Hacker News Reader

I always cringe a bit when folks approach hardware like they do software

Not sure what you mean - was there some terrible mistake in what the author was doing?

Software doesn't short out and burn down your house, for example.

Low voltage systems are a lot safer, but not immune to trouble. Even USB ports occasionally catch on fire if there's a damaged cable or crossed pins.

Actually, they're not because the current is so high. You can easily start a fire.

I mean mostly safer from electrocution. The risk of fire is still there.

But software might have a race condition that causes you to get zapped by a lot of radiation.

https://en.wikipedia.org/wiki/Therac-25

Most people don't have one of those in their house. Also a case could be made that it wasn't software that irradiated people, but the hardware.

If there was the potential for danger, the hardware should have had a limiter built-in. We have fuses in equipment for much the same reason.

Consider the following quote, which concludes the discussion on power -- "The wires we use to provide the power also have resistance. Allright, so what do we do? Let's split the four strips into two segments, and power each one in the middle. This makes the situation much more palatable. On one side I used a PC power supply, and on another two parallel 5V15A supplies. They are both well isolated, so there was no potential difference between grounds or signal integrity issues."

So I don't doubt this works for the author's setup, but it is by no means the correct solution to distributing power across his lights. There are three problems with this, first the PC power supply (being unspecified) may, or may not, actually regulate the 5V line when the 12V line is unloaded. So someone reading this might spend hours trying to figure out why their supply isn't producing 5V when the answer in the general case is to load the 12V supply with a 15W load (a 12V lightbulb works well for this).

Next up he adds two 5V supplies that can provide 15A in "parallel" but the "grounds are isolated". So he is assuming there might be ground cross talk, but what he really should be concerned about is whether or not the power supplies are designed to run in parallel. If supply A produces 5.05V (well within spec) and supply B provides 4.95V, then supply A will try to drive all the LEDS and back drive supply B. Of course in so doing its output will come down, and eventually supply B will get a chance to contribute some power, the supply with the slightly higher output however will run at 100% of its rated output, and can then heat up. This can sometimes cause it to go into thermal shutdown, and suddenly supply B will have all the load. Now it over heats and supply A is called upon again to carry the load. Power supplies that are designed to be run in parallel will have a current mirror which lets each supply see what the other is producing and split the load between them, they are essentially current controlled supplies at that point.

Third, there are lots of places that sell 60A 5V supplies if you want one, although a more typical arrangement might be to bus a higher power and then run independent supplies with a common return at each group of LEDs. So perhaps a 48V bus, with 5V step down supplies every 5M.

The simplest solution is of course to just use a bus bar. Provide a suitable gauge wire (8M of 8Ga wire will have a resistance of about 10 mOhms. Less than .5V drop over the length. Have the bus wire go 8M along two walls, and then out the other direction 8M along the other two walls. That solves his power supply issue caveat he needs a 300W 5V power supply. Since high power GPU boards are so common, if he wanted to use a PC power supply he could run 12V on one half of the room and 12V on the other half split into two 150W feeds, with 12V->5V supplies to down convert at the LED strips.

However since he missed all that (and I think its just missing some of the EE basics) and may (or may not) get a fire in his dual 15A supplies, I cringed.

Know the rules before you break them.

Dmitry has designed at least two of the consumer electronics products you have at your place.

He takes pretty brilliant shortcuts with his personal projects (see his others). While you're cringing, I'm grinning. A true 'hack' should do exactly that.

Are you suggesting that Chuck is wrong, and that it is risk-free to run two computer power supplies together in parallel even if they are not designed for this purpose? Are you also suggesting that all consumer electronic products are representative of good design principles?

Or are you grinning because Dmitry has so far gotten away with a risky approach? You, Dmitry, and Chuck probably all know more about electrical engineering that I do, but I've done just enough with low voltage power supplies to share Chuck's concern. The result is cool, but there might be a better way to get the same result.

Alternatively, perhaps Dmitry meant "parallel" in the sense of "separate", rather than electrically parallel. That interpretation might make more sense, since the "just hook them up in parallel" approach seems so obviously poor.

