I need some basic electrical engineering help with my reactive table project.

Archive FastLED Archive
1

I need some basic electrical engineering help with my reactive table project. It has about 750 WS2812 LEDs, which are mostly lit at 25% to 50% brightness, and mostly colors (not white). I’ve got two questions:

First, I’m currently using a 30A 5V power supply, but it might be overkill. And it’s bulky. Could I get away with less?

Second, I’m wondering about the sizing of the wires. Most of the references I’ve found on matching wire thickness to current draw do not mention voltage at all. So, here’s the question: if I’m drawing 30A at 5V from the power supply, do I need a 30A 110VAC power cord running from the wall into the power supply? I have seen cords rated for 30A (for dryers and ovens) – is this really what I need?

Thanks!

2

First, I am not an electrical engineer, so someone more qualified please chime in. :slight_smile:

750 LEDS @ 60mA each = 45A theoretical max current draw. I don’t think I’ve ever measured this theoretical max, even at 100% brightness, with all LEDs at full white.

So, at 50% brightness, full white, you’d be at roughly 22.5A, assuming it’s linear (probably isn’t)? Have you tried measuring current draw with a multimeter or kill-a-watt style meter? I think it’s best practice to run power supplies well below their max rating.

You could likely use a smaller power supply, and use FastLED’s power management to limit max brightness and power usage.

Regarding wiring, the current is inversely proportional to the voltage, right?

30A * 5V = 150W
150W / 110V = 1.36A

So, even the smallest AC power cords should be able to handle the load, right?

Anecdotally, this certainly holds true for even my largest pieces with 1,000 LEDs. I always load test them at worst-case, 100% brightness, full white, monitoring current draw and wire temperature with an infrared thermometer for a few hours.

I’ve never observed elevated temperatures on the AC side of the power supply.

The only elevated temperatures I ever notice are at the smallest gauge wire, or where I’ve split from a large gauge to several smaller gauge wires. If I observe this, I replace the splitter with a large gauge bus bar or distribution block.

3

OK, thanks. I need to get a meter capable of measuring that kind of current.

4

@Sam_Guyer for my build I have a 1200w power supply which is plug using a normal vacuum cleaner wire.
But for your power supply to your leds you need to big careful of the size of your wires if it’s too small they will heat up. Calculate how much amperes each strip draw and adjust the wire size accordingly. In any case the larger the better because the larger less heat.

5

@Sam_Guyer something most people forget is that the pixels can ask for a certain current amount, but ultimately it’s the power supply that fulfills that request fully, or not.
This is another way to say that I’ve run my 768 panel off my laptop USB port (limited to 2A only) many times, and it works fine.
What happens is that if I ask for the LEDs to be too bright, they won’t be as bright as I asked. Depending on the power supply (true for my laptop) it has a self protection circuit that will cause it to shut off and on, causing flicker. Bench power supplies typically just limit the current going out to what they’re rated for, and you don’t even see that you aren’t getting full amps, through (full as requested by the pixels).
I run as many pixels as you with lots of demos on them (not a full brightness because it’s just too bright), and I usually end up using 2-3A out of the theoretical 30A on white at full brightness.
This is another way to say that a 10A power supply will be more than fine, and you’ll likely find that 5A is enough too.
As for wire thickness, i’s a good question, but ideally you want to run multiple power wires to inject in multiple places on your panels.
Either you have thick wires for the long run and then you split it in thinner wires going to multiple places, or you just run a bunch of thinner wires directly from the power supply to your panel (which one is best depends on your setup).
Hope this helps.
If you’d like fun code to run on your matrix, you can look at http://marc.merlins.org/perso/arduino/post_2018-05-29_EDM-Party-Shirt-powered-with-FastLED_NeoMatrix-and-Adafruit_GFX_-plus-160Wh-10Ah-4S-worth-of-lipos.html
https://github.com/marcmerlin/NeoMatrix-FastLED-IR
Or simpler code split up:
https://github.com/marcmerlin/FastLED_NeoMatrix/tree/master/examples

6

Concerning wire sizes, Jason got it right on the AC side, you do not draw so much current and can use a normal AC power cord. I would just suggest that you need to pay closer attention to the wire size on the 5Vdc distribution side.

Here I would basically use that ‘voltage drop calculator’ tool and aim at never getting a voltage drop to below 4.5 Vdc anywhere in your setup.

