The LED's Challenge to High Pressure Sodium

In my last video, we talked about the highpressure sodium lamp and its ubiquitous use in outdoor lighting.

Already this type of lighting is startingto be phased out with various technologies, with new LED technology among the most common.

Now, given the robust nature of the sodiumlamp and its proven track record in providing an efficient light source reliably, does it make sensethat we switch to newer light sources? Well, as usual in life, there are pros andcons.

So let’s start with a basic question: whydo we use outdoor lighting? For street lighting, specifically, the basicanswer is to improve safety, particularly pedestrian safety.

The odds of a crash of any kind are greaterat night, because it’s harder to see.

With the aid of artificial roadway lighting, a driver can see much farther than their car’s headlights shine, especially to the sides.

Now of course there are disadvantages to largeamounts of street lighting, which we’ll get to, but assuming safety is the goal, aresodium lamps good? Well, no.

Not really.

Remember how I said that the Sodium D-lineis close to our eyes’ peak sensitivity, but only under photopic daylight conditions? Well it turns out that our eyes see quitedifferently at night than during the day.

Under nighttime, scotopic lighting conditions, our eyes actually see bluish light better.

And that makes sense–after all, moonlightand starlight are pretty bluish, so under these dimly lit conditions, having a greatersensitivity to blue light would mean we can see better.

Under scotopic lighting conditions, only therod cells in our eyes are activated.

Rod cells cannot distinguish color, but theyare much more sensitive to light than the cone cells.

The peak sensitivity of the rod cells is around498 nanometers, which is a green-blue color.

Now of course under street lighting our conecells are still active–we can after all see colors and are not exclusively using the rodcells.

This dim-but-not-quite-dark lighting scenariois often called mesopic vision, a mix of the two.

Still, stimulation to the rod cells will befar more visible and is more important.

So then, how well does the light from sodiumvapor lamps line up with our scotopic light sensitivity? Not well.

This is the CIE 1951 scotopic luminosity functiongraph.

The X axis is the wavelength of light in nanometers, and the Y axis is the eye’s relative sensitivity to these wavelengths under scotopic nighttimeconditions.

As you can see, peak sensitivity is aroundthe 500 nanometer mark.

And where’s the wavelength produced by asodium vapor discharge? It’s about here, 589 nanometers.

The rod cells are barely activated by a sodiumdischarge.

While the discharge may be extremely efficientat producing visible light, at night time this light is fundamentally misaligned withour eye’s sensitivity.

See, if you look at the response curves ofscotopic and photopic conditions together, you can see that the sodium discharge linesup great with our photopic vision.

But when our eyes are adjusted to nighttimelighting conditions, it’s actually pretty bad.

This is why focusing on the sodium D-lineemission’s seemingly perfect alignment with our vision is somewhat of a farce.

While it’s true during the day, it’s literallyquite far from the truth at night.

Aside from the simple spectral misalignmentof the sodium lamp, research shows that people can indeed see better under light sourceswith a bluer spectral content.

In Peter Morante’s research for the LightingResearch Center at the Rensselaer Polytechnic Institute (link below), survey respondentsstrongly preferred the light from a 6500k correlated color temperature light sourceover that of high pressure sodium, with metrics of visibility, brightness, safety and security, color rendering, and overall preference all favoring the newer light source, which alsoused only 55% of the energy of the high pressure sodium lamps it replaced.

The study, which by the way is really comprehensiveand worth taking a look at, was done in 2008, which was a tad before LED technology becameeconomically viable.

The study compared induction lighting, whichis sort-of like fluorescent lighting (and worth a video on its own one day because it’spretty neat if not so practical any longer) as well as metal halide lighting to the sodiumlamps for the purposes of creating a recommendation to a local utility company.

The conclusion was that metal halide didn’tmake sense due to higher maintenance costs, but induction would make tons of sense.

But I also liked this final recommendation: Well, it’s been about 10 years time sincethat study, and LEDs are economically viable.

