Luxeon Rebal Dimmer circuit for propeller?
OakGraphics
Posts: 202
Maybe this should go in a different discussion area, but I was looking for a way to make a circuit for the propeller to control many luxeon rebal rgb led stars.
While I was wandering over at do it yourself christmas.com I noticed they had developed a nifty 52 led 8mm 'straw hat' circuit that was basically RGB +W and they were using thier built controllers to dim the circuit.· I thought that would be cool to use for a lighting project I have been thinking of for a community church that is very low on available power.· They have some par-cans but no real 'color' available because they can barely keep the circuit up.· So I thought - maybe LEDs can trim the usage down.··· Further reading and it seems the luxeon rebal might be a good start at making a DMX controlled light thing, but that 'constant current' of 350ma threw me for a loop - circuit wise.· With Gadget Gangster now having a DMX shield available this seemed like the perfect time to ask! :-)
I saw the propeller 128-v2 (now called helix), and that looks interesting, but I don't think it does PWM / dimming of LEDS but I could be wrong.
(I often am. :-) )
While I was wandering over at do it yourself christmas.com I noticed they had developed a nifty 52 led 8mm 'straw hat' circuit that was basically RGB +W and they were using thier built controllers to dim the circuit.· I thought that would be cool to use for a lighting project I have been thinking of for a community church that is very low on available power.· They have some par-cans but no real 'color' available because they can barely keep the circuit up.· So I thought - maybe LEDs can trim the usage down.··· Further reading and it seems the luxeon rebal might be a good start at making a DMX controlled light thing, but that 'constant current' of 350ma threw me for a loop - circuit wise.· With Gadget Gangster now having a DMX shield available this seemed like the perfect time to ask! :-)
I saw the propeller 128-v2 (now called helix), and that looks interesting, but I don't think it does PWM / dimming of LEDS but I could be wrong.
(I often am. :-) )
Comments
Each rebel led is 350ma, and if you burn all 3 at full pwm, thats 350ma X 3. or approx 1 amp.
How many stars do you want? at 1 amp per star, it will add up fast, and also the 3 colors require different operating voltages,
now you need a switching power supply that can output 3 different voltages.
You could use a simple darlington transistor array to do the switching, but you need to figure out how to provide the correct voltages.
If you provide incorrect voltages, the leds will not only have improper color, but they will conduct current differently, potentially damaging them. (and they are not cheap)
Thats where messing with LED's gets expensive.
If you used an adjustable voltage regulator you will only be able to push 2 or 3 leds of the same color per regulator.
Its all going to add up fast, as far as total amperage required.
You could...
Supply all voltages with many lm317's (rated at minimum 1.5 amps) so you could run 4 leds from a single LM317
But that lm317 will get very hot due to the current of 1.4amps. Needs heat sink.
I would run only 3 per lm317 because heat is wasted electricity.
or a lm317 that can handle many amps + ( like the LM138 it can do 5 amps) 350ma X 14 leds = 4.9amp.
Then supply the lm317's with a computer power supplies 12v output. (most have 20+ amps available)
PC supplies are switching so using a non switching supply would cost you wasted electricity, plus pc power supplies are super cheap if you can find an old computer with a working supply. At least when compared to a specially made switcher...(that can to 20 amps)
After you get your voltages right for each color led, then you can feed the voltage thru the uln2003 (darling transistor array) you will need a few uln2003's also to stay in the limits of the uln2003's package.
Also use large 2 watt resistor's on your leds at first when programming it. Then when you have your PWM method down, lower the resistors (only do this if you can see the rise and fall of the LED's amperage using a scope.) Then you can see how long it takes to ramp up to 350ma, that would be your maximum ON time before potentially damaging the led(due to low resistor)
High output leds change the game compared to regular leds.
Oh and if you are burning these leds at 350ma,
you WILL need some kinda heat sink on that star, if you want your led's to last a reasonable lifetime.
(i know all this because I made a luxeon3 led array that has 36 - 3 watt leds. I was not concerned with power use or heat, so I just used a pc power supply, and the correct 5 watt resistors to knock the 5v supply down to the limits the leds could handle. But the 5 watt resistors consume electricity and produce heat.
