There's something going on that I don't understand with the power consumption of my

Archive FastLED Archive
1

There’s something going on that I don’t understand with the power consumption of my new 144 LEDs/m strips.

In theory, full-white, they should pull 86W (144 * 2m * 0.06A * 5V). My cheap-o-watt meter says I’m only pulling 40W out of the wall (with FastLED’s dithering switched off.) When I turn dithering on and setBrightness(128) I get 30W.

Watt’s going on?!

Well, half way through writing this I noticed the far end of the strip was a nice warm white colour, not the ugly blueish hue. After connecting power at both ends, I’m now drawing 56W on full-white. My power cabling is toasty warm too! Hmm…

2

You underestimate the power of the Dark Side! No, wait, that was something else. Maybe it was supposed to be “You underestimate the voltage drop of the Light Side!” Something like that.

Especially with 5-volt LED strips, the voltage drop as you go from from pixel to pixel can be accumulate to significant levels, especially if your animation is running with lots of white (or other highly-desaturated, pastel colors).

If you measure the voltage at the ‘far end’ of the strip, you can usually see the problem right away.

And the right way to deal with it is to do exactly what you did: inject power at multiple points, e.g., at the ‘near end’ of the strip AND the ‘far end’ of the strip. If the strip is long enough, or you’re running with lots of white, you may even want to inject power at one or more mid-points along the strip, too. I have a 1,024-pixel, five-voltLPD8806 setup that I’m pretty sure is going to need lots of power injection points…

How many pixels were in your strip, anyway? 288?

12-volt strips are less susceptible to this problem. Even if the voltage drops, say, two volts across the length of the whole strip, it’s still ten volts, which is plenty enough to fire the LEDs at full brightness. But on a 5-volt strip, a two volt drop across the length of the whole strip leaves you with just three volts at the far end - probably not enough to fire the blue LEDs at all!

At some point I noticed that strings of pixel ‘nodes’ are most typically sold in strings of 50 pixels for five volt strings, and strings of 100 pixels for twelve volt strings. My totally unresearched theory is that 5V strings have too high a voltage drop to sustain a working voltage all the way along 100 pixels, but that 50 is fine, but that 12-volt strings have no such problem.
5V: http://www.aliexpress.com/store/product/promotion-DC5V-WS2811-controlled-12mm-diameter-led-pixel-node-IP66-rated-50pcs-a-string-with-black/701799_804958154.html
12V: http://www.aliexpress.com/store/product/100pcs-DC12V-12mm-WS2811-led-pixel-node-with-all-black-wire-20AWG-IP68-rated/701799_1032319604.html

Also, if you’ve looked at these 12V strings up close, you see that each node actually has a tiny power regulator, and then what appears to be a 5V WS2811 chip! This suggest that 12V strings can be stretched out pretty far, as long as the voltage at the far end is something like 7 or 8 volts, enough to be stably regulated down to 5V.

I’ve run 200 pixel strings of 12V nodes with power just from one end with no problem. 300 pixels and it starts to matter how bright you’re running them. 400 is too many, and you need to inject more power.

Aaaaaanyway, please continue to post your findings about power draw and voltage drop!

3

Given that if I do manage to supply them enough power they’ll take it, I’m not sure what my final configuration should be: 4 x 36 LEDs, 2 x 25W power supplies, ~40W draw and ~3.75h runtime, or 2 x 36 LED strips, 1 x 25W power supply, ~20W draw and ~7.5h runtime.

I’m estimating a 1/3 duty cycle to get those numbers. For reference, the blue LED COB bar I had last year that the GF said “wasn’t bright enough” was in the order of 8W :smiley:

4

(Good thing is I’ve got a shit-ton of LEDs—I could make two of the 2x36 and one 4x36 and still have a whole reel left :slight_smile:

5

How are you measuring watts out of the wall?

6

Kill-a-watt?

7

It also depends on drivers … for example, the WS28x series will sink a full 20mA, whereas the LPD8806 will only sink 18mA per channel. While a small 6mA per (3 channel) pixel, over a length of strip, it makes a difference.

8

I am asking because I don’t understand the relationship between wall power and power supply output power when working with LEDs.

Are you trying to make sure that the wall socket isn’t overloaded?

9

Reminder / refresher:

Watts = Volts x Amps

So 20w = 5v x 4 amps

20w ALSO = 120v x 1/6th amp

So a 20watt lightbulb that runs on 120v draws just 1/6th amp, but 20watts of 5v LED is four amps.

