Hi all,
One of my projects for 2015 is to replace my LED string megatree with a pixel tree. I am planning to use 32 strings of 50 Pixabulbs for this tree, along with an E682 controller and two 12V 30A power supplies, each powering 1 side of the E682. I want to run 2 strings (100 bulbs) off each port on the E682, in an up-down zigzag configuration, to give me 32 strings of 50 bulbs.
I measured a string of the Pixabulbs that I intend to use (6" spacing), a 50-bulb string draws 1.47A at 100% white, I will call it 1.5A.
According to the E682 manual, you can power each side of the controller from a different power supply, and the input for each side is rated for 32A.
If I connect 2 strings per port, this will draw 3.0A per port. The E682 manual doesn't say what each port is rated for, but they come with a 5A fuse, so I assume 3.0A per port is fine.
If I connect 2 strings on all 8 ports of one side, this will draw 24A (3A * 8 ports) at 100% white. This is right at 80% of what the power supply is rated at, but below the 32A rating for a side of the E682.
Is this pushing things too close to the limits, or is it ok, knowing that I probably won't ever run the tree with all strings at 100% white?
- Martin
Looks like you got it all figured out. Math looks good.
80% at full load is good to go. Cause you wont be at full load all of the time. Might of fact only for short period I would presume.
The math seems to work. Reality may different - especially when you add in additional wire from the controller to the start of each string. Don't expect the numbers for two strings to be exactly twice that of one string. And that voltage has to fight gravity to get to the top of the tree. (OK, this last part is not quite true :lol:)
If you power inject at the bottom of each 100 pixel run (something you may want/need to do to ensure those last pixels have sufficient voltage), then you're certain to be in good shape because the amp draw will be shared/split with the injection point source. The wiring will likely be cleaner to power both sides of the controller and the beginning of the strings from one PS and then inject at the bottom with the other power supply.
You make a good point, I didn't measure things exactly as I plan to use them. I have all the parts on-hand, so I took a bunch of measurements and was confused by what I saw. This probably makes sense to an electrical guy.
Why isn't the sum of the draw for the full individual colors the same as the number for full white?
Why is the draw for two strings less than twice the draw for a single string? Does the voltage drop play a role in this?
Why does the draw go down a little when I add a 10' extension? More voltage drop?
I'm also glad you mentioned power injecting, I hadn't planned to do that. I have some long runs on my roofline and they look fine. While I was taking these measurements, I put the 1st bulb and the 100th bulb close to each other, the color difference is noticeable without power injection.
All in all, I think I am still ok with my power supply plans. Even with injecting power, I still only need 16.88A per side, so I think my 30A power supplies will be just fine.
- Martin
All wire has resistance. Teh resistance is generally specified as ohms per some distance. The longer the wire, the greater the impedance (resistance). Passing current through any resistance causes a voltage drop. This means that the voltage at the power supply end of a 50' length of wire that has current going through it will be higher than the voltage at the destination end of the wire. The greater the current going through the wire, the greater the voltage differential. Now take the reduced voltage and apply it to a pixel, since the pixel is getting a lower input voltage and its internal resistance is fixed, then the current through it will be slightly lower and the brightness slightly dimmer. Not noticeable on a single pixel but very noticeable on a hundred pixel string. The next thing to think about is that the resistance per distance rating on most power cord is much lower than the same rating on the wire used in the pixel string (and then there is the impedance in the circuit boards junctions etc...) all of which results in lower voltages and less current draw.
A bit of a ramble, but we cannot escape ohms law and you are seeing the effects very clearly in your measurements.
msiebri wrote:You make a good point, I didn't measure things exactly as I plan to use them. I have all the parts on-hand, so I took a bunch of measurements and was confused by what I saw. This probably makes sense to an electrical guy.
Why isn't the sum of the draw for the full individual colors the same as the number for full white?
Why is the draw for two strings less than twice the draw for a single string? Does the voltage drop play a role in this?
Why does the draw go down a little when I add a 10' extension? More voltage drop?I'm also glad you mentioned power injecting, I hadn't planned to do that. I have some long runs on my roofline and they look fine. While I was taking these measurements, I put the 1st bulb and the 100th bulb close to each other, the color difference is noticeable without power injection.
All in all, I think I am still ok with my power supply plans. Even with injecting power, I still only need 16.88A per side, so I think my 30A power supplies will be just fine.
- Martin
Very nicely done. A nice and concise experiment and documented well. Kudos to you!!
MartinMueller2003 wrote:All wire has resistance. Teh resistance is generally specified as ohms per some distance. The longer the wire, the greater the impedance (resistance). Passing current through any resistance causes a voltage drop. This means that the voltage at the power supply end of a 50' length of wire that has current going through it will be higher than the voltage at the destination end of the wire. The greater the current going through the wire, the greater the voltage differential. Now take the reduced voltage and apply it to a pixel, since the pixel is getting a lower input voltage and its internal resistance is fixed, then the current through it will be slightly lower and the brightness slightly dimmer. Not noticeable on a single pixel but very noticeable on a hundred pixel string. The next thing to think about is that the resistance per distance rating on most power cord is much lower than the same rating on the wire used in the pixel string (and then there is the impedance in the circuit boards junctions etc...) all of which results in lower voltages and less current draw.A bit of a ramble, but we cannot escape ohms law and you are seeing the effects very clearly in your measurements.