My luxeon 3 leds are pushed at around 900ma each.
With the resistors and the leds I pull about 35 amps from my pc power supplies 5v line.
I could save tons of electricity, and heat and perhaps even the leds lifetime by giving up the resistors for pwm + lower resistors.
I just might do this now that you reminded me of it.
And I will most likely do it the same way(LM138's)(darlington's)(pwm), because I have no intent on making a power supply.
One other thing, you can series the leds so they can take higher voltage, but you are then limited to PWMing that entire string of leds, doing that will remove your ability to manipulate single leds.
If I wanted to really get fancy with my 3 watt array, I would put a cheap temp chip on each led's heat sink, and then read those temps with some kinda rctime on the prop. Then you could PWM more or less depending on the temp.
(temperature affects led color and led forward voltage, which then changes the current needed. (the rebel and luxeon datasheets show the temperature graphs) wavelength, lumen output, forward voltage, and current needed all depending on your LED JUNCTION TEMPERATURE.)
Post Edited (Clock Loop) : 12/4/2009 3:35:43 PM GMT
The more I look at the DMX IO - the more I like it!
I was looking at doing something like 7 ~ 10 RGB stars for my final project, but this will help me get off the ground and understand things a bit I figure. Right now putting 3 stars together would be a pretty good test, but getting 1 star to actually function will be a good start.
They are VERY bright, the rebel line is replacing the luxeon 3-watt line.
The luxeon 3 line pulls 750ma.
The rebel line pulls 350 ma
So the rebels will be 1/2 the brightness of luxeon 3.
I CANNOT directly look at a luxeon 3 for more than 2 seconds, and no one probably should.
Like johnny mac says, the dmx IO controller will work for a single star, plus 1 extra I/O (if you buy 3 dmx IO's you can use the 1 extra IO on each one to conntect to another star.
So if you get 3 DMX I/O's you can run 4 starts.
But you are still tasked with dealing with proper voltages, and using the DMX I/O DOES NOT allow you to use different voltages for each color. (unless you organize the leds so that all the same color use the dame DMX I/O)
Also you need to provide the dmx i/o with the proper REGULATED power, it doesn't provide any on-board power management.
The rebel datasheet:
http://www.philipslumileds.com/pdfs/DS65.pdf
states that at 350ma:
Volt = -Min - typ - max
Red = 2.31 - 2.9 - 3.51
Blue = 2.55 - 3.15 - 3.99
Green = 2.55 - 3.15 - 3.99
So actually the voltages are very close, and you could probably get away with powering them all at 3v.
(perhaps even 3.3v, but your pushin it, combine the extra voltage with heat and your current will rise, making the rebels brighter, burn hotter, and lower the lifespan.)
The data sheet says that at 3.3v the rebel stars will pull 500ma NOT 350ma each color. You see how a small amount of voltage allowed the current to rise considerably. Thats why I say, its IMPORTANT to get proper voltages with these stars.
So when adding on 0.3v you increase your current requirements to 1.5 AMP per star instead of 1 amp per star.
So if you can find a 3v regulator that can handle 7-10 amp then your set. (or use one 3v regulator per dmx i/o = 350ma x 4 = 1.4 amp)
Just find a 5v supply that can give 12 amps for the regulators. (pc power supply)
Keep in mind those voltages are if the temp stays low, so use heatsinks on those stars.
Also it might be important to note how big the trace is that goes from the Vin on the DMX I/O to the 4 - TIP125's. If it can't take 1.5 amp then it might be wise to solder in some jumper wires to help handle the current.
Good luck!
Post Edited (Clock Loop) : 12/2/2009 9:05:25 AM GMT
Thanks for the great info - it is appreciated. [noparse]:)[/noparse]
Yeah - I was wondering about the 'current' issue so I did some browsing and found this instructable that demonstrated creating a feedback loop for the transistor so the current is regulated. It seems interesting www.instructables.com/id/EK2XAPS11GEWOF2YSD/
It seems like I could BAM or PWM the circuit with the added ability of not worrying about current load if I attach R1 to the Propeller pin instead of V+. (my 2-second analysis )
Okay - so 3V it is.
