Injecting Power into pixels

Benslights

Dedicated elf
Joined
May 2, 2010
Messages
2,882
Location
Elderslie
Hi Everyone

from chatting with you guys i know i need to inject more power after every 50 pixels. is it possible to see some pictures or a diagram of how your are doing this.

i would like to use rays connectors to do this on my display as i want to make it all easy to pack up and store.
so do i need a mix of 4 pin and 2 pin connectors?
also when injecting power i should do the negative aswell. i this correct.

look forward to your assistance.
 
thought so David thanks for confirming that side of power injection for me.

would love to see some pictures or diagrams of how every one is doing this.
 
I use a 4pin male/female connectors, 2 pins passthrough for the clock/data, then the power seperates out....
PowerInjectionConnector.jpg

The power pins are not bridged at the connectors, so I can fuse the sections in to seperate circuits.
Also all my strips have male & female plugs on either end - so they can be swapped/reconfigured if required - hence why this hasn't been attached directly to a strip.
 
thanks driver that is a great example and the picture makes it nice and clear
 
I was considering a long end-fed 5V pixel string and the numbers just didn't work for distance. By long, I was thinking of asking Ray for 6" between pixels, so a 50 pixel string would be about 25' long and 4 strings (my target) would be 100', fed from one end.
I was concerned about even one string. 50' of 22 ga is 0.8ohm, times average current of 1.5 amps = 1.2 volt drop by the end of the string or 3.8V; the 2801 spec says it works down to 3.3 volts, so the digital logic should be OK, but will the constant current curcuity plus a blue diode fwd drop at that low a voltage on the analog side? Apparently it works for people, tho.
However, 200 pixelx x 60ma = 12 A. Too much for 22ga wires at the first pixels. Also way too much voltage drop at the other end - 200' of 22ga wire (power and return) is something like 3.2 ohms, times avg current of 6 amps = 19 V of drop. Kind of a big hit for a 5v supply, eh? (With constant spacing of LEDs, the voltage drop grows with the square of the string length as both resistance and current increase linearly).
So I thought, I could just parallel the string with some 18 or 16 ga zipcord and inject more power every 50 pixels. I even figured that one can reduce the total drop by feeding power at both ends of a given 50 pixel segment - the lowest voltage would then be near the middle of the string, with only around 25 pixels to support, and more like 0.3 V drop from the 22 ga wire, a 0.9v improvement.
But that doesn't work so well either. The drop along 18 ga wire at the end would be something like 10.5 * 0.32 + 7.5 * .32 + 4.5 * .32 + 1.5 * .32 (using the both-end-fed 50 pixels strings to gain about 0.9v improvement on the 22 ga wire end). Still around 8V drop on the 18ga plus 0.3 on the 22ga - way too much.
16ga doesn't buy nearly enough, and in fact even running a pair of 12ga wires won't - it still drops about 2v on the parallel power line (plus 0.3v on the 22ga). Actually, you can tweak this by using the "both ends fed" only on the outer strings, but it still just doesn't work.
So - you just don't want to run very many pixels on a line fed with 5v power from one end. That square factor kills ya. You could run at 10ma rather than 20 per color per pixel - that would help. Or spread the pixels further apart (eg: every 8 or 10" rather than 6" along the 100' line) so there is less average current. With some power supplies you can tweak the output to 5.5v, gaining a bit more headroom. It's still pretty tight.
So I realized that either you need to run on 12v (more headroom to handle voltage drop), or generate the 5v from something higher (eg: 12v, 36v, 120v) with power supplies distributed along the string.
Or - just don't run that long a line. For example, if you can power from the middle of the 100' run, that helps greatly by making it two 50' end fed runs (this is really just another way to "generate the 5v with power supplies distributed along the string" rather than end fed). 100' of 60ma pixels at 5v fed from one end just isn't a good idea.
I'd appreciate somebody checking this math. I rounded to 1-2 significant figures as our basic values are not known to any more accuracy, so don't worry about more precision - it's the magnitude that matters here. If I use 5v pixels I'll not push the length of the string nor of the supply line from the supply.
Does anybody know how much voltage drop the 2801 5v RGB pixels can tolerate before the colors distort? I'm guessing that even if the 2801 is handling the digital side OK, the blue and green LEDs are going to run into problems before the red, and that could happen at more than the 3.3v the chip itself needs.

Does
 
Without all the tech calcs which is not my cup of tea - I used a hands on approach 100 pixel string at 12.5 cm spacing fed power in both ends and on white the pixels each side of 50 were starting to show some pink so given that you would need to inject every 6mt or alternatively and it would probably be easier go 12V
 
For long strings I'd be looking at a 12V pixels.

I might be going slightly off-topic, but another option could be to create a higher-voltage DC distrubution supply, then have multiple DC-DC stepdown regulators to tap 12/5V supply as required. This allows your lengthly runs to use cheaper thinner gauge cable.

J1SYS also have this new kit - http://www.joshua1systems.com/ecg-ppx/ - which combines the regulator idea along with 'signal extension' for the data/clock. The outlay for this gear would be counter-acted somewhat (or completely?) by the fact you can then use cheaper cables everywhere. And remaining cost spent on also not having weirdness from voltage-drop or clock corruption will also be money well spent in my books :)
 
all good information driver i have looked at those before but always good to refresh
 
Zeph


Maths all seems to be close to reality. I run 112 count 100mm spaced pixels and feed the string from both ends with local 24V - 5V converters that have no more than 6 feet of 16G wire from converter to string ends. If one end is unplugged 1 and even 2 colours are fine but white will start showing Pink around pixel 65 odd.


It's not just the drop on the pixel string but the drop on the feeder cable.


For your scenario I would be tempted to feed power not from the ends but inject at the 1st and 3rd quarter points with local 24-5V converters.


Cheers
Phil
 
Back
Top