This is a good write up, read it a couple of times and come back to it tomorrow and read again! 🙂
The two things I would amend to it (more for others reading) are 1) that ohms law is voltage =current*resistance (V=I*R) which can be rearranged to solve what you need. A VERY useful equation for lots of things.
and 2) if you are using one of those online calculators (or doing the math by hand) for voltage drop, keep in mind that the drop occurs in both directions. Meaning that if you have a 10m light strip, you will get a drop on the voltage on the power going out, AS WELL AS the ground path coming back. Some online calculators will set your voltage drop for a 10m path to be 20m because of this, others leave it to you, so keep an eye out if some sort of calculation doesn't make sense to you.
The other thing to measure is the voltage at the start and end of each string. That would be very interesting to know in your test cases.
When all pixels get close enough to the nominal 12V, the total current should scale pretty much linearly with the number of pixels. However, if the voltage at a given pixel sags enough, there will be less current going through that pixel and it will be dimmer. A small voltage sag won't do this - the pixel driver will compensate, but that has limits. When you are getting dimming, the total current will be less than N* the current of an undimmed pixel. (Talking about voltage based dimming, not PWM here).
Yes, it's the resistance in the wiring that causes the voltage sag. For a small voltage sag, that's fine - it just means that the wasted power between 12V source and what the LEDs use will be dissipated in the wiring resistance rather than the WS2911 chip. For larger voltage sags, there won't be enough voltage left to fully light the pixel.
If you do power injection on both ends, then the dimmest pixels will be in the middle - but the difference between start and middle brightness won't be as large as the difference between the start and end without power injection at the end. You might want to eyeball pixel 50 vs pixel 1, tho.
If your total current for 100 is less than twice the total current for 50, then some pixels are getting less than their minimum voltage for full brightness (and are thus dimmer and drawing less current).
You can play with some numbers using the calculator at http://blinkyflashy.net . However this calculation assumes that each pixel still receives enough voltage to be fully bright. If it calculates an ending voltage which is "too low" for full brightness, then the calculations are guaranteed wrong (high). The purpose is to help you stay in the "normal full brightness" range of input voltage for each pixel (which depends on the pixel). (once you get below the normal range, it would become very hard indeed to calculate the total current because each pixel will be on a different point of a non-linear curve of voltage vs current.)
It doesn't calculate for feeding both ends - which gets interesting because the feeder wires will usually have different lengths. If the feeder wires are short for both ends, you can just calculate for half the string (and double the current).
I don't know the minimum voltage for a Pixabulb to be full brightness. It's higher than most 12V strings because the WS2811 drives two LEDs in series rather than one (for more brightness at the same current, a good thing). But the pixabulb has larger wires (18ga IIRC) than most, which would help some.
As a quick hint - for a pixels string (or strip) with fixed spacing between LEDs, the voltage drop at the far end is proportional to the square of the number of pixels (while each pixel is still in the normal range). So a 100 pixel string fed from one end has 4 times the drop of a 50 pixel string fed from one end (with the same spacing, and a short power lead).
Useful measurements.
So with long leads, bulb 1 vs 100: 11.07 vs 8.45 - visibly dimmer at end
And fed both ends, bulb 1 vs 100: 11.01 vs 9.85 - less dim at end (almost the same as bulb 1?)
I am thinking there's a substantial brightness difference between 8.45v and 9.85v, right (bulb 100 without/with injection)? (bulb 1 should be about the same, for reference)
How about the brightness of bulb 50 vs bulb 1? In the both ends case, it's down to 9.41.
I'm still puzzled tho by how large the difference is between r+g+b vs white - even for the 50 bulb string where ending voltage on white is 10.76v, which I would have thought was plenty for full brightness. Could you check the start/end voltage with just blue on full (say for the 50 bulb case)?
I note that R, G and B currents are close. That implies something like a constant current regulation, since the two red LEDs drop less voltage than Green or Blue. But even at voltages tapering from 11.96 to 10.76 (50 bulb string), the white current is lower than we'd expect.
Suppose the chip electronics draw 0.10A for the 50 bulb string, lit or not. Red, Green,Blue would add .58, .59 and .56 amps to that. So for white we'd expect 0.10+0.58+0.59+0.56 = 1.87A as a simple estimate, IF all bulbs had sufficient voltage to self-regulate the current. It only drew 1.47, so this is 0.36A less than naively expected.
I'm beginning to think that getting a handle on this may require plotting the current vs voltage for a single bulb at varying voltages. Brightness should be fairly close to proportional to current drawn (even more so if you first subtract the dark current value to get an estimate of the LED currents). (Of course we *perceive* brightness on more of a logarithmic scale, but that's another story). It's looking as if the current starts dropping sooner than I would have expected, as the voltage sags even a little bit.
Any chance you can measure the end to end resistance (of the power or ground wires obviously, not the data wire)?



