Comments on “MORE LUX: light bars for SAD”


Full Bridge Rectifiers

J's picture

I see some reasonable-looking luxmeters on amazon for <$20, so you might consider getting one, for science! Also, I'm a big fan of full bridge rectifiers, but high power linear power supplies require great big transformers, and 300W 24V transformers tend to be more expensive than the switching supplies you bought. (And you're right you'd need a huge rectifier as well as a huge smoothing cap). A better option for people working on the cheap might be used PC power supplies, which often have 12V rails in the tens of amps.

Light hacks

That's a great find. I got inspired by your earlier post and got parts for a 100W (actual power) LED box, along the lines I had suggested, using current drive LED strips. But your solution is both cheap, nicely packaged, cost effective, and doesn't use high voltage.

You do have to take care that most LED lights are advertised as "incandescent equivalent watts", rather than actual physical LED watts. That's almost universal for AC fixtures or replacement lamps. Given the discussion below of "LED watts" marketing term below, my guess is is that your actually input power is probably 1/2 that or less. Note that they are also quoting "LED lumens", which I would assume would mean the light output if driven at the theoretical max. You can pretty easily find the actual input power by measuring the DC current and voltage. Is each bar really drawing 12.5A at 24V?

What is the difference between "LED Watts" and "True Watts"?

LED diodes are rated based on the amount of power (wattage) they can theoretically handle, if they are perfectly cooled. Excessive heat causes LEDs to degrade ("burn out") and makes them shift the color of light they are giving off, so in the real world LEDs are never run at their full rated wattage. This means that "LED Watts" is a completely useless number for comparing the light output from two different lights-- for example, you can have two "500 LED Watt" lights, with one running 100 watts of actual power through the LEDs, and the other running 300 watts of actual power.

Ignorant when it comes to electronics

I failed my graduate oral qualifying exam because the examining committee asked me about capacitors. I had, obviously, only the vaguest idea of what one does.

At the next department meeting, the faculty as a whole un-failed me (and, the rumor mill said, informally failed the committee) because, as a CS student, I wasn't supposed to know what a capacitor does. I'm not quite sure why the committee didn't know that, and so asked me about capacitors instead of the Schönhage–Strassen algorithm.

@J —Thanks, I've added a link to your comment from the post!

@Rob — " Is each bar really drawing 12.5A at 24V?" Hmm, I don't know. The fixtures seem to have internal voltage regulation (since they are designed to run on 12VDC but will take up to 30). They do claim 30k lm; if that is true, then they'd have to actually draw about 300W. (LED luminous efficacy is ~100lm/W.)

Very nice. I was looking into

Very nice. I was looking into LED shop lights for similar reasons and to save floor space (I thought it would be nice if I could get rid of my old broken torch lamp with the giant CFLs), and also saw the car lights, but the electrical wiring killed them for me. Big LED tubes + shop lights might be a better way for people not confident in their electrician skills.

500W light bar

Adam's picture

I found this:
It was $55.59 when I bought it, $60 now. Says it's 500W 50,000 lumen.
I also got a 12v 500W power supply for $35 off amazon. I'm going to hang it directly over my computer desk. I hope the power supply is high enough, I'll be running it at max. Did your 300w supplies drawing 300w all the time have any issues?

500W FTW!

Nice find! The only 500W ones I could find on Amazon were quite a bit more expensive.

The power supplies I got are fine. I was afraid they might get hot, but the internal fans are more than adequate; they're not more than a little warm.

The light bars themselves have massive heat sinks, but still get quite hot. Borderline of too hot touch—OK briefly but would get painful after a few seconds.

Actual power draw

Adam's picture

So, I was reading all the comments from the first blog entry. I saw the one about the LED strip not actually drawing all the wattage it says it should and the reply link to

There it suggests that since it's for a car, it might expect the 12v voltage to actually be ~14v like a battery would provide. I was wondering if that might be the case with these as well. If you have a multi-meter handy, could you check if your 300W lights are actually drawing 300W off the power supply?

Actual power draw

My $12 multimeter only goes to 125 mA. I'm not sure how to measure serious DC wattage; if anyone does know, and it's inexpensive, I will try, because I'm curious too.

The LED bars I got apparently have an internal voltage regulator / power supply, because they will accept anything between 10V and 30V. LED strips, by contrast, are just bare LEDs wired together in clear plastic.