I think the end-result will have to have a PCB made, so I will use a 2oz copper pour and nice thick traces to the terminals.
I am also looking at having the stars mounted on an aluminum plate stock that will fit into a par64 can. I haven't done the drawing yet so I don't know if I can fit it in a smaller can just yet.
You should definitely operate them off current limited supplies otherwise you are asking for trouble. The supply will lower the voltage as appropriate to ensure they do not draw too much current. With a sufficiently high voltage power supply you can put multiple LEDs in series and current limit them all together, but only if you wish to switch them together. I do this for two arrays of 6 red K2 Leds that I use for capturing high speed videos of insects in flight.
If the board is not able to switch the currents then just use it to switch mosfets that can perhaps.
Cheers,
Graham
I am not familiar with the output capabilities of the DMX unit and I'm not sure how he intends to wire up the LEDs but he may be exceeding the rated output which is why I suggested a solution just in case.
I will also mention that with many of these high power LEDs you can lower the current from the rated without a massive loss of light output but with quite a decrease in heating, often you don't really need all they can give out and it pays to run them a little cooler if you can.
Graham
I like it except for this:
R3 will cause a ton of heat and use power to do it, total waste because all current the LED(s) use passes through R3.
YOU only need a VERY VERY SMALL RESISTOR.! (because your using the correct voltage + pwm.) (if you were using a 5v supply to power the rebels, you would need a large R3. (and r3 would need to be pretty big 2W to 5W, because it has to take 2v @ 350ma and discard all that electricity as HEAT. WASTE OF POWER!!!! (you mentioned that your limited on total power)
NOR DO YOU NEED THE FEEDBACK CIRCUIT. Just lower the Pulse Width to lower the current.
Yes connect R1 to the prop output.
You should also put a resistor BEFORE R1 that pulls the props output low just in case funny things happen to the prop. (high z state)
I guessing Q2 turns ON when the gate is high.
(but look up Q2's gate current needs, and if its more than 20ma(doubt it), then you can't connect the prop directly to Q2's gate, you will need to do something similar to what the DMX i/o does, which is use a to-92 transistor to push the gate of Q2)
When you FIRST START TO PROGRAM THE PROP, USE a large R3 (if you can't calculate it, let me know)
This is just in case you hit bugs in your software that cause Q2 to stay on for longer than your maximum PWM time.
But with a large R3 in place the rise time of the current will be different than when using a small R3 so thats why I say get a scope on that led, so you can see the rise time with a small R3 to determine your max pwm for that color. (pwm will vary between colors due to forward voltage drop differences in the rebels.)
The only reason to use a large R3 is if any rebel led were to go crazy on you(short) then Q2 could possibly blow also. (not likely, it might just use tons of power from your supply and heat up.)
You will need an o-scope to see the current rise to 350ma, once it hits, that is your ceiling, turn off that channel. And make note of the time for the color and channel. You will have small timing variations due to differences in Q2, and the rebel's construction. (and also heat)
So make sure you tweak the time the max pwm is after you let the led's heat up for a few minutes at the max pwm.
(if you ever let your Pulse Width be longer than that MAXIMUM window, your led will start to heat up and fry.
The dmx I/o is kinda overkill if you don't need DMX input/outputs.
The stars will thank you. You can calculate the surface area needed on that aluminum heat sink using luxeons datasheets on thermal management of the rebel leds.
But remember you have 3 rebels per star, so you will need 3x the surface area on that heat sink.
If you can't figure out how to calculate the min surface area on your heatsink for all 3 rebels, just let me know and i'll look into it.
The datasheet on both will tell you that you are mistaken.
The Lux output of the STRONGEST Rebel(red) is around 60lm.
The Lux output of the STRONGEST Luxeon3 (red) is around 140lm.
Perhaps you saw the strongest rebel, and the weakest luxeon3.
(the reason different lumen outputs are available for each color is because of binning.)
Much like processors that were on the very outside of the DIE,
The lower lumen LED's are sold cheaper because they were the chips that were not in the very center of the die during manufacturing.
