OC & Modding DIY LED Dimmer control - Need suggestions

[cranky]

The resistor is only to drop voltage *if* Vs is higher than Vf. It is not mandatory - in actual fact it is a current-limiting device, not a voltage control device. It only prevents you from frying the LED with too much current (*not* voltage).

All semis are not fixed loads - they are 'active' devices and behave differently at different voltages and currents, from not functioning at all to being destroyed. 'Passive' components like resistors (or lightbulbs) have linear V/I curves, and obey Ohm's law faithfully.

The junction forward voltage is the minimum voltage required to operate the LED, that is correct. If you look at the V/I curves for a typical semiconductor it will become clear:

LED center How is LED brightness related to current?

This is why PWM is probably the best way of controlling a LED, better even than current control. Since you are varying duty cycle (above 60Hz ideally 200+Hz so the eye does not detect flicker) the control over brightness is much better most PWM controllers use 400Hz as the control frequency, so you should be fine with one of those.

 

[Gannu]

@cranky, I dont even want to point out the other issues with your post! But I like the way you mix some relevant facts with BS to make your posts look credible.

Raghu, please do away with the mudslinging part from the forums. I have seen this before and do not appreciate the way it goes about - not at least in our forums. Very well, I do not know what has been posted in the thread but least of all let us ensure that a healthy debate progresses. Instead of stating 'mix relevant facts with BS' and all - totally uncalled for IMO. Some things work out of intuition and practice and may not warrant all the theory we have learnt in our graduation by far - I'm sure you know of it already.

No offense meant.

 

[greenhorn]

Wrong! LED brightness varies almost linearly with the current (all the way from ~0 to max. current). Check any LED datasheet.

in terms on luminous intensity, yes. but visually, doubling the current hardly doubles the 'brightness'

Wrong again!! Resistors dont damp out the control/voltage. This circuit will work very well indeed.

@cranky, I dont even want to point out the other issues with your post! But I like the way you mix some relevant facts with BS to make your posts look credible.

I dont find any problems with his post. its quite correct technically. I'm not sure what you think is BS. I'd say you're the one mixing BS with your posts...

regarding the point about resistors damping out the voltage, his point was that, if you try to use a voltage regulator to control the voltage across the LED, adding a resistor in series would result in the voltage changes being dropped across the resistor, with the voltage across the LED being constant.

without the resistor, and if you ensured that the voltages were safe, you actually had a shot of forcing voltages across the LED.

with a resistor now in series, any increases in voltage will be dropped across it instead.

 

[PhOeNiX]

Just talked to cranky and everything is clear he made me go back to my Semiconductor basics though

Well this is the plan :

1. Buy 4 LED's of the colour which i want to use in the end (mostly blue, so around 2.6V each)

2. Connect them all in series and try with the fan controller if it is able to control the brightness sufficiently. Will add a 100ohm 1W resistor to give it sufficient Voltage drop.

3. If it is, then will cut off the cables of the LED's on the fans, connect them in series with a resistor, and then connect each [4xLED+Resistor in Series] set of each fan in parallel and run the single channel to the FC. Will add some shrink wrap etc to keep the cable connections looking nice

Thanks to Cranky, greenhorn, NinByChoice Will keep this post updated with my progress

 

[cranky]

@cranky, I dont even want to point out the other issues with your post! But I like the way you mix some relevant facts with BS to make your posts look credible.

Well, thank you seems like you're back in the word game again. Maybe we should just put each other on our respective ignore lists. Which is what I intend to do after this post.

Let me return the favour...

1) Resistor in series - requires 1 fixed resistor and 1 variable resistor.

Variable resistor = 9V/3mA = 3 kOhms = ~3.3k pot should work well.

First, nobody makes a 3.3K pot (trimmers maybe, but not pots). Please point it out if someone does - it only exists to those who know only theory. Most small 5K pots will simply not work with the ~100mA of current this application calls for (20LEDs, assuming 5 chains in || at 20mA max per channel). A 5W pot is required, and even that will run warm. Very.

Second, 3mA is not enough to show *any* visible light from an LED. Maybe you should actually set up a working circuit for once, add an LEd and a resistor, and take some photographs like the nice man in the link above, who actually tries things out and then shoots hi mouth off (or doesn't).

Third, visual perception of light is does not exhibit a linear function with luminescence output (mind you, this is a different emission phenomenon than light radiation which happens from light bulbs, fluorescent lamps and the sun - so perception of light is also very different).

2) Voltage regulator (like the 7805 circuit you had earlier or as a CCS) - 1 IC, few resistors, capacitors.