So, my best guess is that this issue with LED strips does not apply to the bars—but I could be entirely wrong.


Adam's picture

Just googled it. What's needed is a shunt.
It's put inline on the live wire and scales down the high amperage to the mA range. As long as what it steps it down to is lower than what your meter can handle, it'll work. There are 50A -> 75mA shunts on Amazon for ~$6
I'd probably have had to figure that anyway when I went to measure mine once I get it, good to know! 50A multimeters are $$

Using a shunt

Thank you! I will probably do this, eventually... I'm unusually busy at the moment, so it may not be for a few weeks.


David: yes, a light bar with a wide-range input like yours definitely has an internal switching power supply to convert the input to constant output current. r.e. using an external shunt, it's a resistor, so you get volts out for current in. 75 mV @ 50 A means it's a 1.5 e-3 Ohm resistor.

Disclaimer: for some reason I find thermal design interesting, and I tend to end up designing things that run cooler than they really need to.

If the external heat sink is getting that hot, then the LEDs are probably running hot enough that they aren't going to last 50,000 hours. This is a reasonable tradeoff for the original intended use. No matter how cool it seems, I doubt that many people who buy these are going to run them all that much without really pissing people off. Also, on a vehicle at night it's likely to be cooler than indoors and there will be some forced air cooling due to vehicle motion. If you were to aim even a pretty small fan in the direction of the lights, it could make a big difference in the operating temperature.

BTW, for high-power LEDs, end-of-life is conventionally defined as output less than 70% of rating. As well as long term output degradation, higher operating temp also produces a temporary output reduction, which is typically 10-20% as long as the rated max operating temp is not exceeded. Higher temp also tends to shift the color temperature in the blue direction.

In the LED strip datasheets I was looking at, light output was quoted at a hotspot temp of 50C, and life at 70C. To touch, that's the difference between "hot" and "bloody hot". Depends on your finger, but here's one calibration:

I vaguely remember an engineer mentioning an old rule of thumb
for quickly estimating temperatures on circuit boards: you put
your finger on a component, and it went something like this:

Feels cool ... 25 to 35 deg C
Neutral ... 37
Warm ... 40 to 50
Very warm ... 50 to 60
Hot ... 70
Makes a blister... 100 and up

Does anybody have the numbers that go in here?
I added some above. For most folks, 70 is about as hot as one can stand for a second or two (does depend on calluses, though).

There's no way of knowing how much hotter the LEDs are than the heat sink, but we can be confident they are hotter.

r.e. StackExchange comments about "12V" strips, most of those comments seem on point too. Drawing less than rated power at 12.0V makes sense from both marketing and engineering perspectives. On the marketing side, there seems to be a lot of spec creep, and you have to compete. On the engineering side, there is the issue of 13.4V vehicle operation and just generally providing a margin to improve reliability (derating). Most customers aren't going to notice if it draws less than rated power, but are going to notice if it blows out or runs so hot that it smells like burning plastic.

r.e. power supplies. If you buy enclosed switching power supplies off of EBAY then you'll probably end up with a power supply that is rated reasonably conservatively, and should run fine at 100% of rated output. I've been pretty happy with EBAY as a source for cut-rate Chinese electronics; the next notch up might be Jameco, which stocks "Mean Well", a low-cost brand with a decent reputation.


If the external heat sink is getting that hot, then the LEDs are probably running hot enough that they aren't going to last 50,000 hours. This is a reasonable tradeoff for the original intended use. No matter how cool it seems, I doubt that many people who buy these are going to run them all that much without really pissing people off. Also, on a vehicle at night it's likely to be cooler than indoors and there will be some forced air cooling due to vehicle motion.

That makes sense! Does seem like they may have limited lifetime, then.

It doesn't say it's switching or universally regulating or anything.

I'm reasonably sure it is, and not to worry. (But, I don't know anything about electronics. Rob would know better.)

Power supplies

Almost all power supplies are switching these days. Universal input just means you can operate on 120 and 240 VAC. As long as you are in the US, the one you got is fine.

I referred to David's bar as having wide-range input because it was spec'd to work over 12-24V or an even wider range. That would be unreasonable for LEDs without some sort of internal current regulation. And at these power levels, a switching converter is the only sane option.

