurious and counterproductive for us to swap because even if they fit my face and look awesome, the optics don’t match my eyes (even if I think I can see OK with them).”

“And it’s not because I picked the neighbour to the left instead of the neighbour to the right. The same holds true for both neighbours’ glasses even though the one pair has glass lenses and the others are plastic, the one set has round lenses and the others are rectangular, the one set is photochromic and the other isn’t, the one neighbour is more farsighted than I am and the other is more nearsighted, etc.”

“The details are different, but the basic problem is still optical incompatibility, and the scale of the relevant differences is much smaller than ‘these lenses look the same to me!’”

To say it again, slightly more sciencey: LEDs in housings designed for halogen bulb replacements put the wrong amount of light in the wrong places.

But I have seen comparison tests where LEDs looked like a clear upgrade.

“Looking at beam patterns is not an accurate or adequate way of assessing them. Looking at pictures or videos of beam patterns is markedly worse,” Stern explained to me.

I showed him Diode Dynamics’ goniophotometer test, and he agreed that this clip does a good job explaining why this tool is important in overcoming visual tricks to understanding how well a light is actually working:

I still don’t understand how a goniophotometer works exactly, but essentially it ingests data from a light source at every possible angle to analyse its performance. Without such a device, it’s impossible to really know what kind of light quality you’re looking at.

“Low beam cutoff presence, sharpness, and shape are not go/no-go gauges for whether or not a beam pattern is good, and even if they were, they’re impossible for even a legitimate expert to accurately assess by looking at pics and videos,” Stern wrote me.

Why can’t pictures or videos capture headlights accurately?

Stern: “If we’re talking about pics and vids, they’re very misleading even if the photographer has the best of intent. Pixels and film have a much narrower dynamic range than the human visual system, so even a photographer who carefully keeps all the camera settings identical when photographing different beams cannot present more than broadly general comparative information: a sharper cutoff versus a fuzzier one, for example, or a wider beam versus a narrower one.”

But even if you have comparable lights in front of your face, you might not be looking at a better light just because one “looks brighter.”

So I can’t even trust my eyes? Why?

As Stern explains: “Light and seeing work in counterintuitive ways. For one example… glare and seeing light is not 1:1. It takes a whole lot more additional light than intuition suggests to begin to get an improvement in seeing, and it takes a whole lot less additional light than intuition suggests to cause a significant increase in glare.”

“So if we shine a headlamp at a wall and we notice there’s a little more light above horizontal, it seems like just a little more light… but in fact, we can very easily have doubled, tripled, or quadrupled the glare and the upward stray light which causes backdazzle in rain/fog/snow.”

Beyond blinding yourself, increasing glare is also a factor in why some LED-upgraded cars could blind oncoming drivers too, even if so-upgraded cars seemed to have their lights aimed correctly.

I don’t buy it. I know what I can see.

I was confused by Stern’s assessment that human eyes are a bad judge of what makes a good headlight. Like, if you’re looking at better-looking light, isn’t that… the better light?

As Stern elaborated: “One big difficulty is that what we feel like we’re seeing isn’t what we’re actually seeing. The human visual system is a lousy judge of how well it’s doing. ‘I know what I can see,’ seems reasonable, but it doesn’t square up with reality because we humans are just not well equipped to accurately evaluate how well or poorly we can see (or how well or poorly a headlamp works). Our subjective impressions tend to be very far out of line with objective, real measurements of how well we can (or can’t) see.”

Then how can I ‘see’ that some lights are brighter than others?

“The primary factor that drives subjective ratings of headlamps is foreground light, that is light on the road surface close to the vehicle… which is almost irrelevant; it barely even makes it onto the bottom of the list of factors that determine a headlamp’s actual safety performance.”

“A moderate amount of foreground light is necessary so we can use our peripheral vision to keep track of the lane lines and keep our focus up the road where it should be, but too much foreground light works against us in two ways: it draws our gaze downward even if we consciously try to keep looking far ahead, and the bright pool of light causes our pupils to constrict, which severely degrades our distance vision. All of this while creating the feeling that we’ve got ‘good’ (or ‘much better’) lights. It’s not because we’re lying to ourselves or fooling ourselves or anything like that, it’s because our visual systems just don’t work the way it feels like they do.”

So do LED bulb replacements ever work?

Not all incandescent-type halogen bulbs are equal, and as we’ve touched on, there’s quite a variety of LED bulb replacements on the market now too.

The problem here, Stern told me, is “a lack of predictability.”

More specifically: “…Occasionally it’s possible to happen upon a combination of one of these ‘LED bulbs’ and a particular headlamp that works acceptably.” (For example, DiodeDynamics has a particular ‘LED H11’ Stern recalled being successful in a particular Ford truck headlamp housing.)

“But there’s no predictability to it; it’s nothing at all like ‘oh, as long as you have projectors you’re fine,’ or ‘reflectors are fine if they have a bulb shield,’[a “bulb shield” is a piece that blocks unwanted light coming from certain angles of a bulb] or anything like that. And ‘Oh, no problem, on our bulb the LEDs can be rotated relative to the base so you can focus them’ is, if anything, worse, not better.”

The whole point of standardisation of bulb types, Stern explains, is “so any headlamp designed and built to take [for example] an H11 will work safely with any bulb designed and built to the H11 specifications. That doesn’t mean all H11s are alike… but the standardisation ensures at least adequate safety.”

“And that really is the way it has to be, because think of a bulb that fits in any H11 headlamp but works safely in only a few of them.” That’d be pretty crap.

So what really makes a good headlight?

Stern: “Headlight beam safety performance involves multiple interdependent variables. For example, an amount of foreground light that might be just fine with a strong, well-focused hot spot, will limit the driver’s seeing to a totally inadequate 50 or 18.29 m if the beam’s hot spot is weak or nonexistent. So just saying ‘Yep, the cutoff looks good’ doesn’t even begin to be close to good enough.”