Quite simple and easy to build. Easy to find ICs too. Works great, almost linear brightness response. Can be designed to work as a constant current source if you want. Will work out the designs if you need it.

Maybe you should be called out on this one. I'd like to see some pics like the link I posted, installed on some LED fans in a PC case, and working with this nice voltage control circuit which you will 'design' after a schematic is posted on the thread

3) PWM - 1 IC, few resistors, capacitors, maybe 1 MOSFET switch.

The most efficient circuit of the 3 designs. Works very well. Might be a little trouble finding a good IC which integrates the switch too. Otherwise, can always use a 555 timer and a MOSFET. Will post the design if you want.

LED current driver ICs are available, but prolly end up costing more or harder to get.

The man already has one of these in a fan controller, or are you only reading my posts?

Look we've had this altercation before. The fact is that you are simply misreading everything I write - I may have missed *one* word and that does not make my post BS, any more than your hypothetical (as usual) 3.3K pot. I suggest you start ignoring what I write. It will help you come forth with your own perspective without needing to look for targets, and as a result you may develop a little less bile and a little more thought. My sentence should probably have read "Once it hits 6-6.5mA it will work at 80-85% of full perceived brightness". And that would be it.

In the meantime, Sudhir has his solution and will post back on the thread.

I suggest you buy a fan controller and wire the fan with a small series resistor (about 20 ohms should do it, I used 22ohms, the closest value I could find)) in the positive supply line, and report back on RPM measurements. The thing is, I've actually tried it (good heavens, sacrilege!!!). The controller simply goes crazy, and can only switch the hitherto working fan on or off, removing the resistor simply makes it work again. With a voltage fan control source using series pass, the resistor introduces no anomalies.

Good Night, all!

 

[NinByChoice]

in terms on luminous intensity, yes. but visually, doubling the current hardly doubles the 'brightness'

Having used a lot of LEDs in the past, I would actually say its quite smooth and not as drastic as cranky made it out to be. The difference in brightness between 30mA and 6.5mA is a lot!

regarding the point about resistors damping out the voltage, his point was that, if you try to use a voltage regulator to control the voltage across the LED, adding a resistor in series would result in the voltage changes being dropped across the resistor, with the voltage across the LED being constant.

Ok, let me explain this. A voltage regulator isnt trying to control the voltage across the LED (which remains a constant as cranky has correctly pointed out), but the voltage across the LED+Resistor. The increase in voltage would be lost across the resistor, like you said. Now, heres where you overlook the crucial point.

The resistor would now be carrying a higher current because it has a higher voltage across it!! This same current flows through the LED too, since they are in series. With higher current, the LED will glow brighter indeed.

Well, thank you seems like you're back in the word game again. Maybe we should just put each other on our respective ignore lists. Which is what I intend to do after this post.

I wont do that, nothing personal anyway. Some of your posts are informative. Besides, I dont know when I might be wrong and you might be the one to correct me.

First, nobody makes a 3.3K pot (trimmers maybe, but not pots). Please point it out if someone does - it only exists to those who know only theory. Most small 5K pots will simply not work with the ~100mA of current this application calls for (20LEDs, assuming 5 chains in || at 20mA max per channel). A 5W pot is required, and even that will run warm. Very.

Ofcourse! The point wasnt the 3.3k or the 300 Ohm resistor. For that matter a 4.7k would work well too, at a small loss in the adjustment accuracy.

With 20 LEDs and a current of 100mA, the design values will change. It wont be 5kOhms anymore, will be much lower infact. Redo the math to get to the new value. I dont think we will need a 5W pot in either case.

Heres a 3.5k pot though. Please ignore the cost on Farnell, they are extremely inflated.

HONEYWELL S&C / CLAROSTAT|JA1L040S352UC|POT, CARBON, 3.5KOHM, 10%, 5W | Farnell India

Second, 3mA is not enough to show *any* visible light from an LED.

That was a design example. If 3mA doesnt cut it, then redo the math with 5mA. Thanks for letting me know tho, will verify it when I get a chance.

Maybe you should be called out on this one. I'd like to see some pics like the link I posted, installed on some LED fans in a PC case, and working with this nice voltage control circuit which you will 'design' after a schematic is posted on the thread

Please sponsor this!! I will be glad to build the circuit over a weekend and prove you wrong

Will post the circuit in a while, though its similar to the one in Post #9.

I suggest you buy a fan controller and wire the fan with a small series resistor (about 20 ohms should do it, I used 22ohms, the closest value I could find)) in the positive supply line, and report back on RPM measurements. The thing is, I've actually tried it (good heavens, sacrilege!!!). The controller simply goes crazy, and can only switch the hitherto working fan on or off, removing the resistor simply makes it work again. With a voltage fan control source using series pass, the resistor introduces no anomalies.