7

Also… about PSU size… I would not go for a smaller one myself but if you really need to reduce the bulk of your setup, you simply need to control the total current draw by playing with brightness. The built-in FastLED power management does that for you but will you be happy with the brightness that it would automatically select ??

8

OK, thanks all. I have a good system on the 5V side using fat wires as busses to inject power every 60-70 LEDs. I just don’t know as much about the 110VAC side.

9

@Sam_Guyer sorry, I missed the 110VAC side question. So, think about watts. 30A x 5V is 150W. If you ignore losses, to get 30A on 5V, you need 150W from 110V, which is a mere 1.3A. Any power cord will do, even if you round up to 2A.
If you were using 30A at 110V, that would be 3300W, which does require a thicker AC cord (air conditioning and so forth) and would equate to 660A on 5V, that’s more amps than my Telsa deals with (it can do 400A at 400V, and it uses huge copper connectors to pass that amount of amps)

10

@Marc_MERLIN That makes total sense. I had a feeling that the 30A dryer cord was overkill!

11

Wire ratings are done for amps, not voltage - 30A from a 5V power supply is likely around 150 watts, if you want to assume a terribly AC/DC conversion, call it 300 watts - which is only about 3A at 110V.

I think the WS2812’s are around 60mA when full white? So you’d need 45A to run 750 of them at full white. At quarter brightness white you’d be fine with 12-15A - and presumably you aren’t going all white often?

Personally, I like to size power supplies for either full white brightness, or, barring that, I’ll pick a power supply size and then use power management to make sure I don’t over draw the power supply.

For reference, Flow, in the veracode office, has 480 WS2812 leds and runs off of a 10A supply, I believe - but I’m pretty sure I have the code limiting it to about 2.5A.

12

@Sam_Guyer a wire does not know how much tension is applied to it, it only knows how much current passes through it. A cable also does not know how much power is consumed at either of its ends. Voltage is not really relevant it only impacts the isolation and there is no big difference between 5 and 110V. Wire diameter is irrespective of applied tension. Lower voltages are harder to handle because they result in higher currents, there is no way to get around this . Thats why we smart Europeans use 220 rather than 110v, its actually safer, against intuition. More people die of fires than from electrocutiom In short if you really want to transport 30 A the cord to your washing maschine is the only safe way. The risk is not so much to burn the LEDs but to cause a fire by overheating thin wires, melting isolation resulting in shorts.

The norm in Austria requires 2.5 mm2 cross section for 30 A which is about 1/20 of an inch in diameter (evidently for each GND and VCC) on 30A. With all due respect for the FASTLED lib. Don‘t rely on software for limiting current and preventing fires. Fuses are cheap and save your from damage the insurance won‘t pay. Don’t calculate on average consumption but peak. Due to software errors all Leds may get on inadverntently. Being too bright even for only msecs can bring them in a state where you no longer can switch them off or otherwise control them because the data from the data line are no longer recognised and the chain stays in this suicidal mode. Where you place the connections every 2m to the chain sacrifice a 100 microF Elko, and a reverse diode against inductive spikes on switch off.
Ws2812 are reasonably cheap by now, but 750 start to appear in the budget of your project and it may be a major mechanical task to replace them. My Ws2812 dont‘t survive 5.3 V for more than seconds. And yes I tried.

The usual advice is to supply power and GND from both sides at least once every 2m.

While the Ws2812 and friends are easy to handle from the MCU for the purpose of wiring 12v or even 24v would have been much better. Thats why ready made products usually run on 12 V, so as to limit the currents. Bulky power supplies cannot really be avoided if you want to drive them anywhere just close to acceptable brightness over a long distance. Use a seperate power source for the MCU and the Leds. Don‘t supply the power to the LED through the voltage regulator of the arduino (that is through VIN to 5v), you‘ll barbacue the arduino.

13

Ps when using stranded wires for distributing such high currents which is your only choice to stay flexible enough, don‘t just stick the strands into the screwable connector but crimp all the strands together at the ends. The problem is not so much shorts between stray strands between ground and VCC. But a single strand might have to stand the full current and disintegrate. Soldering the end is NOT good enough. Local current induced temperature may rise above melting point of the solder and single strands may jump off.

Ps Ps. I hope i did not frigthen you too much. 50 A can be handled with amateur means. Spend some money on the components and the crimping tools, they‘ll pay off on your next project.