In fact, they’ve become incredibly economicallyviable.

But before we move on to them, let’s goover the main issue once more.

Because of the spectral misalignment of thesodium vapor lamp with our scotopic (and mesopic) light sensitivity, it takes more light outputfrom a sodium lamp to produce the same visible light level from a bluer light source.

In fact, Dr.

Alan Lewis of the New EnglandCollege of Optometry measured individual’s response time to a hazard approaching fromthe sides, and found that in well-lit areas such as a major motorway, a high pressuresodium light system needs to produce 3.

9 times as much light as a metal halide source toachieve the same response time.

The effect is even greater in dimly lit areas, where he found that 7.

8 times as much light from high pressure sodium was required tomatch response time under cooler, metal halide light sources.

So it seems as though the high pressure sodiumlight is less efficient that it appears on paper, and that roadway safety is greatlyincreased when light sources are used that are tuned to our scotopic and mesopic lightsensitivity.

It’s looking pretty bad for high pressuresodium.

But there’s one thing that HPS technologydoesn’t do that bluer light sources might.

That, my friends, is circadian rhythm disruption.

Research has suggested that the color of lightwe are exposed to has a great impact on what time our biological clocks think it is.

The shorter wavelength of daylight sun andblue sky may help keep us awake by suppressing melatonin production, and the long wavelengthlight of the sunset may signal our bodies to sleep by allowing melatonin to seep intoour blood streams.

To help us understand the impact differentlight sources can have on circadian rhythm disruption, light sources are characterizedby their melanopic content.

There’s a great source from the US Departmentof Energy linked down below which goes over this in much greater detail, but as a generaloverview, with high pressure sodium technology normalized at 1 for both scotopic light contentand melanopic light content, a metal halide lamp with a color temperature of about 4, 000Kwill have about 2.

5 to 2.

8 times as much scotopic light content, but also produces 3.

16 to 3.

75times as much melanopic light content.

If you look at the various color temperaturesof LED light sources, you can see that as the color temperature increases, both scotopiclight content AND melanopic light content increase.

However, melanopic light content increasesat a greater rate than scotopic light.

What this means is, there’s a trade-off.

And it’s complicated.

The higher the color temperature, the morescotopic light it produces, which means you could use less light (and thus less energy)to get the same perceived brightness and safety levels.

But, because melanopic light content increasesat a greater rate, although you may need less light of a higher color temperature, it willdisrupt circadian rhythm to a greater extent.

You can see that high pressure sodium hasamong the lowest melanopic light content of any light source, with only amber LEDs andlow pressure sodium producing less melanopic light.

So then, we’re presented with a choice.

We can clearly use less energy and expecta safer nighttime driving experience with light sources of higher color temperatures, but this may cause unwanted side-effects that sodium lighting largely doesn’t.

And this is completely ignoring aestheticpreferences.

I myself generally detest cooler lighting, as I find it harsh and unpleasant.

But I can acknowledge its safety advantages.

In my area on Interstate 88, many of the roadwaylights have been changed from high pressure sodium to LED.

The change is dramatic, with visibility greatlyenhanced once under the cooler light.

As much as I don’t like the color, I cantell it’s a lot safer.

Visibility in my periphery is tremendouslybetter, and I’m certain these new lights are using less energy than those they replaced.

Before I move into the conclusion of thisvideo, let’s discuss the issue of light pollution.

Light pollution is exactly what it soundslike–excess light in our environment that is irritating, unnecessary, poorly distributed, or in general unwanted.

Outdoor lighting is by far the most prolificsource of light pollution, and it has gotten so bad that people living anywhere near acity can barely see the night sky.

There’s even an anecdote about a widespreadblackout in Los Angeles causing many panicked 911 calls from lifelong residents who hadnever seen the Milky Way before and were a little scared of it.

Light pollution is a complicated problem, but the LED may actually help to solve it.

Now to be clear, the solutions I’m aboutto offer aren’t exclusive to LEDs, but the way light is emitted from an LED chip makescontrolling it relatively easy.