Yes, thats because the fan needs a certain voltage to startup (say ~6V, for the 12V fan). Also the impedance of the fan (say 0.5A rating = 24 Ohm) is similar to the 22 Ohm you connected. So, with the 22 Ohm in series, the fan might not have been getting enough voltage to start up, except when the controller is set to the full speed. Hence you felt that the controller was working like a switch. Maybe you should have calculated your fan impedance and noticed it being similar to the resistor value. The controller doesnt go "crazy", it just works as explained by theory.

1. Buy 4 LED's of the colour which i want to use in the end (mostly blue, so around 2.6V each)
2. Connect them all in series and try with the fan controller if it is able to control the brightness sufficiently. Will add a 100ohm 1W resistor to give it sufficient Voltage drop.
3. If it is, then will cut off the cables of the LED's on the fans, connect them in series with a resistor, and then connect each [4xLED+Resistor in Series] set of each fan in parallel and run the single channel to the FC. Will add some shrink wrap etc to keep the cable connections looking nice

Sounds like a good plan. The max current is only ~16mA through the 100 Ohm resistor, so you wont need a 1W rating. A 0.25W, which is the regular size, will work just as well

 

[greenhorn]

Having used a lot of LEDs in the past, I would actually say its quite smooth and not as drastic as cranky made it out to be. The difference in brightness between 30mA and 6.5mA is a lot!

Um, I've used a lot of LED's in the past, and over the past year or so. I've been working on a pet project ( you could read about it here )

I've been trying to find the sweet spot for current/light output, and the difference isnt *that* much. noticeable, but not a *lot* We've had more usable light output running multiple LED's with low current in parallel than running higher current over a single set of LED's ( and you'll see most users in the thread agree)

Ok, let me explain this. A voltage regulator isnt trying to control the voltage across the LED (which remains a constant as cranky has correctly pointed out),

It is when its set up as a constant current source, and the LED is directly connected to the outputs.

The resistor would now be carrying a higher current because it has a higher voltage across it!! This same current flows through the LED too, since they are in series. With higher current, the LED will glow brighter indeed.

of course, but like you said earlier, the voltage across the LED in this config remains pretty much the same.

@ cranky. 3mA is plenty to drive an LED. I've designed RED LED's to run off just 1mA & they ran just fine

 

[NinByChoice]

and the difference isnt *that* much. noticeable, but not a *lot*

Well, I was assuming that 5mA is the point where the LED becomes noticeably bright. Do you get a drastic change from 5mA (off point) to 6.5mA (30% from off point)? Or if it becomes visible at 1mA, do you notice a drastic change at 1.3mA itself? My point was that the brightness doesnt change drastically over such a short range. If it were so, it would be a pain controlling the brightness with a knob.

It is when its set up as a constant current source, and the LED is directly connected to the outputs.

The regulator produces whatever voltage is needed to maintain the set current. The voltage appears to be almost constant, but is infact determined more accurately by the slope of the V-I curve of the LED (which is still finite for real LEDs). For hand calculations, this slope is considered to be infinity and hence the constant Vf.

of course, but like you said earlier, the voltage across the LED in this config remains pretty much the same.

Yes, its the current thats varying!! Thats the whole point!! The LED intensity is dependent on the current, not the voltage. The voltage is a constant, well almost.

Ok, let me work out the current/brightness in each case.

Circuit: Voltage regulator is connected to 1k resistor+2V LED in series. Voltage regulator can supply from 3V to 12V.

Case 1: Voltage regulator is supplying 3V to the circuit.

Voltage across diode = 2V

Voltage across resistor = 1V

Current through resistor = 1V/1k = 1mA = Current through LED.

LED glows with 1mA brightness.

Case 2: Now, voltage regulator is supplying full 12V to the circuit. The extra 9V over Case 1 is now across the resistor (The diode voltage remains constant).

Voltage across diode = 2V

Hence, voltage across resistor = 10V (See that all the 9V is added to the resistor voltage. This influences the current through the LED)

Current through resistor = 10V/1k = 10mA = Current through LED (They are in series)

So, LED glows with 10mA brightness, which is brighter than 1mA.

So by varying the voltage regulators output from 3V to 12V, we can change the LED brightness from 1mA to 10mA.

Not sure if I can explain better than this. Would a simulation help?

You can redesign the circuit to provide the output voltage range for the required LED current/brightness.