(where the focus of the lithography is perfect, and diminishes as you move out from the center.)
If he wants each can to have the full range of the rainbow, he cannot series the leds.
I agree 100%, but thats why you PWM the thing.
Just make the Pulse a bit shorter and bam, you now have lower current running through the led, thus lower temps and less lumens.
You can do this, but you need 2-5watt resistors. They are huge. WHY? They must take all that extra power the led can't use and pull it out of the circuit in the form of heat, and lots of it.
You can see what I am talking about when you look at LED based Christmas lights.
They pwm. (and you can see it... they "ever so slightly" flicker when compared to a regular Christmas light.)
Post Edited (Clock Loop) : 12/4/2009 3:28:56 PM GMT
My prototypes are too "snobby" to share space with pcb's.
(see attachment)
Dead Bug style! Cool! What is it? It looks like a 12 channel controller to something. 4 props?
I could not follow all of what you wrote but the idea of simply setting the voltage is not good, if they rise in temperature current will go up beyond the rated, at the very least you will have no proper control over it. The circuit that Oakgraphics linked to is much better than what you are suggesting. Better still is to use a current limiter based on switching. Of course the less voltage dropping across the limiting element the better, if the LED takes nominal 3v and you supply 3.3v say then you are only talking a 0.3v drop whereas a single LED running from 20v would be bad.
I don't know which datasheets you are looking at but I cannot see the numbers you are quoting on Philips website for the Luxeon and Rebel LEDs.
He could series all the reds together and all the greens together and all the blues together and still have a full range of colours although all would be the same.
The wattage of current limiting resistor depends on the voltage drop required, if the supply is close to the LED forward voltage then you will not be throwing away much power (no need for 5W resistors) but that said I am not a fan of the solution either for high power LEDs.
And I think you should buy some veroboard :P [noparse];)[/noparse]
Graham
Its nice, but does use more power due to more parts that are not needed, the large current sense resistor consumes current which generates un-needed heat, and wastes electricity. Try them both and keep what you like.
I design to be as efficient as possible. Electricity is expensive, and additional ambient heat(from resistor) adds to LED heat.
Uhh yea,,, one would call that PWM. A switcher will turn on and turn off. Depending on current needed, the switcher turns on longer or shorter... this is also known as PWM. . Because the supply is a 3v regulated supply, the only issue is management of the current through the LED. And that again, is done with PWM.
I wrote that basically, if you provide the correct LED voltage, a VERY VERY SMALL RESISTOR is needed.
The LED will take X amount of time to ramp up to 350ma. At which point you turn it off.
Then you can turn it back on after a small "off" wait time and repeat..... aka... pwm.
The pwm length is on the order of milliseconds, but thats normal when your doing PWM to control current.
If I had a scope, I would show you.
I don't know which datasheets YOU are quoting, mine are shown below.
You DO realize that the phillips sells every led in different "quality" chips. The quality of the chip makes its operating efficiency better, thus increasing lumen output even though the same current is used.
Lower quality chips still use the same current, but they are located around the outside of the die, where the lithography quality is less than perfect. Sure the chips still work, but they work with much less efficiency.
(this is why you will see there are many colors that have a few "versions" that state different lumen output for the SAME CURRENT.
I'll pay the extra money to get a higher quality rebel that puts out more lumen using the same amount of current as a lower cost rebel of the same color.
LOOK AT THIS PAGE.
http://www.luxeonstar.com/luxeon-rebels-red-rebels-c-142_143.php
Ever wonder why the two are different in lumens, but consume the same amount of current? (and notice the price)
This is called "binning". The higher lumen version is exactly the same chip as the lower lumen version but the quality of the chip is lower in the lower lumen. (high end graphics/cpu chips do this also, the core generation is the same on many different graphics cards even tho those graphics cards all state different core frequencies. The companies just test the chip for defects and then underclocks the ones that fail at higher frequencies, and sells them as "lower end" graphics cards. I have an ATI radeon x850 pro, which boasts 520mhz core with 12 pixel piplines. The ati radeon X850 XT Platinum Edition's core runs at 540mhz with 16 pixel piplines. But both graphics cards have EXACTLY the same IC design. Known as "chip R480" The x850 pro card's core chip was not at the highest quality point on the DIE after lithography.