Outdoor floodlights

Thanks! Yes, outdoor floodlight of this general sort are a plausible alternative. The best lumens/$$ ratio I was able to find a couple months ago was this one—which is actually a pair of 30,000 lumen floods. You can buy more than two at a time, but not only one, of that particular model, for some reason.

Before Rob's most recent message, I would have said that the bars were a better bet, in terms of cost and ease of use, but if those are going to have a lifetime problem, the floods may be the way to go. Presumably they are designed to last, since they're meant for lighting parking lots and such.

So, got my light all wired up

Adam's picture

So, got my light all wired up, still need to hang it once I find a ladder.

Using the lux meter on my phone I measured 100,000 lux from ~12in away, and 10,000+ lux at 5ft! So I'm pretty happy with it so far :D

Reliability, lifetime, etc.

It's really the wild west with LED lighting right now, both because things are moving so fast, and also because far east suppliers are eagerly selling whatever seems to sell, without worrying much about the long term. For me, for light therapy, it isn't a disaster if it fades after a year or two.

The one concern that I do have with the flood lights or light bars is in how you use them for light therapy. With the light box, you are supposed to frequently glance directly at the light. I'm not sure that is such a good idea for extremely bright sources, but if they are bright enough, you may not need to do that. Yes, sun light is quite bright, but you don't stare at the sun. If you have a 100k lux flood light, that's your sun. What you might want to do is make sure you have some white surfaces in your work area that get brightly lit.

From wikipedia Daylight:
120,000 lux Brightest sunlight
111,000 lux Bright sunlight
20,000 lux Shade illuminated by entire clear blue sky, midday
1,000 - 2,000 lux Typical overcast day, midday
<200 lux Extreme of darkest storm clouds, midday
400 lux Sunrise or sunset on a clear day (ambient illumination).
40 lux Fully overcast, sunset/sunrise
<1 lux Extreme of darkest storm clouds, sunset/rise

Reflective diffusing surface

Note also that diffuse reflection is still directional. You'll get the most reflected light if you are viewing the surface at right angles vs. a more oblique view. You should be able to get a sense of this with the phone lux meter.


If you wanted to vary output lumens, what would be the easiest way to put a dimmer switch on the lightbar?


Most LED lighting systems are not dimmable, unfortunately. You can assume they are not unless they say they are. These bars don't say they are, so they almost certainly aren't.

If you get several, you can turn some off.


Installed one of the 500watt light bars that Adam linked to. It's not as bright as I would have thought, but I have it currently aimed up at the ceiling.

Also this power supply is NOISY. I'm going to have to replace it with a computer psu with larger/quieter fan, or I'll have to work wearing earplugs all the time.

Silent power!

I found an virtually silent power supply - (at least up to 500w)
1) Buy an HP Platinum 1200W platinum efficiency power supply - such as the DPS-1200FB-1 A. You know it's platinum because it has the blue plug socket. Example -
Then buy an adapter to use it like (You could DYI, but I'd rather just buy an adapter).

Virtually silent at 500watt draw! (This could power up to 900watt at 120ac, 1200w at 240, but no idea how quiet that might be) It's also really small. 10x6x2, including the adapter board.

Worrying info about retina damage

Alex's picture

Heads-up: there's a study from the NIH linking LED lighting to macular degeneration. The experiment used "cold" (>4500K) and bright (>= 6,000 lux) light, meaning most people probably don't need to worry and get rid of all their LED lights, but this seems like one application where it would be a cause for concern.

Macular degeneration

Thank you for the heads-up! Definitely a matter of possible concern.

There's extensive discussion of this paper on Hacker News. (Thanks to Daniel Klein for pointing this out to me.) Here are some thoughts... bear in mind that I'm utterly non-expert on any of the relevant science.

There's weak evidence that blue light is key for SAD treatment; and weak evidence that blue light causes macular degeneration (in general, not just from LEDs). So this might be awful, in the long run.

The evidence for blue light causing macular degeneration in humans seems to be sketchy:

The main observation from HN is that even the brightest SAD lights (such as the ones I've built) are less bright than winter sunlight, and have pretty close to the same spectrum. That suggests that exposure to white SAD lights is probably no worse than going outdoors on a sunny day in winter. (Eye doctors do recommend that people with early-stage macular degeneration wear blue-blocking sunglasses when going outdoors. However, the evidence for this seems to be doubtful currently. UV is definitely a risk, but in humans blue light is not so clear. The clearest risk factors for macular degeneration are actually the same as for the "metabolic syndrome" diseases such as diabetes. Losing weight is good for you.)

