Power 3W LED help requried!

[kl22]

Update:

For future reference for those that might need something like this.

The circuit I used was http://www.instructables.com/id/Circuits-for-using-High-Power-LED-s/step8/a-little-micro-makes-all-the-difference/ + Soeren's addition, suggestions and modifications.

So we ended up with the attached circuit.

The parts are as follow for 2 3W LEDs connected in series:
R1 = 220 ohm (1/4 or 1/2 Watt)
R2 = 0.47 ohm (5W)
Q1 = Fairchild BC337-40 (45V 800MA)
Q3 = Fairchild FQP50N06L-ND (60V 52.4A)
Vin = NOT 5V, I used 2cell lipo ~7.4V (full charge ~8V)

Note: R2 can be changed depending on the current required. Caluclation for R2 = ~0.65/I.


Heatsink R2, LEDs and Mosfet.


-----------------------------------------------

Hi All!

So i just bought myself 2x Power 3W LEDs. They are placed on a star heat sink and have the following specs: 3.2-3.7V 200-1400mA.

Update: hooked up a meter and I get the following from my circuit:
3.3V @ 600mA across LED.

So I designed a circuit that I thought would drive it (probably not efficiently). This is the description of the circuit:

1x N-Channel power MOSFETs (P30N06LE http://www.fairchildsemi.com/ds/RF/RFP30N06LE.pdf)
2x 4 ohm resistor (I think it is rated at 2W? its a huge resitor)
1x 5.1k resistor
1x 6V battery

Connections:

N-Channel Connections
G - connects to 5.1k resistor then to I/O of Axon
S - connects to ground
D - Connects to (1x LED then to 1x 4ohm then to 6V), another set of what is in the brackets is in parrallel so

D---+---LED-----5ohm----- 6V
      |
      |---LED-----5ohm----- 6V

So my problem is that the resistor is heating up like crazy! to the point it was just about to start smoking. Wondering what I can do to fix this problem. Or if anyone has a circuit that uses a switching voltage regulator that they can share with me. (I want to be able to drive it via PWM).

Thanks in advance

[rbtying]

You need a resistor rated for the amount of power you're running through it - power is I²R, so (0.6A) * (0.6A) * (5ohm) = 1.8W.  The most common type of resistor (and as such the one you probably have) is rated to 0.25W - find/buy/source some power resistors and you'll be fine.

[kl22]

Hi I am using a power resistor, looks kind of like this
http://www.kpcomponents.ca/images/2W%20Metal%20Oxide%20Resistors.gif

It about 1/2 cm in diameter.

The guy told me it rater at 2W... but it's over heating fault maybe?

any suggestions on a voltage regulator that would do the trick?

[rbtying]

Try one of these per LED?  It seems like it'd work, with some extra overhead available.

[TrickyNekro]

You need a resistor that's rated at least 2 times more than your current power usage, as a failsafe.

Alas, you'll need a 5W resistor. And you gotta see the datasheet of the resistor for its heat dissipation.

You may as well need a heatsink... To help it dissipate more heat.

And that's all it's not that's something is wrong with your circuit, it's that you'll need a higher Watt rated resistor...

Best Regards, Lefteris
Greece

PS. What's the on resistance of your MOSFETs? You know that you can control it via the voltage at the gate.

You can PWM it, but's generally that's NOT recommend without proper protection, you'll need some Schottky diodes.

[Soeren]

Hi,

If the MOSFET can handle the excess power (and is mounted on a suitable heatsink), a small signal transistor and a resistor can set the max current and then you just use PWM to dim it.

Another possibility is to drop some of the excess voltage in 2..3 power diodes.

Best way is to use a current mode switcher. You can find a simple one at Roman Blacks website - don't remember if it's dimmable though.

[kl22]

Thanks TrickyNecro and Soeren

Ill try the 5W resistor first. If that works I'm gonna try out what Soeren suggested seems it would waste less power.

Ill post a reply once I get it working!

[TrickyNekro]

If the MOSFET can handle the excess power (and is mounted on a suitable heatsink), a small signal transistor and a resistor can set the max current and then you just use PWM to dim it.

I'd recommend a totem pole circuit for that... Don't forget there is a gate-drain capacitance that can cause excessive problems, especially on the transistor ( of course you still need schottky diodes)


Not that Sören's idea is wrong. By any means No.... But using totem poll is just a little bit finer for the expense of complexity.

[kl22]

Hi,

Thanks for the suggestion TrickyNekro.

I'm alittle caught up with my motor controller right now I'll let everyone know what I do/and if it works:D

[Soeren]

Hi,

I'd recommend a totem pole circuit for that... Don't forget there is a gate-drain capacitance that can cause excessive problems, especially on the transistor ( of course you still need schottky diodes)


Not that Sören's idea is wrong. By any means No.... But using totem poll is just a little bit finer for the expense of complexity.

It's two different things and not mutually exclusive.

The purpose of the current limiting transistor I suggested is just to pull the input down, keeping the max current through the MOSFET constant (as good as it gets anyway).