> Know the rules before you break them.

Yeah, well, none of his "calculations" made sense to me. Actually, there is no calculation, just random figures dropped randomly.

There is ABSOLUTELY NO WAY he gets 4V at the end of a 4m strip of 240 LEDs when they are at full intensity if the strip has a resistance of 2 Ohm.

Without even doing the full calculation, let's just calculate the drop caused by the single last diode group at the end of the strip: dV = R * I = 2 Ohm * 60 mA = 0,12 V

The others diodes groups will face less and less resistance as they get closer to the power source, but still, with that one single group, we ate 0,12V out of a 0,50V budget. Just imagine adding the voltage drops caused by the... 239 remaining diode groups.

(The total voltage drop supposing all groups pull 60mA is actually over 14V, so it just does not work. It only _seems_ to work because LEDs emit light with less than 4V and much less than 20mA each.)

He also writes: "Lessons learned and improvements to make: Really, what I should do is construct better power supplies, with sense lines running down from the top of the wires, since even now there is a significant voltage drop over the wires that run the height of the room. With sense lines, a supply could at least compensate for that drop."

Yeah, sure, and how do you compensate the voltage drop at the end of the line? By rising the source voltage. Which means you fry the LEDs which are closest to the source...

No, just add a pair of wires behind the strip and connect it to the power bus of the strip from place to place. Its resistance will be much lower than the 2 Ohms of the strip (supposing he did not screw this measurement too).

> Dmitry has designed at least two of the consumer electronics products you have at your place.

One must be the new satellite receiver that takes 2 minutes to boot and crashes every other day. As for the second one, I wonder.

RGB LEDs aren't particularly suited for general purpose lighting anyway. Good white light is fairly broadband, which is why "white light" LEDs use a phosphor to get a broad emission spectrum.

RGB+W LEDs exist, but I haven't seen them used in an LED strip configuration yet, unfortunately.

> RGB LEDs aren't particularly suited for general purpose lighting anyway.

This. A few years back a client I was visiting had an RGB LED setup similar to this in their conference room, where our team was temporarily "housed" during the trip. It provided monumentally annoying light no matter the settings.

If you're doing lighting for a space that will be used as a club or lounge where funky effects are desired and the inability to really see are OK... go for it. But even as "mood" lighting, full RGB control provides a vast, vast range of options that are truly terrible and only a few that maybe aren't.

The thing is, the good current white LED bulbs are using blue+phosphor approaches to get broadband color in the warmer end of the spectrum. This trounces RGB approaches thoroughly, and they're getting remarkably good at this. Philips even has a dimmable "white" bulb that gets warmer as it's dimmed.[1] It looks fantastic, IMO. It's that kind of light quality manipulation that I find really interesting, much moreso than the effects-oriented lighting.

[1] http://www.lighting.philips.com/us_en/lightcommunity/trends/...

The “blue” part of blue + phosphor is unfortunately a huge spike, and for any bulbs bluer than about 2500K CCT, ends up being awful at night. Unfortunately there are now many such (sometimes even up to 5000K) being used as street lamps and other outdoor lighting at night, and the result is extremely unpleasant.

The bluish “white” LEDs might be okay for daytime task lighting though, or for help waking up in the morning.

The one in your link that goes down to 2300K sounds more reasonable, but not too many people are using those as far as I can tell.

>and for any bulbs bluer than about 2500K CCT, ends up being awful at night.

How so? And as I've posted previously, you probably shouldn't use anything below ~4000K anyway if you want to have healthy eyes.

Light in the blue part of the spectrum is very distracting in peripheral vision and causes more glare because it saturates rod and short-wavelength cone photoreceptors, and it causes the eyes to adapt to an expected higher light level, which wipes out night vision.

For both drivers and pedestrians, blue-wavelength glare impedes vision into the shadows, causes distraction, and slows reaction times.

If you want more reasons not to have blue-wavelength outdoor lighting at night, it also disrupts melatonin production and sleep rhythms causing all sorts of human health problems, heavily disrupts wildlife circadian rhythms (both animals and plants), and creates more light pollution, making life dreary for skygazers and amateur astronomers.