The luxeon III: (LXHLPD09) not the luxeon. (this would be a chip that was near the center of the die after lithography)
Page 4 on http://www.philipslumileds.com/pdfs/DS45.pdf
Typical lumen flux for RED is 140lm
The red rebel: (LXML-PD01-0040) (this would be a chip that was near the center of the die after lithography)
Page 4. http://www.philipslumileds.com/pdfs/DS65.pdf
At 700ma, the flux on the red rebel is 80lm.
Naw, I like my dead bugs.
Also makes it easier to recycle prototypes, repair, modify....
Post Edited (Clock Loop) : 12/4/2009 3:24:33 PM GMT
Interesting to see the insides.· I like your idea of 'light' switches.
Post Edited (OakGraphics) : 12/4/2009 2:57:14 AM GMT
The sequencer is also known as the BlackBox.
http://forums.parallax.com/showthread.php?p=831833
Link to project page with code and schematics and video and pictures.
Thanks, this idea was proven by Forrest M. Mims III.
(http://www.forrestmims.org/biography.html)
He discussed the concept in his very popular book that radio shack published.
"Getting Started in Electronics by Forrest M. Mims III"
OakGraphics, you might like this, To allow the detection of light using a LED, one has to be able to "read" the led with VERY LITTLE resistance. (so you can't use the correct current limiting resistor)
What your forced to do, is basically use a VERY VERY SMALL RESISTOR.
This is ok when "reading" the led for light.
But when it comes to lighting the led up, you would normally blow the led due to a VERY VERY SMALL RESISTOR.
But it works because I PWM the led.
(a.k.a, only let the led burn for X amount of time before I shut it off (again, on the order of milliseconds)
And this is exactly how I told you to light your rebels.
If your PWM pulse is short enough, you can safely light the LED with a VERY VERY SMALL RESISTOR.
(but with 350ma, the length of your ON time is very small)
Post Edited (Clock Loop) : 12/4/2009 3:14:30 PM GMT
You can believe what you like, but electronics is not a faith system.· A current limiting resistor is required·or the limiting as such·is, de facto,·that of the impedance of your supply and its inability (mercy·=·undeserved forgiveness) to provide unlimited current, the resistance of the wires connected, the diode's own impedance, and other sources of imperfection in the circuit and that's all that stands between your contentions and a smoldering heap.
Cmon dood, faith system? Believe what I like? LOL.
Fine, everyone use a very small resistor on your led's, or you will make the circuit police mad.
(i agree, most high current devices can't switch fast enough to allow no resistor, but its pretty darn close.)
You could try to help educate me, and other people here on why I am wrong, instead of suggesting that my lack of knowledge is somehow because I have faith or a belief in some circuit god.
I suppose if I had a scope I could "correct" myself by just trying it.
So, lesson to us all, get a damn scope.
I'll go back and change every part that said "NO RESISTOR IS NEEDED" to a VERY VERY SMALL RESISTOR is needed.
But I can't tell you how small because I don't have a scope to see the rebels profile.
Compared to the size that is needed when not using pwm, its on the order of magnitudes smaller.
I'll do you one better -- Please tell me how you calculate current using the means you've advocated, an oscilloscope is NOT essential,·and then I will take everyone to school.
I lick the circuit, turn around 3 times, smell my thumb, and then ....
Oh never mind, I fixed my jacked up suggestions on using no resisisisisisisisisisitor.
Just educate us already.
PWM can be used to control current but that is not what you have been advocating, rather you suggest setting a PWM level and then just leaving it. If the device resistance goes down current will go up and you can have problems such as if overheating occurs. Ambient temperatures of course affect this too.
You do not have to have a resistor but you should have some means of current limiting ideally, heat sinking will help but it is bad practise to just hope for the best at least with such high power hot running devices. You could of course just pulse them for 10us once a week and even I would not suggest that a resistor would be needed then but this approach even if watered down will reduce the powers you can expect to see. If for a specific application that works out then fine but as a general catch all solution I'm not so sure.
