DIY solar powered rgb led strip

[erickarlop]

I want to make a solar powered rgb led strip. My thought is something like christmas light. I would like the lights to have effects like fading, blinking, and chasing.

What equipment do I need and what steps do I have to take?

I was recommended to use arduino and have started looking at that and would like to use that as it seems easy to use. The led would be RGB led so that I can change colors and stuff like that. I was thinking of having 2-3 independent "lines" which would make up one strip. For example, one line would be red and the other would look green. it would look like this: r-g-r-g-r-g-r-g with r meaning read and g meaning green. I dont know if it would be best to run each line in serial or parallel. Then I would like to provide power for the whole strip with a small solar panel. Sort of like the ones used on solar lawn decorations or solar motion sensing security light. Hopefully I wont need a big solar panel and a small one would be enough to power all the leds that I want. I'm thinking of having about 100 led per strip so 50 led per line. After this is done I might want to do a strip of 200 led. And finally, I would like for the light to turn on and off automatically at dawn and dusk. I don't know how practical it would be running so many led bulbs on small solar panels and am open to suggestions. If I have to make it smaller I would do that. I could see me adding this to the trim of the gutter on my house so I cant have a huge battery up there or be running wires that would be too visible

Is anyone willing to help me out with this project? I am a newbie and need lots of help.

 

[exdeath]

Youll need a line driver circuit (74LS541, STP16C596, TC4428, etc). A CMOS microcontroller IO pin won't sink the current for that many LEDs.

The solar circuit will be completely separate and trickle charge a battery, it won't be able to drive a MCU let alone 100s of LEDs directly from the solar panel. People think LEDs don't take any power to run, but enough of them together or bright enough will pull quite a bit of current that you won't be getting from a simple photovoltaic cell.

A small PIC chip would be perfect for something like that, and is the most popular MCU for POV LED clocks, flashers, etc.

 

[erickarlop]

I should have clarified when I mentioned solar. I plan on using a battery since I would like the lights to turn on from dusk to dawn. What I cannot do is having a big car battery or something like that as it won't fit in places where I want to put the LED lights. The biggest I could do would be about the size of batteries found in small UPS. I plan on probably using panel(s) from this http://www.harborfreight.com/interests/solar/36-led-solar-security-light-98085.html or something similar in size

What do you mean by the MCU cannot sink the current for so many LEDs? How many LEDs can a MCU power? I would like to stay with Arduino as PicKit seems more complicated than I can deal with for now.

I was talking to a friend of mine who has played with LEDs before and he mentioned that the RGB LEDs need 3 resisters which would change the current and light the correct light to change colors. Is this true? It would be a pain to add so many resisters even if I only did something like 20 LEDs. If that is the case then I will just go with single color LEDs and just buy different ones. Would it be easier to just buy a string of LED christmas lights, remove the power plug and hook it up to the MCU and still be able to do all the light effects?

 

[exdeath]

What do you mean by the MCU cannot sink the current for so many LEDs?

IO pins on CMOS MCUs are low level logic signals, they aren't capable of driving a large load, eg: lots of current. You need line drivers, motor controllers, injector drivers, etc for driving power into anything more than digital logic or simple LED that draws more than 25 mA or so.

On average a single simple small red LED draws 20 mA when forward biased and turned on.

Look up the datasheet for the MCU you plan to use for "max current sourced/sunk by IO pin"

Additionally there is also a rating for "max current sourced/sunk by IO port" which is the total max current for ALL the IO pins at the same time, before you think about splitting them up to multiple IO pins.

And an IO pin is 3.3v or less, you'll need much higher to drive a series string of LEDs.

You need an external line driver chip that takes the low level digital IO state and then acts as an "amplifier" to supply the power to actually drive the device.

Simple projects with a couple LEDs are ok, but an entire string of LEDs is going to be too much to connect directly to MCU IO logic pins.

 

[mindless1]

I want to make a solar powered rgb led strip. My thought is something like christmas light. I would like the lights to have effects like fading, blinking, and chasing.

What equipment do I need and what steps do I have to take?

While DIY often has its rewards such as learning things, a sense of accomplishment, or building something unique you can't buy anywhere, there is also much sense in recognizing that we as a species have achieved what we have because we build upon the work of those who came before us. In other words reinventing the wheel doesn't seem to make much sense in this case and where would it end?

For example you'd need a mode IC for the effects but if you're REALLY into DIY, won't you want to build that all discrete too? Let's take it a step further... you'd be using some transistors so why not build those from scratch too? It might seem far fetched but the same is true of every step in the process when you can get something ready made for any particular subsystem if not the whole thing ready to use.