The best outdoor lighting is low-pressure sodium lamps. High-pressure sodium lamps, “warm” incandescent lamps, and LEDs at a CCT of ~2000–2500K are also not too terrible, especially if well shielded and not too bright: In general outdoor lighting can safely be much dimmer than most cities make it, because human vision can adapt to extremely low light levels just fine.

Bright white LEDs of ≥3500K CCT are absolutely awful outdoor nighttime lighting: unpleasant, unhealthy, environmentally destructive, and wasteful of energy.

For more info, see http://www.csbg.ca/articles/APPLIED_SCOTOBIOLOGY.PDF and http://darksky.org/wp-content/uploads/bsk-pdf-manager/8_IDA-... (and the rest of http://darksky.org)

Almost none of that is true, or it's only true if the lighting is designed incorrectly. Try to try it for yourself.

The recommendation of using red light to preserve night vision is mostly a myth. While at the same absolute brightness red light indeed reduces night vission less, you need a much more powerful red light source to reach the same subjective brightness, which cancels out any potential benefit.

The same is true for outdoor lighting. The eye is the most sensitive to ~505 nm blue-green light, so you need much dimmer white/blue-green light sources to reach equal visibility with orange ones, which both saves energy and reduces light polution. White LEDs are only a problem where no compensation was made for their higher subjective brightness. When LED lighting is used correctly, it's quite pleasant and it looks kind of like unusually good moonlight.

No you can't use dim orange/"warm white" lights, because human vision can't night adapt to them, it needs the short wavelenghts for that which are missing from these light sources.

It's not true that high color temperature LEDs are wasteful, on the contrary, low color temperature LEDs are inherently less efficient. Warm white will always be less efficient than more bluish light and there is nothing that can be done about that, since it's a feature of our vision.

And what I meant in my previous comment is that it's likely that the exposure to unnaturally low color temperature light is what causes nearsightedness: http://www.nature.com/news/the-myopia-boom-1.17120 and http://iovs.arvojournals.org/article.aspx?articleid=2127682

I’ve just spent several months researching this topic, and doing lots of informal experiments on myself and other people. In other words, I have tried it for myself. :-) If you’re somewhere in the SF Bay Area and want to discuss in more detail, I’d be happy to give a mini-lecture about the subject, or about human color vision more generally.

At the typical brightness of streetlamps outdoors in most cities, humans are using photopic rather than scotopic vision (that is, using the “medium” and “long” cone photoreceptors), which is fairly sensitive to anything between about 500–650 nm. For photopic vision, the bright blue spike of a white LED provides almost no useful contribution to brightness perception.

Under starlight or moonlight and no artificial lights, you would be correct, but that’s not the condition we’re discussing. If outdoor lamps cast no more total light in aggregate than the moon, then bluer wavelengths would be fine. In practice they are several orders of magnitude brighter than the moon, which means they saturate the rods and mostly wipe out scotopic vision.

The cause of the nearsightedness epidemic (e.g. in East Asia) is children not going outside in the sun during the day. The details of the lamps they use at night (indoors or outdoors) are irrelevant to that.

I became interested in this topic because the streetlamps on my street were replaced by white LEDs at ~4000K CCT, and they are truly awful. They cause extreme amounts of peripheral-vision glare, and they make the previously charming street feel like a prison yard or sad industrial warehouse. The previous high-pressure sodium lamps are so so soooo much better.

> the typical brightness of streetlamps outdoors in most cities, humans are using photopic rather than scotopic vision

Becase they would be useless otherwise. Take a 650nm led and a 500nm led and try turning them down slowly. The 650nm one will just go off with very limited, if any, dark adaptation. You will be able to see the led glowing, but the room won't seem to be illuminated. With the 500nm light, there will be a wide range to which the eye can adapt easily.

>The cause of the nearsightedness epidemic (e.g. in East Asia) is children not going outside in the sun during the day. The details of the lamps they use at night (indoors or outdoors) are irrelevant to that.