The "daylight" LED spectrum is not exactly the same as sunlight, and it's possible that the difference is critical. But this seems relatively unlikely to me.

Another observation is that this study was done in rats, which tend to avoid sunlight, and whose eyes, some other studies suggest, are much more prone to light damage. Further, they found that LEDs were particularly bad for albino rats (whose eyes are abnormal anyway), and that fluorescents were worse for non-albino lab rats. Humans are probably more like non-albino rats, so the upshot might be "bright light is bad for your eyes," and sunlight is the brightest light, so you are probably no worse off with a SAD lamp than if you lived in a subtropical locale and got enough sun.

I'm not sure what to make of this, for now. I would recommend not looking directly into bright SAD lights. It is probably better to avoid blue-only SAD lamps. It may be worth experimenting with using "warm white" lights rather than "daylight" ones—although my weak guess from experience is that warm light is less effective for SAD treatment. If you have early-stage macular degeneration, or know or suspect you are particularly at risk, avoid bright LEDs, or at least consult a retina doctor first.

Thank you for this article.

Peter's picture

Thank you for this article. One question: Do the led strips give off UV light? And if so, what is your thinking around that? Seems like there is a common sentiment that UV light should be avoided in SAD treatment lights. I haven't dove into any research on that though, so it's more a curious question.

Thank you!

LEDs and UV

Thanks, that's an excellent question!

Normal LEDs emit very little UV—less than any other type of artificial light. (And much less than sunlight, of course.) It's not zero, but it's close.

In the past few years, specialized UV-emitting LEDs have been developed for UV-specific applications; but they are expensive and wouldn't show up in any ordinary LED lamp or bulb.

Fluorescent lights generate a lot of UV. Ones intended for general lighting include a filter to eliminate most of it. However, some "full-spectrum" and especially "grow light" florescent bulbs deliberately allow UV through. Those should not be used in SAD lamps.

Only 140W

Wow, that's disappointing!

Are kilowatt meters inexpensive? I'm tempted to check mine.

Do you have any subjective report on whether the bar has helped with SAD?

January and February have been quite unusually dark where I live (most overcast on record, I believe). I haven't been depressed, but my focus and productivity have been poor. I do think the lights help (because I'm not depressed, and not entirely paralyzed, and because there's a sense of relief as soon as I turn them on). But, it's totally possible I'm fooling myself!

I have "900W" in the three bars I'm running, and have definitely been tempted to add more.

AC power meters

Not expensive. You can buy for maybe $40 at home store, for home energy efficiency surveys. Kill-a-watt is one brand.


Luminous flux

Daniel P's picture

Where are you getting the luminous flux rating of 30,000?

The Amazon product details list it as 13,200 lm. Or is it 13,200 lm at 10 V, rather than at 30 V?

Product switcheroo, and lm/watt

Whoa... The product I bought was the WoneNice 300W 30,000lm bar, which had the Amazon ID B01M1FV2DA. That's what my Amazon order list still shows. (Screenshot here for the record.) Amazon is redirecting that to product ID B016LTDPLO, the Tenviii 32" 180w 13,200lm bar. (I'm not sure if this is a screw-up or a scam.) I've updated the post to reflect this.

When I wrote the post, all the auto light bars advertised around 100lm per watt. Rob and others above expressed skepticism about this, and it looks like they were right. Most are now rated around 60-70lm/watt. So, the 300W bars are now advertised at 17,000 to 22,000 lm.

Either they were lying before, and got forced into the update, or it may be that (as Rob also suggested) the lights were running too hot, and they've decreased the internal voltage (and light output) to increase their lifespan.