I was confused by the numbers you quoted for the Lumileds because I genuinely did not see them, it turns out I was looking at the min not typical values for the III. It is actually quoted for 1.4A drive current for the RED LED, for many of the other colours (like for like comparisons with current and colour) the Rebels are brighter than the IIIs. I didn't really appreciate the manner of your response, I don't like being talked to like an idiot. I think my tone was as neutral as it could be, sticking to the facts.
Except for the veroboard comment [noparse]:)[/noparse] But really "easier to recycle prototypes, repair, modify..." looks like a nightmare to me but whatever works and each to their own of course! I also do dead bug sometimes though usually to save weight in things that fly.
Regards,
Graham
The internets suck for communications.
Perhaps you thought that when I said "determine the max pwm to ramp up to 350ma..."
Thats nothing more than the MAXIMUM pwm. Anything beyond it would subject the led to damaging current.
Anything less just makes the led not as bright.
I HAVE been advocating using pwm to control current.
The idea behind pwm IS to adjust it.
Pulse Width Modulation.
Pulse Width : The amount of time a signal on a wire stays high or low.
Modulation : Changing one waveform in relation to another waveform.
I did mention about getting a RAINBOW of color.
To get a rainbow of color from R G B leds, you need to pwm each color differently to have them all mesh and make a color from the rainbow.
If one uses a single pulse width, and doesn't change it, its not called pwm, its just a pulse of "x length" at that point.
If the pwm was set to a single constant, the brightness of each led would be constant,
thus creating a single color, depending on the pwm that each led was set to.
Even before I replied to this thread, I knew what he was trying to do.
He is trying to replace high output "stage" lights with leds.
Stage lights usually have many or a single color available, and depending on what "cans" you have on, and how bright, will determine what everything on stage gets lit with.
His goal being full color output from a single "can".
I didn't mean to offend you, I just KNEW from prior experience that the rebel leds have less lumen output.
But that depends on the color as I now see, ... and you too.
And it just so happens that my design used red only, so I didn't even look at the other colors.
We are both wrong, and right.
Smaller package, lower current, and higher lumen output.
Im sold. The rebel line is better, and probably why phillips discontinued the luxeon3 line.
Post Edited (Clock Loop) : 12/4/2009 10:48:01 PM GMT
· If you don't have a current limiting resistor, as I stated before,·then the effective resistance and limiting factor is the sum of the wire and solder, the diode's resistance (impedance) and other factors, all of which add up to be very little (not enough.)· And that is simply not desirable.
· With PWM, you can/do dim the LED output, but the current going through it remains the same, the same as with continuous current only it's pulsed.· The PWM current does not get divided by a factor of the Duty Rate (No.)
· Please see attached dwg.
I was thinking he might use PWM for brightness control and a current limit to prevent any problems, setting the supply voltage close to the LED's forward voltage minimises the power lost.
I only mentioned the Rebels being bright because for many of the colours they are but sure for RED you can get a hell of a lot more out of a III. I am similarly disappointed about the loss of the K2 leds.
Graham
I think the point is that as a given PWM point with no resistor or current limit you get a certain current but that might change in ways that cannot be controlled as it depends on the device properties alone.
Graham
It talks about current spikes.
But in this case it is using a current regulated supply.
In our case we use a voltage regulated supply.
www.philipslumileds.com/pdfs/NA04.pdf
Post Edited (Clock Loop) : 12/4/2009 11:12:33 PM GMT
I think that you and I are not in fundamental disagreement (I hope.)·
One can pulse at higher currents, but that was not·a point of contention.· The crux of the discussion is/was unsafe operation: no current limiting other than the ineffable (aforementioned, twice already.)· If somebody has been doing so then he can thank current-foldback for saving his bacon.
All,
And, right,·operation from a·current-regulator (NA04.pdf) is resistorless because the current-source is the "resistance", but, in higher-power·situations this·brings unique concerns, as the article declares.·
Note their use of Ohm's Law there just as I had.
Graham