sample circuit designs for rgb led strips

[Trent3]

I'm working with a hobbyist group and I want to create some simple circuits using different components and techniques to demonstrate using RGB LEDs. I'm using standard surface-mount rgb led strips like the 5050s that are operating at 12vdc (non-digital/individually-addressable) and I want to create some simple, cheap circuits that can make the LEDs cycle through red-green-blue fade up/down. Can anybody help me with some different circuit designs for something like this? I'd like to start off with something simple that is done "old school style" and also a demo of using some higher-level stuff like ICs. Any help would be most appreciated!

[mrkev]

It's not really that easy task if you wanna use ICs, the most used thing for this is uC.
Do you wanna make them cycle automaticaly, or is f.e. solution with three pots that would adjust the brighteness of each collor enough?

[Trent3]

I'm not necessarily interested in using ICs.   My end result is I want to make some small light fixtures that automatically cycle through fading up/down red/green/blue without intervention.   But I want to show a number of different, progressively more-sophisticated approaches towards doing this.  A simple circuit using pots would also be fun to demo.  But what I need most is something that will make a small string of these LEDs automatically change color.  I know there are individual LEDs you can buy that will also do this if you simply supply voltage, but I'm using this to show off electronic circuit theory/examples.

Any help would be appreciated.

1.  What's the simplest way to do this?  (obviously one way is with pots but what about automatic changing?)
2.  How many different ways are there to do this? 
3.  How does the cost/complexity of the components change over the different types of circuits?

Thanks!

[steve750]

Try this site for a colour changing circuit.
http://www.instructables.com/id/LED-Rainbow-RGB-LED-PWM-Controller-Construction-/

[mrkev]

Trent3: There isn't much that you can actually came up with. Basicaly for a small LED string, you can use either PWM or analog voltage diver. You will need some ramp or triangle generator in both methods and it would be really difficult to make it with discrete stuff only...

Steve: I think that's just some uC with three power transitors... Easiest way, but not really any fun...

[zapta]

Blinky Tape is open source hardware with repository on github.

http://blinkinlabs.com/blinkytape/

[Trent3]

Trent3: There isn't much that you can actually came up with. Basicaly for a small LED string, you can use either PWM or analog voltage diver. You will need some ramp or triangle generator in both methods and it would be really difficult to make it with discrete stuff only...

Steve: I think that's just some uC with three power transitors... Easiest way, but not really any fun...

Very interesting.. I will order a few of those kits to play with.. but at $18 each, I was hoping for something a little lower-cost.  It doesn't have to be as sophisticated.    Anybody have any other options?   I was hoping to get some kind of circuit in place for < $10 if possible.     I will look into this but also as part of the project I want an outline of all the different ways this can be done, with what features and at what cost... I'm hoping there are also some cheaper ways?  Thanks very much for this though!   Keep the great ideas coming.  I'll  be taking all of them into account.   What I want to do, is in the process of learning the best way, show how many different approaches can be taken.

[Thor-Arne]

Simplest way is to use a  micro-controller and some mosfets.

You can get clones Arduino Nano ($4) or Arduino Pro Mini ($2.5) on ebay.
These are small and convenient, remember you need a extra FTDI serial module for the Pro Mini.

The rest of the parts doesn't cost many cents

[mrkev]

I love how he wanted something and you all post the very same solution that is the exact opposite of what he wanted. 
And after all this nonsence he wrotes that he is gonna order some already done solution that has just simple microcontroler and three transistors. Just epic 

[Thor-Arne]

@mrkev: I was replying to the statement:

I will order a few of those kits to play with.. but at $18 each, I was hoping for something a little lower-cost.

Which I did answer with a practical solution to reduce cost.

Oh, and for the record I'm still thinking about this, so give me a break here.

[Trent3]

I appreciate everyone's suggestion, even the critical ones. 

I do like the packaged-controller thing, but I'd also like to "homerew" some circuits as well.

So please keep the suggestions coming!  When I get done with all this, I want to put together a demo to show off all the different ways this can be done, using both low and high-level solutions.

Thanks for everyones' replies - keep 'em coming!

[Thor-Arne]

To change the intensity on the leds it is best to use some sort of PWM control to avoid excessive heat, this also allow better control over the fading.