10

What @Mark_Kriegsman said: I’m using a Kill-a-Watt (not literally, but a similar device), and the power coming out of the wall is an upper bound on the power being consumed by the strips. I just have a power supply to turn it from lots-o-volts and a bit-of-an-amp AC to bugger-all-volts and a shedload-of-amps DC in the middle.

11

Except for the bill, I still don’t understand. What is the benefit of knowing how much power is being drawn on the AC side of things?

Like you said 20 watts is 20 watts but you get there with different V and I.
For the input side (AC) I focus on the max rated current to keep from overloading a breaker. And multiple power supplies are powered on in sequence - not all at once.

For the output side I focus on keeping the power supply unstressed with a maximum draw of 80% of the rated output current.
This means using the right size wire and short cable runs.
Injecting power in multiple points in longer runs.

12

Kill-a-watt meters are rubbish as measuring the power usage of switched mode power supplies because they don’t have a power factor of 1. i.e. current and voltage not in phase.

This is why some equipment has a Watt rating AND a VA rating to account for the power factors created.

Use these meters only on resistive loads - any induction throws them right off.

13

@Michael_Sharnet I’m running this piece off a wall-connected PSU while it’s in development, but eventually it’s going to be bike-mounted and run off a battery. So the power consumption in watts influences my runtime. The power usage displayed on the Kill-a-Watt is a “worst case” power consumption (there are some losses in the PSU), but it helps me calculate my “worst case” run time off the battery. Of course I do need to take account of the current going through the 5V wiring as that influences the voltage drop in the cabling.

@Adam_Sharp I did notice my meter (again, not a Kill-a-Watt™, it’s a “Velleman” brand) displaying a power factor of 55 or so, as it’s reporting the 40W draw. If it knows what the power factor is does that mean it’s probably compensating for it?

14

@Robert_Atkins for my battery operated creations where my rough power estimates are not enough or if I want an accurate view of the current/watts/volts, I use a remote control battery tester. http://hobbyking.com/hobbyking/store/__22631__Turnigy_7_in_1_Mega_Meter_Battery_Checker_Watt_Meter_Servo_Tester.html

15

I did have a quick read of the manual from Velleman for their plug-in energy meter (presume this one). It does mention the ability to measure power factors. But it does not say how well and if it uses the results in any calculations.

It does state the meter is 0.1% accurate for power with a resistive load and no mention of inductive loads. So I would would suggest it is only a ‘indication’. As it is a class 1.0 meter. It can’t be used for billing, really stating it is ‘inaccurate’.

A PF of 0.55 is pretty much voltage and current about 56 degrees (Cos-1 0.55) out of phase. Making a the reading a long way off power wise if it is only designed to do resistive (1.00).

You could read chapters 11 & 12 here: http://www.allaboutcircuits.com/vol_2/index.html
Then you can see why it is very difficult to measure (and display) true AC power. Cheap meters are not going to bother.

16

Actually, thinking about this problem. It is pretty much impossible to measure to any accuracy power consumption of LED strip systems.

If you measure the AC supply side, you have to take into account power factors, even for old linear supplies. Switchers are even worse.

If you try to measure the DC side, then you are dealing with a fast pulsing on the rails due to the PWM drive of the LEDs. And this requires a fast sampling and averaging ammeter. Involving factors like Nyquist.

If you use a battery supply, then maybe a general ammeter on the ground side might work. As this will see a better signal, presuming you are using lots of fast smoothing across the rails. Certainly for AC chopped circuits (like lighting dimmers) you should ‘fuse’ the neutral and not the line side, as this gives a better current sense for the circuit cable.

17

@Adam_Sharp Thanks for sharing the link. I’ve been looking for an electronics refresher. It’s been more than 20 years since my last electronics class.

18

@Adam_Sharp Interesting. Just as a data point, for my panel project, this meter reported the consumption at the wall as being pretty close to the theoretical consumption if you multiplied out by the number of LEDs.

19

As another data point, with these strips the wall meter says 4W with no LEDs connected (just the Arduino Uno); with 72 LEDs connected I get 11W, 8W, and 11W for full-on R, G and B and 24W for white. If you subtract the 4W “idle” power for those, you get just about what you’d expect. Notably the green takes a little less juice than red and blue—is that some kind of gamma correction built into the WS2812B?

20

I don’t have time to do the computations now but I expect that you have hit a balance between the number of leds being driven and that this is nothing but a coincidence. I would expect the wall to show a higher wattage when compared with the LEDs.

The conversion of AC to DC is not 100% efficient.