LED Strip Power

[BryceKarlins]

Hello! I'm doing some blingification with Adafruit Neopixels this year, and I'm trying to find out the best way to power them. Because FIRST locked down the other 5V 2A port from the VRM, I see a couple good options. There's the BEC converter from the old control system that powered the radio, of which we have PLENTY to spare. There's also a small UBEC I got for another project a while ago.
Does anyone else see other options?

BTW: Neopixel strips need a relatively clean 5V, and anything over 6V will fry the whole strip instantly. Also, I'm using 4 meters, which should have a max current draw of ~8 amps (all LEDs on full white 100% brightness). I plan on using only 1/3 of that regularly, but don't what to fry anything if we accidentally use more than we expect.

I guess I just want some thoughts and suggestions. Thanks!

[FRC2501]

We are using the same LED's and we are using the 12V-5V converter from the old control system connected to the PDP with a 20amp breaker. It outputs 5V-5amp's and the pixels will only use as much current as they need, but anything over 6V will fry the pixels.

[wmarshall11]

Would throwing a couple of 5v regulators in parallel work for your application? Something like 4 LM323's on a perf board (with appropriate filtering) should be enough for your application.

Edit: See below and apparently don't do this. Also I may have a soon-to-be-crispy board floating around and should actually learn to EE.

[GeeTwo]

Originally Posted by wmarshall11

Would throwing a couple of 5v regulators in parallel work for your application?

AAAAAAAAAAAAA!

Putting regulators in parallel is never a good idea unless you put enough resistance between each of the outputs to allow the variation in output voltage to go somewhere other than frying electronics.

[Mike Bortfeldt]

I've used this successfully in the past.

[philso]

Originally Posted by Mike Bortfeldt

I've used this successfully in the past.

Use this or one of the 12V to 5V Voltage Converters from previous years (check the output current capacity first).

Originally Posted by wmarshall11

Would throwing a couple of 5v regulators in parallel work for your application? Something like 4 LM323's on a perf board (with appropriate filtering) should be enough for your application.

Edit: See below and apparently don't do this. Also I may have a soon-to-be-crispy board floating around and should actually learn to EE.

A single LM323 would be sufficient if you have a large enough heat sink. Putting multiple linear regulators in parallel will not give you more output current capacity than with just one. Increasing the output current capability of a linear voltage regulator is done by adding a large output device.

Originally Posted by GeeTwo

AAAAAAAAAAAAA!

Putting regulators in parallel is never a good idea unless you put enough resistance between each of the outputs to allow the variation in output voltage to go somewhere other than frying electronics.

The internal circuitry of all the linear regulators I have ever looked at are such that they can only source current and cannot sink current. Because of this, they cannot "fight" each other. The one that has the highest output voltage, by even a few millivolts, will end up supplying most or all of the output current. The other ones will sense the higher output voltage and shut off their output stage so that they supply little or no current to the load. In this mode of operation, they are not likely to be damaged.

The output isolating resistors will cause the output voltage to be lower than what you may want unless there is circuitry to compensate for them.

[sparkytwd]

I second the 9A Pololu converter. This is a newer model than what we used last year, and handles over-voltage better. Make sure you add a decently beefy capacitor on the output. For the further paranoid, you can add a reverse biased zener diode to short power and ground when you go over voltage.

Originally Posted by philso

Use this or one of the 12V to 5V Voltage Converters from previous years (check the output current capacity first).

Nothing in previous KoP's goes over 3 amps I believe.

Originally Posted by philso

The internal circuitry of all the linear regulators I have ever looked at are such that they can only source current and cannot sink current. Because of this, they cannot "fight" each other. The one that has the highest output voltage, by even a few millivolts, will end up supplying most or all of the output current. The other ones will sense the higher output voltage and shut off their output stage so that they supply little or no current to the load. In this mode of operation, they are not likely to be damaged.

The output isolating resistors will cause the output voltage to be lower than what you may want unless there is circuitry to compensate for them.

Correct, you get pretty vicious leader/follower cycles. There are some switching power supplies that do support paired regulators, but they're not very common or cheap.

What you can do is independently power a subset of the strips. Last year we had 10 meters of WS2812 broken up into 5 strips, and used 3 separate regulators. You can see the cable bundle here: http://hackcasual.io/images/bling_installed.jpg

[dellagd]

Originally Posted by FRC2501

We are using the same LED's and we are using the 12V-5V converter from the old control system connected to the PDP with a 20amp breaker. It outputs 5V-5amp's and the pixels will only use as much current as they need, but anything over 6V will fry the pixels.

Yep. Also using a strip of NeoPixels with the old converter and it works a treat.

[VeqIR]

We've used NeoPixels on our bots the last couple of years and found a couple of things that might be helpful to you. We used a the 5V/5A radio voltage regulator for powering the lights from last year and found that the dropout on those can be pretty high (we've measured >6V). On our bot last year, when the batteries were anything less than full, under load, the LED strip would display rainbow colors because the regulator couldn't keep 5V steady. This year, we tried a VRM (5V/2A) as suggested earlier and it was working great, especially since the buck/boost has a much lower drop out.

The Pololu solution was also one we were looking at, but we ended up going with some cheap UBECs from Amazon due to their low dropout and small size:

http://www.amazon.com/NEEWER%C2%AE-U...s=5v%2F5a+ubec

Another thing to mention is that make sure you include the resistor on the signal pin to the first LED. The signal line has some strange input characteristics and exhibits bad overshoot if you don't include the resistor to provide some simple filtering. We ended up frying the first LED in the strip that had already been taped inside of our arm and was an hour job to fix it.

Here's an interesting article on it:

http://www.eetimes.com/author.asp?do...&page_number=2