So first, let’s discuss one of the biggestcauses of light pollution; lights that point up.

This is less obvious than it seems, but it’sincredibly important.

Right outside my apartment are drop-lens cobraluminaries containing high pressure sodium lights.

Because the lens protrudes downward from thefixture, a lot of light escapes to the sides and indeed upwards.

I’m on the 4th floor of my building, andmy eye-level is above these street lights, but I can still see the source of the light.

This is not ideal for a number of reasons.

First, a lot of light is being wasted by lightingup things that are not the road.

That’s kinda dumb.

But secondly, a lot of this light is goingup into the sky.

Granted, this style of fixture isn’t theworst offender, but a better design would be a flat lens that does not allow light toescape above the horizontal.

This may still throw light farther to thesides than necessary, but none of it will end up lighting the sky.

The worst offenders for this kind of lightpollution are lights that illuminate buildings by shining upwards, wall-pack lights withoutshielding, and these decorative fixtures.

I’ll admit they’re pretty, but they’rereally wasteful.

Recently I was on an airplane flying intoChicago at night.

I took some video as we landed, and you cansee the difference between a well-managed light and a poorly managed light.

This roadway is lit well.

I cannot see the actual light sources, I canonly see the reflected light from the road.

That’s what we want.

As we got closer to the airport, these neighborhoodshad tons of lights that were visible from above.

Much of the light produced by these lampsis shining into the sky and being wasted.

I should not be able to see the actual lightsource from an airplane, yet I can.

This is contributing to skyglow.

Skyglow is what makes the night sky hard tosee when you’re close to a city.

This is probably the most widespread lightpollution problem, and while it’s not caused exclusively by poorly designed fixtures, theyare a major component.

But once again, the solutions to skyglow arecomplicated.

Largely because light sources that cause theleast skyglow are high and low pressure sodium.

In fact, low pressure sodium is used widelyaround large astronomical observatories because their nearly monochromatic light output caneasily be filtered out, eliminating any skyglow they create.

I saw a large number of people saying thathigh pressure sodium can also be filtered out, but I don’t think that’s true dueto the pressure broadening and their spikier output.

Someone correct me if I’m wrong but I couldonly find references to low pressure sodium being used around observatories.

Anyway, where it gets tricky is that an LEDlight source has about three times as much sky glow impact than a high pressure sodiumlight.

But also, you need less of it, so perhapsthe sky glow impact is similar.

In addition, the sky glow impact of incandescentlighting is barely higher than low pressure sodium.

So it could be that LED street lighting correlatedto a 2700K color temperature causes less sky glow than high pressure sodium.

But I think further research needs to be donethere.

In any case, what makes LEDs potentially muchbetter at reducing skyglow is the optical systems that can be combined with them.

Early LED fixtures may have used a large numberof small 1 watt LEDs and tiny lenses to direct their light.

Some really bad designs may have simply hadan exposed chip, prodiving little directional control.

You still see this a lot in cheap flood lightfixtures.

But newer fixtures like these from Cree willuse large chip-on-board emitters, like these 10W chips but larger, and because they onlyemit light in one direction, it’s very easy to control their output with a lens.

You can see on the spec sheet that there are5 lenses in total, though larger fixtures have more.

Most importantly, the optic system iscustomizable.

Depending on fixture height and spacing, youmay need a wider spread of light or a shorter one.

Due to the customizable optics, you can getwonderfully consistent lighting on a road surface such as this area here.

This may also help to prevent light pollutionbecause less overall light is needed.

The hotspots of light you see from above heremean some areas get too much light, and others get too little.

A more scotopic light source with better andmore consistent control may not only reduce light pollution, but may use less energy andprovide safer driving.

Now many of these developments are rathernew, especially the newer optic designs.

But the advantages of LED lighting becomeconfused when drop-in replacement bulbs are used.

I don’t have any major issues about goingthis route–after all replacing the entire fixture can be costly, and some drop-in designsare fairly good.