Baers's picture

First off thank you for starting this blog. I read both the corn bulb post and this one regarding the LED bars. It is really relieving to see that other people are also suffering from SAD, and are actively looking to find a solution to it. In the past few years I've become more aware of the symptoms and now am almost certain I suffer from SAD. Initially I thought my cognition just slowed down in the winter time compared to spring/summer, but it extends to so much more. My drive, my outlook, disposition, demeanor, concentration are all so drastically different in the colder/darker months than the warmer/brighter months, and I have serious issues with sleep in the winter (insomnia and crazy night anxiety), all of which basically go away when sunlight is at a certain strength. I do notice a bigger impact when the UV is at higher number (late march - september), but am very curious to see if I can see improvement without UV exposure with these LED bars.
Just had a few questions/comments:
1. Amazon now has 30,000 lm led bars - so it looks like they switched back from the previous 13,200 they had tied to that original link.
2. I read that these light bars aren't drawing the full 300 watts from the power source - is it possible to power these with just two 300 watt power sources?
3. I'm so unfamiliar with these power sources - so after I order and receive them - is it pretty straightforward how to set them up? Do they plug into anything or are they self contained power sources that just run out over time?

Any clarification on some of the above would be GREATLY appreciated. Thanks David for creating this webpage, can't tell you how good it is to find other people who have to deal with this.

Power supplies

Yup, SAD definitely sucks!

We don't currently know how much power the bars actually draw. You could try using less power than they are rated for, to see if it works. Or, they are inexpensive enough that maybe it's not worth experimenting.

The power supplies plug into the wall. However, the ones I got don't come with a power cord. If you are completely unfamiliar with wiring, you might want to get help to hook them up. If you know how to use wire strippers (the only tool needed), it will be straightforward.

For each power supply, you will need a power cord with a plug. The simplest way to get that is to buy an extension cord from any hardware store, and cut off the "female" plug end. Leave the "male" end to plug into the wall. You strip the insulation off the bare ends of the wire and stick them in the input side of power supply. Then there are screws you tighten so they don't fall out.

The light bars have a cord coming off them with two wires inside. You can cut it to length, strip some insulation off the end of the two wires, and connect that to the output of the power supply.

Received all equipment just need to set it up

Eric's picture

Thanks for the explanation David. Most of it makes sense, but unfortunately I am unfamiliar in this space and calling myself an amateur with this type of equipment would be an understatement. With that said, on the power supply the "input" is the L&N (labeled A/C) on the left side of the power supply, correct. Also what voltage should I be setting the power supply at, I'm assuming the 110v it is set at is ok. The other larger power supply I have is preset with no option to adjust at 100-120. The bars have a chord coming off of them but it is pretty short, with a red and a black wire. There is also a separate cord that came with each light bar that is extremely long, the light bars red and black cord I believe connect with the red and black wires from the longer cord and I believe on the other end of the cord there are red and black rings which probably are where you connect in a vehicle. There is a third piece that sticks out of the longer cord that looks almost like the female end of an extension cord but has four inserts and the insert size is very thin... I have no idea what this is for was hoping this could somehow just connect the lights directly to the wall (bypassing the power supply) but I'm guessing that's not its purpose.

Any advice would be great. Thanks

Wiring it up

on the power supply the "input" is the L&N (labeled A/C) on the left side of the power supply, correct.

Yes; but actually there are three power input connections. You should use three-wire power cord with a three-prong plug. The third input is "ground," labeled with a symbol that has several parallel horizontal lines. "L" is "line" and "N" is "neutral."

So which is which on your power cord? Different countries have different ways of labeling them, and some countries have several. If you are in the US, line will be black, neutral is white, and ground either green or bare (no color). Sometimes they aren't color coded, but instead the neutral wire has a rib or groove in the plastic insulation around it, and then usually the ground will be middle of the three wires.

If you are in the US, then plug power is 100-120V, and you should set the power supply for that.

The bars have a cord coming off of them but it is pretty short, with a red and a black wire.

Here the red wire is positive (+ voltage) and the black one is negative (- voltage). You should connect those to the +V and -V outputs of the power supply. (One of my power supplies has the negative output labelled COM.)

The other long wire thing is specifically for auto use, and not helpful for your purposes.

Add new comment


You are reading a metablog post, dated November 22, 2016.

The next metablog post is What they don’t teach you at STEM school.

The previous metablog post was The Court of Values and the Bureau of Boringness.

General explanation: Meaningness is a hypertext book (in progress), plus a “metablog” that comments on it. The book begins with an appetizer. Alternatively, you might like to look at its table of contents, or some other starting points. Classification of pages by topics supplements the book and metablog structures. Terms with dotted underlining (example: meaningness) show a definition if you click on them. Pages marked with ⚒ are still under construction. Copyright ©2010–2017 David Chapman.