Um, I've used a lot of LED's in the past, and over the past year or so. I've been working on a pet project ( you could read about it here )
I've been trying to find the sweet spot for current/light output, We've had more usable light output running multiple LED's with low current in parallel than running higher current over a single set of LED's ( and you'll see most users in the thread agree)

Havent been through all the 19 pages, but I think I get what you are trying to do. Nice to see DIY work! Multiple LEDs are an easier approximation to a diffused/omnidirectional light source than a single LED, and hence the perceived improvement in illumination. There will be no measured difference in their light output if their currents are the same.

 

[PhOeNiX]

Ohh. Will read this through too

Well was talking to a friend and this is what he suggested, instead of connecting even the individual fan LED's [4] in series with a resistor and then connect all these sets from different fans in parallel, he suggested i connect only 1 resistor, near the fan controller to be in series with all the fan LED's [20]
which in themselves will be in parallel groups. Something like this.

Also he suggested, instead of pairing each fans LED's in series, i connect them in parallel too, so that even if one blows, the circuit is intact. Can you tell me what resistor to use in both cases (individual 4 LED sets in Series as well as in parallel). Also how do you calculate the Resistor values too, if you guys can mention here so i dont bug you :bleh:.

 

[greenhorn]

I would strongly suggest using induvidual resistors( will help distribute the current better)

what colour LED's are we talking about?

have you calculated the value of the resistor

 

[PhOeNiX]

have you calculated the value of the resistor

No Haven't calculated them yet Read through boy

Also he suggested, instead of pairing each fans LED's in series, i connect them in parallel too, so that even if one blows, the circuit is intact. Can you tell me what resistor to use in both cases (individual 4 LED sets in Series as well as in parallel). Also how do you calculate the Resistor values too, if you guys can mention here so i dont bug you :bleh:.

I would strongly suggest using induvidual resistors( will help distribute the current better)
what colour LED's are we talking about?

Blue LED's. The only gripe i have with individual resistors is that if i want to add another set etc, i don't have to go through the entire hoopla again. And even during the installation, i have to change only a single resistor instead of multiple ones.

 

[greenhorn]

for Blue LED's @ 12V, I'd suggest 100 ohms and three blue LED's in series.

at 3.5V per LED, the LED's will come to 10.5V, and 1.5V across the resistor

at 100 ohm, this will work out to a current of 15 mA

you can parallel any number of these together

when you want to add another set, just add the resistor for that one

 

[PhOeNiX]

Well won't it increase the number of wires ? I was thinking of it like this.

Each fan's 4 LED's in parallel (As series won't work). Best to best i can add a resistor here in series here before i connect them to the fan controller. Basically if i use the PT-FN05 Fan Controller, it gives me a single knob control, but gives me 4 x 3-pin fan connectors which i can use to connect the LED sets to the FC without actually having to join the individual sets to each other in parallel. How does this sound ? This makes each LED set of 4 be seperated from the other sets, but lets me control them all at once too.

And if i go this way, what Resistor would i need in each set of 4 parallel LED's ?

Basically the circuit will be like this.

 

[NinByChoice]

Parallel LEDs are a bad idea!! Dont do that, you will likely burn them quickly. Or a few LEDs will be dimmer than the rest. This is due to the small variation in Vf of these LEDs.

Like you said 4 in series wont work. So put them as 2 in series with a resistor and 2 such in parallel.

| -> Resistor -> LED -> LED -> |
| -> Resistor -> LED -> LED -> |

Tie the ends together and then connect to the fan controller.

 

[NinByChoice]

Posting the schematic for the Design 2 from Post #20, as requested by cranky.

First is using 7805 as a variable voltage regulator.



R2 is used to adjust the voltage from 5V (minimum for 7805) to 10V (maximum with a 12V supply). The resistor in series with the LED (at the output) will limit the current to the required limits at these 2 voltages.

The other trick to use is to connect the ground reference of the regulator to the anode of the LEDs. The problem is you won't get a variable control over the brightness, but it is a constant current source.

Variable control is easily obtained by making the current source adjustable. Circuit below.



The brightness can now be controlled by the variable resistor, R1. The standard 7805 has a 4mA minimum current supply, so you might not be able to turn off the LED very well. This an be bypassed with a resistor parallel to the LED though.

I would suggest a LM317 to have better control for both these designs.

All circuits from datasheet of TI UA7805.

 

[caldran]

try mc34063 . . efficiency is about 80 percent . .

from the photo it looks like mc34063 IC is used and there is no use in splitting with multiple resistors as all the 4 connectors are in parallel(connected to the same point). a normal high wattage resistor will do the trick..