But the optic system of a sodium, mercuryvapor, or metal halide fixture is designed to reflect and project light emanating froma tiny arc tube, which an LED drop-in can’t recreate.

Many of the complaints regarding poor lightdistribution in LED replacement lamps may simply be from the use of these replacements.

Then there’s the issue with existing ballasts.

Some drop-in lamps claim to work with existingballasts, but I have a few misgivings regarding how well their power supplies deal with thevoltage the ballast provides–particularly if the high voltage ignitor sends some crazyvoltage spikes to the LED drop-in.

I’m sure they can be designed to cope, butit still worries me somewhat.

So then, we have a series of complicated choicesto make.

Sodium vapor lights are pretty efficient, have only a moderate contribution to sky glow, cause only minor circadian rhythm disruptionif any, and have a proven track record of reliability.

But their color also makes them far less effectiveat improving safety, and due to the misalignment of their output with our scotopic light sensitivity, they require more light (and thus use a lot more energy) than a whiter light source.

Perhaps less an issue but still importantis that they contain both mercury and elemental sodium, meaning their disposal is far moredangerous and complicated.

If we were to switch to a white LED sourcewith a color temperature of about 5, 000 K, nighttime visibility would be greatly increased.

Studies have shown that people see hazardsfar sooner under this light, and as the bluer wavelengths match our scotopic and mesopiccolor sensitivity more closely, we can use less of it while also achieving a greatersafety benefit.

This reduces the need for energy.

However, this bluer light contributes moreto circadian rhythm disruption and skyglow, but some of that is mitigated by the loweroutput required by this light source.

Still, many people may not find the aestheticsof this light source pleasing.

One possible compromise would be to use awarmer color temperature LED light source.

Data from the US Department of Energy tellsus that a 3000K LED light would produce 1.

89 to 2.

39 times as much scotopic light as ahigh pressure sodium lamp, while increasing melanopic content between 2.

1 and 2.

99 times.

Because of its greater scotopic output, a3000K LED replacement should only need to produce about half as much light as a highpressure sodium lamp.

This effectively cuts the circadian rhythmdisruption potential in half, too, placing it near about the same as high pressuresodium.

The greatest downside to using a warm colortemperature LED is in their efficiency.

The efficiency of these lights is very similarto that of an average high pressure sodium lamp.

At 67 lumens per watt, these 3, 000K LEDs arejust slightly less efficient than the sodium lamp featured in my last video.

Although you would need only about half asmuch light output, there are HPS lamps which approach 150 lumens per watt.

This would mean that a 3000K LED will usejust about as much energy as a very efficient HPS lamp.

I’d still call that good, but it makes replacementless compelling.

To normalize the effects of scotopic lightcontent, I’ve multiplied the lumens per watt number by the scotopic light contentfor the following light sources.

As you can see, the normalized efficiencyof the LED goes up considerably as the color temperature does, due to both luminous efficiencyand greater scotopic content.

This is likely why most LED street lamp installationsare done with the blueish 5700K and higher color temperatures.

You can use the least amount of energy toproduce the same amount of perceived brightness and safety.

However, the normalized efficiency of eventhe 3000K LED is very similar to that of high pressure sodium, and few HPS lamps actuallyoutput 150 lumens per watt.

Also, the calculated lumens per watt of theLED is based on the input power of the fixture, so the losses in the ballast (which are fairlyhigh for high pressure sodium) aren’t accounted for here.

To conclude, although the high pressure sodiumlight is very efficient, its primary output color is misaligned with our nighttime visibility.

Only about a quarter of its light is actuallyeffective at stimulating the cells in our eyes.

Although the cool color temperature of manyLED replacements is harsh and aesthetically displeasing, studies have shown that it isnot only more efficient but also makes driving at night safer.

There is however the potential for greatercircadian rhythm disruption and larger amounts of skyglow using these bluer light sources.

Still, it seems clear that the high pressuresodium lamp is on its way out.