14

already a lot of great answers here, but FWIW, I run up to twice that number of LEDs frequently using a 2.1A output on a beefy USB power bank and have no problem in that 25-50% brightness range (especially with patterns that are mostly colors without much white)

As far as wires are concerned - as others have mentioned, if you use a smaller power source, then it really won’t matter too much,because you won’t be putting out serious current. I use 24 gauge for my power bus on big projects, but I’m highly confident I’d be fine with 26 gauge, and I’m transitioning to that for just about everything. Just always make sure you connect your power directly to the strips rather than through your controller/arduino

15

@chad_steinglass I tend to agree with you. Hobby projects with the Neopixels are really fun and easy and do not create too much worries. A bit more care is required when you allow to run your projects unattended or when you give them away. My daughter wanted music control leds in her room, and that was the first time I worried about amps and wire diameters. Before I went by trial and error, I burnt a couple of Neopixels, blew a few fuses under my fingers which is a bit unpleasant but no more than that. Yes, I do have a fire extinguisher in the room I play with my gadgets, just in case.

16

@Walter_Rorschach Very good point - there’s a big difference between something that you are always supervising vs an installation or a gift.

17

@Walter_Rorschach Thanks for your input. You hit on exactly the question I was struggling with: I feel like the current-carrying capacity of a wire should not depend on the voltage because it is just expressing the number of electrons that can pass through it. In fact, the current-rating charts that I have found do not mention voltage at all (and if they do, they just say “valid for voltages up to 1000V” or something like that).

18

Have fun with your project. After 20 years of technical abstinence in an administrative job I really enjoyed diving back into technology with the arduino. The learning curve with the arduino was plesantly steep. And I was surprised how little changed in the meantime. It is still C and Java, there are still S Records, and Unix as a cross developement system is what it used to be, and most of all I enjoyed finding out how much I remembered over such a long time and unlike Latin it stayed useful. Some advances are nice, like breadboarding being nicer than wirerapping. And the components and the tools starting from multimeters, soldering irons, and even scopes got dirt cheap as compared to then. The other week I sent a handful of IR diodes for 20cts and a microphone for 1.29 to a sick friend to cheer him up. My wife could not understand that he could enjoy such a trifle and called me a pennyturner. We makers know better. Old mens‘ hobbies are very affordable these days, at least more affordable than those of the youngsters. The laptop I bought for the arduino projects so as not to burn the USB ports on my PC cost 1/10 of that of my son‘s gaming laptop.
However, the greatest advantage over the glorious past are fora like the present one and youtube and arduino cc where people readily help each other and do not just showoff their projects but are geneorusly ready to share them for luv. Capitalism will die over it when people work for mere recognitiom rather than bucks. But then it did not die over Linux which is now part of the big fat expensive macs on which Jobbs earned the money for his yachts and planes.
Back to technology: Having the memory on the CPU rather than on the board makes things more predictible, a lot easier to wire. Some 50pins almost all of which are usable without decoders, multplexers.
In the interpreted languages having popped up since, I see limited benefit. Lua and MYPhython are unnecessarily complex for prototyping. I hope with the new chips having more memory someone is charitable enough to implement FORTH on them and I am completely back in the glorious past. And if not, maybe this is my project when I retire.

19

One thing that’s good to keep in mind is that the brightness of the LEDs is very non-linear. As we know, an RGB value of (0, 0, 0) is “off”, and (255, 255, 255) is full brightness. But if you compare that to (191, 191, 191) (75%) or even (127, 127, 127) (50%), you’ll probably observe that it’s still awfully bright, and you can likely go a good bit lower than that before you start getting levels that don’t hurt your eyes.

On top of that, our eyes are naturally very sensitive to lights when we are in a darkened environment. So it’s worthwhile to explore just what maximum brightness you really need for a given project.

Now, on the flip side of that, on the low end of the brightness settings, that non-linearity can work against you. At very low settings, color mixing becomes way less accurate that what you can observe on a high quality RGB monitor, and you’ll see “muddy” colors that aren’t what you expect. Also, you might see extreme perceived differences in overall brightness over a relatively small numeric range.

20

@Sam_Guyer @Walter_Rorschach - so after my comment in this thread saying don’t worry about it… I’m now just having my first taste of power issues and realizing my wire gauge is too thin😂

But I’m driving 2400 LEDs! Definitely the most I’ve ever tried. Sam, I’d still say you’re prob good with like 24AWG for 700 or so