The evidence for light chromaticity seems to be more convincing. It's supported by animal models and it provides a plausible mechanism. The light intensity on the other hand is suggested by nothing but weak correlation, which could be bacause children who spend more time outside are on average exposed to less yellow light.

You can have a rising/setting sun in your peripheral vision without it feeling unpleasent, the lights are either way too bright, poorly constructed or there is something wrong with your eyes. (a cataract maybe?)

I have RGB+W LED strips that light my home office.

http://imgur.com/a/JgWV8

The LEDs are 5050's (not individually addressable), and it alternates between white and RGB LEDs, controlled via analog/PWM (there are +5V, R, G, B and W connections).

I bought this from ebay, and it came with a remote and controller (and a crappy wifi controller that I've never gotten to stay connected and working for more than 2 minutes). The remote is kind of crappy, but it does allow independent control of white brightness, RGB hue, and RGB brightness.

My office is a 10x10 room with a vaulted ceiling, and I have two boxes each with two 1m strips (4m total) mounted on top of cabinets on opposite sides of the room, shining up and mostly reflecting off the ceiling.

I'm very happy with this. That said, I primarily use the white (which is a "warm white") though often mix in some subtle color (eg: 50% brightness of the white). At night I'll often have more of a 1:1 mix or even the color slightly stronger, just because. When turned up full, it's significantly brighter than the 50W halogen floor lamp I used to have in here.

The key thing that makes this usable for me is the white LEDs. Just to see if the RGBW vs RGB strip was worthwhile, I did actually try turning the RGBs on full when I was building it (the controller I have doesn't let you do this) and it's just not quite the same. Even though I have the strips projecting onto the ceiling (so it's indirect, soft light), the color is just.. off.

If you're considering RGB LED strips as general purpose lighting, I'd highly recommend to either go with RGBW or mount a white LED strip alongside the RGB one.

Yeah, except it occurred to me that for the price of Hue bulb, I could buy a RasPi and a Unicorn Hat that could be controlled via Wifi.

OK, I don't know the brightness difference, so maybe I'm talking out of my hat, but those high-tech controllable RGB LED light bulbs are still way out of reach for anyone without a lot of disposable income.

For a lot cheaper you can use an esp8266 and a ws2812 LED strip. The esp8266 is <$5 and includes wifi and a microcontroller that can be programmed in JavaScript, Lua, or C. Various ws2812 forms (strips, rings, matrices) are available for cheap on eBay. You can reproduce the functionality of an Phillips Hue for less than half the price. This is assuming that you can write your own smartphone app to communicate with the esp8266 and have decent soldering skills.

What I've noticed about the maker community isn't really related to disposable income. Most people with skills like programming and soldering have disposable income. The people that could benefit from saving money by making their own stuff don't usually have the skills to do so because they didn't have good schooling.

Actually I think price does have a significant role in the Maker mindset. I'm sure when they see one of these cool, but really expensive products, like the Hue, a lot of folks will say, "Oh, I could make my own for much, much less." and some will inevitably add "With blackjack and hookers!"

Anyone could benefit from saving money by making their own stuff, if they are interested (i.e., their time is worth it) and they have the knowledge and skills... and the project I suggested wouldn't take more than hardware skills than plugging things in and no more software skills than a few lines of Python.

I can't imagine the maker movement is made up of the idle rich, but of everyday people, some with mortgages and kids (like me), or some who are themselves kids. Given how much one of the main selling points we hear about neat new devices like the Pi is the price, I think your characterization isn't correct.

I'm sure in 10 years, something like the Hue will cost as much as a regular light bulb does today, but for those of us who don't have hundreds to drop on a cool colored lighting system for their homes, there are other options, just like you described, that are much cheaper and often not complicated at all.

True, although a lot of people don't know what to do with them and so get rid of them. I found a starter set (3 bulbs and base station) for $25 on Craigslist.

The "Mi-Light" brand stuff you can find on Aliexpress and eBay is inexpensive (like this: http://www.ebay.com/itm/221527712289 ). No idea as to the quality, though.