Advancements in LED technology are happeningat a breakneck pace.

Just 10 years ago they weren't seen as viable.

But today, even the least efficient of LEDreplacements ends up meeting the efficiency of high pressure sodium when scotopic lightoutput is considered.

As it stands in 2018, we are faced with achoice of efficiency over aesthetics.

I’m pretty sure I’d enjoy roadways litwith the relatively warm 3000K LEDs, and these also wouldn’t disrupt sleep much.

But you can save a lot more energy (and potentiallyhave safer roadways) with 5700K lighting.

Either way, it seems clear that LED technologywill very soon overtake the tried-and-true high pressure sodium lamp, just as the HPSlamp itself replaced the mercury vapor lamp.

And in 40 or 50 years, who knows what technologymight light our roadways.

So I have a couple of things to close out, first you may have noticed in my chart that the mercury vapor lamp had a normalized efficiencyof over 100 lumens per watt, and the 50 watt sodium lamp in the last video was only 78lumens per watt.

Mercury vapor bulbs do have considerable operatingdisadvantages compared to HPS, most notably their steady decrease in light output as theyage, but I think it is somewhat humorous that our current understanding of the visual systemsuggests that sodium light may have been a step backwards in some situations.

You may have noticed that I didn’t talkabout the blue light from LEDs and how this is supposedly ruining our eyes–that’s becausethe “science” behind this is questionable at best.

You can clearly see in this chart that thereis blue-light content in nearly all light sources, and lower color temperature LEDshave less blue-light content than their higher color temperature varieties.

I don’t doubt that blue light can disruptcircadian rhythm–that much seems certain.

But considering that our eyes can withstandthe intensity of sunlight, which is far far greater than any normal artificial light sourceand also has a lot of blue light (and ultraviolet which definitely IS harmful), I think theblue light thing is just fear mongering.

If someone can point to some verified, peer-reviewedresearch supporting this, and not a dodgy website, I’ll consider changing my stance.

In any case, the high flexibility of LED technologymeans that it can be tuned in pretty much any way you like.

I also want to give a shoutout to VWestlifefor the suggestion of LED fixtures with both high and low color temperatures that willswitch to the warm light later in the night.

I think that’s a great idea, though obviouslyit would add expense to any fixture.

However, I was surprised to learn that theCree LED fixtures I’ve been using as a reference are all capable of dimming, and they havea 0 to 10V control input to enable this.

Reducing light levels to perhaps 50% of normalafter midnight might become a common practice, and I think that would be pretty wise.

Maybe this will get combined with technologysimilar to Philip’s warm-glow and we’ll get incandescent-like lighting at night.

For those worried about light pollution forastronomical observatories, there are amber LED street lights available designed to replacelow pressure sodium lights.

This is also great news for wildlife–manyanimals cannot see the wavelength of light produced by low pressure sodium, so this lightsource is used where lights may be disruptive.

One particular example is near beaches wheresea turtles lay their eggs.

After they hatch, baby sea turtles followmoonlight to the ocean, and street lighting was confusing the poor things and they weretravelling inland.

Since they cannot see the wavelength of alow pressure sodium light or its amber LED equivalent, they aren’t confused and successfullymake it to the ocean where they belong.

As a last little tid-bit, the spec sheet fromCree says that their LED cobra head replacements should produce at least 95% of their originallight output after 100, 000 hours.

Assuming the driver and heat sink are robustenough, these fixtures should last well beyond 20 years.

That is impressive.

Thanks for watching, I hope you enjoyed thevideo! If you haven’t subscribed to TechnologyConnections yet, and you like my nerdy deep-dives into whatever floats through my head, whatare you waiting for? Hit the button! Please? As always, thanks to everyone who supportsthis channel on Patreon! You are all making a big difference and it’ssuper awesome of you.

If you’re thinking of joining these awesomepeople and becoming a patron yourself, why not check out my Patreon page? Thanks for consideration.

And I’ll see you next time!.

Leave a Reply