You want at least 100 LEDs and they'd be driven to roughly 75mW for good output. 7.5W isn't a lot in the grand scheme of things but that's only the LEDs, not the resistor or other electronics, wire, or charging losses. Look at the Harbor Freight setup you linked. That solar cell is paired with a 6V, 600mAH battery, the entire capacity of it is 3.6VA, it couldn't even power your project for 2 hours. You can use a larger battery of course but the point is they're not giving away excess solar cell capacity for fun, you should plan that if you want this to run for only 5 hours every night you might need 3 or more panels which adds up in cost and size.

The amount of time and expense makes this project unworthwhile. Soldering 20 resistors is about the least time and expense of any part of the project... and solar power makes no sense unless you are off the grid, a string of 20mA LED lights uses so little current that you're not really even saving any power once you consider the computer time, room lighting to build this, manufacturing/delivery/etc of a battery and solar cell, disposal of these things some day when it stops working, etc. Also if it is cold outside the most cost effective battery would be SLA or the better performing in low drained apps would be NiCd, both of which have toxic metal, special disposal issues.

I am certainly not against DIY, just trying to suggest that your time is finite and this isn't worthy of it. We haven't even mentioned weatherproofing yet. This project could easily take a couple months to get from where you are now to finished and cost in excess of $100... or buy a $10 ready made plug-in set from a local store and do a different electronics project with more return on the time and money spent.

 

[Modelworks]

Fading or blinking an entire strand is easy to accomplish but chasing 100 LED on a home made strand will require you to wire up each LED separately meaning you will have over 180 wires going back to whatever you want to use to control it

 

[exdeath]

Fading or blinking an entire strand is easy to accomplish but chasing 100 LED on a home made strand will require you to wire up each LED separately meaning you will have over 180 wires going back to whatever you want to use to control it

Depends what you are doing. You could use a local latch at each LED and wire the whole thing as a 100 bit serial shift register. Surface mount 4014s for example could be hidden in the stand with ease. Time multiplex the register contents and rely on POV and you could generate pretty much anything. 16 Mhz MCU could refresh all 100 bits 160,000 times per second, and you only need a little as 30 hrz. Flicker will be less apparent on a moving display anyway.

4 wires to control the whole string: power, ground, clock, data. (or 3x data for rgb). Brightness controlled by duty cycle of each bit, and can be controlled directly by the MCU software (varying analog voltages with a digital MCU is much less straight forward)

It would also eliminate the need for a line driver since the 4014s would be powered and be able to drive 8 LEDs each directly from a voltage regulator instead of the IO pins.

 

[Modelworks]

Depends what you are doing.

It is easily done by someone with the knowledge, but when you are talking about someone like the OP who doesn't even know how to use something like an arduino you are way over their head. There are commercial versions that have a n addressable switch integrated into each led. The switches are developed by maxim and have up to 65535 possible addresses. You only need supply VCC, VSS, CLK, DATA wires to control up to 65K LED.

The newest chip from maxim is the Max6966 and would be great for LED control. Sells for under $3 each.

http://www.maxim-ic.com/datasheet/index.mvp/id/4539

2.25V to 3.6V Operation
I/O Ports Default to High-Z (LEDs Off) on Power-Up
I/O Port Inputs Are Overvoltage Protected to 7V
I/O Port Outputs Are 7V-Rated Open Drain
I/O Port Outputs Are 10mA or 20mA Constant-Current Static/PWM LED Drivers, or Open-Drain Logic Outputs
I/O Ports Support Hot Insertion
Individual 8-Bit PWM Intensity Control for Each LED
Any Output May Use or Not Use PWM Control
Exit Shutdown (Warm Start) with Simple Active-Low CS Pulse
Auto Ramp-Down into Shutdown
Auto Ramp-Up Out from Shutdown
0.8µA (typ), 2µA (max) Shutdown Current
Tiny 3mm x 3mm, 0.8mm High Thin QFN Package
-40°C to +125°C Temperature Range
Underwriters Laboratories (UL) Recognized

 

[Mark R]

The newest chip from maxim is the Max6966 and would be great for LED control. Sells for under $3 each.

Tiny 3mm x 3mm, 0.8mm High Thin QFN Package

Nice chip, but good luck soldering that! QFN is exceedingly difficult to hand-solder. For extra-small QFN like this (4 lands in the space of 1.5 mm), I wouldn't even know how to begin. (And I'm quite competent at soldering QFPs, SSOPs, 0603s and even 0405s if I'm feeling brave).

 

[exdeath]

Lots of flux, solder the entire side at once drawing the iron in a line, then go over it with wick. Lots of flux. Basically don't worry about bridges at that level.