Designing something like this with only discrete semiconductors seems like quite a challenging task, at least for my skill level.

As I see it this contains of three parts:

• The driver, basically just a mosfet.
• The PWM generator, that must be able to modulate the duty cycle from 0 to 100% by external source.
• The modulation source, basically a very slow sine or triangle wave oscillator.

If you go for a IC circuit perhaps the 555 timer and/or some jellybean op-amps can do the job.

For now, this it the most helpful I can be. Perhaps someone else got a idea how to implement these steps.

[mrkev]

@Trent3 Do you wanna build all of them? And how long should the maximum strip be?

There are many ways how to do it, but as I wrote before, not really any easy ones. For very short strip (less than 20cm), its ok to use linear drivers (current mode is best since they react on voltage from cca 7V to 13V and each collor respond differently to voltage, but +- same to current).

My suggestions:
1.  http://www.elektrodilna.wz.cz/data/majak/majak-schema.jpg  will do the changing of collors, but it won't really make much shading. But I can't think of anything more simple and you could use this schematic directly (as the common pin is usually +12V).
2. You could use simple logic components. The biggest problem will be swapping to next chanel, ofc. if you don't wanna make everything three times... I made some drawing, it's not really great but I don't have time to make it any better
So first idea is to take 4b counter. You could use it to generate  kind of ramp signal. Four resistor at the output are actually I/U converter. As it counts up, there is more current going through R, 2R and 4R. This makes more voltage at the input of the MUX. Driver would be U/I converter, or you could use some U/PWM converter (with digital mux). The other counter is just for switching the collors.

3. You could use simple triangle generator. Fast one generates base for PWM, slow one makes the shading of signal (they should have time constants like <1Hz slow and 1kHz fast). You will need something to swich chanels, square output (f.e. falling edge) from slow one should give you the signal when to change.

4. The same thing could as 2. could be done with shift registors. The difference is that you get less shades (just 4 different stages). Transistors at the upper right corner (i draw bipolar, but they should be unipolar) generates first 1 stage, that is then shifted. through.

5. You could make ramp generators from three 555. Transistors with diode make current source that is charging cap. You could probably use falling edge of that ramp signal to swich another 555 (you may have to add another resistor from each 2 pin of 555 to ground). output can again go to pwm generator (simpliest is to use triangle gen. like in 3. and three comparators) or use linear current sorce as i draw.

I don't have time to build any of these so there will be some debuging, but I think that those concepts could work.

[Tac Eht Xilef]

I love how he wanted something and you all post the very same solution that is the exact opposite of what he wanted.

To be fair, what the OP asked for was pretty contradictory - "simple, cheap circuits that can make the LEDs cycle through red-green-blue fade up/down" "done 'old school style'", before leading into "higher-level stuff like ICs".

And, to be just as fair, they didn't know that at the time.

In their application, "simple, cheap" = uC, while "old school style" = complicated & more expensive, with "higher-level stuff like ICs" somewhere in between.

I like the general idea, and you could run with it several different ways to make various points:

• Low-frequency (< 1Hz) free-running transistor oscillators driving individual power transistors for each R, B, & G channel - difficult to control the pattern, & feel the heat (inefficiency)!
• PWM using 2 x transistor oscillators per channel - still difficult to control the pattern, more efficient, high component count & cost.
• PWM using IC oscillators / dividers - maybe lower component count, more accurate timing & controllable output colour pattern.
• PWM using uC - minimal component count, most controllable & flexible patterns, cheapest of the lot.

[Trent3]

I've got 555s and they're cheap so that would make a great component.  I'd be all for that.

By the way, I see what you all mean now.. I did not necessarily know that "the simple way" is actually more complicated and probably more expensive to produce...  or does someone disagree with this?

To change the intensity on the leds it is best to use some sort of PWM control to avoid excessive heat, this also allow better control over the fading.

Designing something like this with only discrete semiconductors seems like quite a challenging task, at least for my skill level.

As I see it this contains of three parts:

• The driver, basically just a mosfet.
• The PWM generator, that must be able to modulate the duty cycle from 0 to 100% by external source.
• The modulation source, basically a very slow sine or triangle wave oscillator.

If you go for a IC circuit perhaps the 555 timer and/or some jellybean op-amps can do the job.