“What matters much more is the amount and distribution of light under the cutoff, and that gets pretty scrambled (randomised) with most ‘LED bulbs’ in most halogen headlamps. To give just one of many examples of how easy it is to get tripped up on this point: Sometimes you’ll get a reasonably sharp cutoff with an “LED bulb”, but the hot spot (assuming it still exists) is moved.”

“Every last little bit of downward and/or rightward movement of the hot spot decreases the seeing distance for the driver, but the beam on the wall looks like it has a nice cutoff and hot spot.”

“Another example: say we’re dealing with a headlamp that didn’t start out with a very sharp cutoff. Put in an ‘LED bulb’ and the hot spot moves upward/leftward. The typical advice: ‘Re-aim the lamps to keep the bright part out of other drivers’ eyes.’ But by doing that, we’ve shifted the entire beam pattern so a bunch of other stuff no longer has appropriate amounts of light on it.”

“What’s the scale of these effects on seeing distance with different effective lamp aim (whether it’s by dint of how the lamp is adjusted or how the lamp is distributing its light)? Well, if you’re using the shine-on-a-wall method, aiming a low beam just 2.3 cm (0.9 inches) lower than it should be cuts 26 metres off your seeing distance at night!”

Is there anywhere I can see this for myself?

If you’ve read this far, you might be realising what Stern spelled out to me: Most LED reviews on the internet are not helpful, nor are the criteria they often cite (sharp cutoffs, colour temperature.)

As far as amateur, layperson field testing of LED headlights, Stern pointed to this thread on the Tacoma World forum as a pretty good real-world explanation of why “thoughtfully selected” halogen bulbs beat even name-brand LED retrofits.

“That’s an amateur instrumented test set up to be as favourable as possible to the LED bulb… It’s a major brand product, not some $US20 ($29) no-name trinket, and it’s tried in a projector lamp, which enforces a sharp cutoff with just about any light source crammed in,” Stern wrote me.

What about LED upgrades that change out the whole housing?

Older cars and Jeeps with sealed-beam lights might, somewhat ironically, be in a better position when it comes to LED headlight retrofits in the current technological ecosystem.

If you’re replacing the bulb and housing, I asked Stern, wouldn’t you be able to get a good LED retrofit with today’s tech? His response:

“The concept is correct — this is the right way to do it, an LED headlamp engineered, designed, constructed, tested, and certified/approved as such. There are excellent ones on the market, and of course also a great deal of junk. The king daddy of them all [at the time of publication] is the JW Speaker 8700 Evolution-J3, which brings almost Star Trek-level technology to the old seven-inch round headlamp format.”

Stern also shouted out the “JWS 8700 Evo 2” as a good option one step down, and “another step or two down from there, [is] the Peterson Manufacturing 701C (in Peterson or Sylvania Zevo packaging — same lamp) and the Truck-Lite units are reasonably good with or without heated lens.”

JW Speaker was cited as the optimal choice right now for rectangular sealed-beam retrofits too and “Truck-Lite also makes respectable lamps in this size.”

Any other important features on those lights?

Stern made a good point I’d heard before: Since LEDs don’t heat headlight lenses the way halogen bulbs do, you might want a light with a heated lens to melt snow.

“Wise to get heated lenses if one does a lot of wintertime driving in heavy wet snow and slush; without a lens heater, the LED headlamp lenses run cold so slush can build up on them and freeze/occlude the lamp instead of melting off like they do from a warm halogen or BiXenon lamp lens. Short of that, though, there’s little reason to fret about it; less severe winter conditions with lower-volume, drier snow won’t make problems as the cold snowflakes will just glance off the cold lenses.”

“The situation with small round lamps is different and more difficult. Here yet again, the JW Speaker entries are terrific with or without integral LED parking light, daytime running light, and turn signal — all functions safety-approved; this is not the greasy kid stuff one finds all over the motorcycle/hot rod/chrome shop sites. Expensive, though, and most headlamp mount cups would need their central holes enlarged to clear the relatively large rear body of these lamps. No entry in this size from any of the other major makers, because of relatively minimal demand. However, I am now testing some rather promising ones very nicely made in Korea. I hope they pan out; they’d be an easier fit in most mount cups.”

“Sometimes one runs into electrical incompatibilities in trying to install LED sealed beams. Toyota-made vehicles, for example, have very unusually-configured headlamp circuits, some of which won’t play nicely with LEDs. There are workarounds that don’t involve hacking the vehicle.”

Will LED Headlights Ever Be A Worthy Upgrade For More Cars In The Future?

“There are technical working groups worldwide (SAE in America, GTB in Europe/Asia) actively working to develop a technical standard for LED retrofit bulbs to replace halogen bulbs in headlamps, fog lamps, and other such,” Stern told me, as member of such groups himself.

As for the current state of technology, Stern says: “Unlike ‘HID kits’ where there is no possibility of optical compatibility, that possibility does theoretically exist with LEDs. The products presently on the market are not close to acceptable; there are still some very substantial technical hurdles to overcome… but eventually, there will be legitimate products of this kind.”

“It’s hard to wait (believe me, I know!) but the ones on the market now just don’t cut it, no matter whose name is on the box and what promises and claims are made.”

PSA: No Matter What Bulbs You Run, Aim Your Headlights!

Mr. Stern had a parting thought he asked me to pass along. When you do find a good set of bulbs for your car, don’t forget to make sure they’re aimed correctly.

“…headlamp and fog lamp aim is by far the main thing that determines how well you can (or can’t) see at night with any given set of lamps, so no matter what lamps are in use they must be aimed carefully…” If you’re not sure how to do that, check out Stern’s own guide.