For now, this it the most helpful I can be. Perhaps someone else got a idea how to implement these steps.

[Trent3]

@Trent3 Do you wanna build all of them? And how long should the maximum strip be?

I'd like to have a number of different circuit ideas to demo different approaches towards solving the problem.  If they are all fundamentally the same, then maybe I don't need different approaches?  But the idea is to reveal through example, some understanding of how the circuit works in a progressive fashion from low-tech to high-tech.

The strips would basically be anywhere from 15" to about 90".   one single trip to light a small panel that would be approximately 6"x18", to a larger item like a pinball translight/backglass that would be somewhere in the area of 24x36.     Probably the maximum power for the larger ones would be 40w of 5050 LEDs or something smaller.

There are many ways how to do it, but as I wrote before, not really any easy ones. For very short strip (less than 20cm), its ok to use linear drivers (current mode is best since they react on voltage from cca 7V to 13V and each collor respond differently to voltage, but +- same to current).

My suggestions:
1.  http://www.elektrodilna.wz.cz/data/majak/majak-schema.jpg  will do the changing of collors, but it won't really make much shading. But I can't think of anything more simple and you could use this schematic directly (as the common pin is usually +12V).
2. You could use simple logic components. The biggest problem will be swapping to next chanel, ofc. if you don't wanna make everything three times... I made some drawing, it's not really great but I don't have time to make it any better
So first idea is to take 4b counter. You could use it to generate  kind of ramp signal. Four resistor at the output are actually I/U converter. As it counts up, there is more current going through R, 2R and 4R. This makes more voltage at the input of the MUX. Driver would be U/I converter, or you could use some U/PWM converter (with digital mux). The other counter is just for switching the collors.

3. You could use simple triangle generator. Fast one generates base for PWM, slow one makes the shading of signal (they should have time constants like <1Hz slow and 1kHz fast). You will need something to swich chanels, square output (f.e. falling edge) from slow one should give you the signal when to change.

4. The same thing could as 2. could be done with shift registors. The difference is that you get less shades (just 4 different stages). Transistors at the upper right corner (i draw bipolar, but they should be unipolar) generates first 1 stage, that is then shifted. through.

5. You could make ramp generators from three 555. Transistors with diode make current source that is charging cap. You could probably use falling edge of that ramp signal to swich another 555 (you may have to add another resistor from each 2 pin of 555 to ground). output can again go to pwm generator (simpliest is to use triangle gen. like in 3. and three comparators) or use linear current sorce as i draw.

I don't have time to build any of these so there will be some debuging, but I think that those concepts could work.

Thanks very much for this, I will study your contributions.

Let me ask you, what do you think is the most cost-effective way to accomplish this? 

[Thor-Arne]

I've got 555s and they're cheap so that would make a great component.  I'd be all for that.

By the way, I see what you all mean now.. I did not necessarily know that "the simple way" is actually more complicated and probably more expensive to produce...  or does someone disagree with this?

If you use two 555's and set one up with resistors for both high and low pwm ratio (in series) and use the discharge pin of the second one to drive a transistor to "short out" one of the resistor, you'd basically get what you want.
Downside is the PWM frequency might not be particularly constant.

Yes, "the simple way" is the most complicated one to make, and it will probably be more expensive when you include the labor needed.

If you need to drive 40 watts of led-strip, I think you'll have a lot of difficulties dealing with the heat dissipated if you try to implement a linear fade.
IMO, the only option is to use PWM and drive some mosfets hard-on so it's not wasting the energy as heat.

[mrkev]

Let me ask you, what do you think is the most cost-effective way to accomplish this?

The most cost effective way is probably using few 555, in case if you are making just one prototype with short strip and you don't own a uC programmer. If you are making more than one, the uC with power fets is deffinitely less expencive.

The thing here is that if you wanna build your own, you will have to learn how to program that uC and also you may have to build/buy programmer. It's pointless if you wanna make just one project with it, but if you are gonna use it more than once, it will pay off... Over all, any of these solutions shouldn't be more expencive than 5$, but any old-school solution will have to be purpouselly build thus making just one function. And it won't be simple. With uC, you can change the programm and make any kind of collor mixing and changing of shades you'll want...