The gate capacitance (input capacitance) is called C_iss and is the sum of the g-d and g-s capacitance, and for the REP30N06LE it is 1.35nF.
To turn it around a bit, Q_g at 5V is 28nC (typ and 34 nC max).

Had the transistor been chosen a bit more carefully (i.e. a 3..5A logic level device), C_iss and Q_g would have been substantially lower, which is why I never recommend using monster devices for light jobs - but if that's what is at hand...

Don't forget that it would be operated in a combination of switching and linear modes.

The transistor I suggested doesn't determine whether you use a totem pole or not, as the "pole" would be for amplifying the current to the gate, while what I suggested is just pulling the gate down (so would just have more work to do) and since it needs to ride in the MOSFETs linear range to be used for current limiting anyway, I'd suggest using a gate resistor of 220 Ohm (and no totem pole, but if one is at hand, a smaller device would be my choice).

The fact that the MOSFET won't reach the 47mOhm mentioned in the datasheet is really of no consequence at all in this application.

And where do you want to use Schottky diodes??


BTW. Do you have a Swedish or German keyboard?  The letter "ö" (o-umlaut) is used in Sweden and Germany), but the Danish equivalent is "ø" (0-slash) 

[TrickyNekro]

Don't forget that it would be operated in a combination of switching and linear modes.

Alright, I didn't have that in mind really... If he's going to use the MOSFET in the lineal area when totem poll is no good.
I was talking about switching.

In fact in the past I was using a simple transistor NOT with 1k pull up and had slower On time than Off time. I had in mind,
that I used that configuration for an H-bridge and had been proved a total disaster.Of course that won't be a problem here.

The transistor I suggested doesn't determine whether you use a totem pole or not, as the "pole" would be for amplifying the current to the gate, while what I suggested is just pulling the gate down (so would just have more work to do) and since it needs to ride in the MOSFETs linear range to be used for current limiting anyway, I'd suggest using a gate resistor of 220 Ohm (and no totem pole, but if one is at hand, a smaller device would be my choice).

I get someway around... Not all the way. How can he ride the Mosfet to the lineal region without being able to determine the voltage
at the gate? To drive it to the lineal region he would need an RC circuit and PWM. He probably has pwm, but no RC?
If he's driving the MOSFET with a direct PWM signal then he's using it in the switching region.
And why such a high gate resistor, it's generally most needed to have on, but I usually use one at the 18 - 50 Ohm range.

And where do you want to use Schottky diodes??

Hmmm, Imagine the totem poll circuit driving a motor (the one lead), don't you usually place Schottky there.
I'm talking about these Schottky. Just before the gate resistor to protect the driving circuit, from the evil capacitive load. 

With a 220Ohm resistor though, he may not need to use them.

BTW. Do you have a Swedish or German keyboard?  The letter "ö" (o-umlaut) is used in Sweden and Germany), but the Danish equivalent is "ø" (0-slash) 

Mein Freund lerne ich Deutch, weil ich werde nach Bremen gehen. Und ich werde Robotik für sechs Monate dort studieren.
Erasmus, oh wunderbar, Erasmus!!! 


Best Regards, Lefteris
Greece 

[kl22]

Wow Thanks for all the great info!

I'm going to do some research with what you two said, and maybe I'll understand .

I'm slowly leaning away from the totem pole idea (seems a little complicated), but I'm going to try it sooner or later since it's something I've never heard of before. I'm still trying to look for a suitable MOSFET + transistor + resistor to do the job for me (maybe simpler?). Worst comes to worst, I"ll go with a suggestion that Soeren mentioned about the 3 diodes. Please correct me if I am wrong but the diodes draw 0.7V each correct? And if I do do the diode route, will I need some form of cooling for them and can I PWM those?

BTW, PWM is not mandatory.

I'm a soft eng, a lot of this electrical stuff goes way..wayyy over my head.

[TrickyNekro]

Just for being exact, saying that you're gonna use totem pole, I've uploaded this picture with some
"instructions" for you to see. That's all.



Best Regards, Lefteris
Greece

[Soeren]

Hi,

How can he ride the Mosfet to the lineal region without being able to determine the voltage
at the gate? To drive it to the lineal region he would need an RC circuit and PWM. He probably has pwm, but no RC?

When the current reaches the preset value, the BjT gives negative feedback to the gate of the MOSFET. This keeps the MOSFET from switching through to its saturated value of 47mOhm.

If he's driving the MOSFET with a direct PWM signal then he's using it in the switching region.

As I wrote, it's a combination. It's not really called a switching region, but I get what you mean.
When the PWM pulls down, it's indeed a switching signal, but the feedback means that it never reaches "the other end", but stays in the linear region, acting as a current-regulated resistor.

And why such a high gate resistor, it's generally most needed to have on, but I usually use one at the 18 - 50 Ohm range.

"Usually" depends on where you look
Without a (small-ish) gate resistor (or a ferrite pearl), the MOSFET may start oscillating at HF in some instances.
For switching use, it should be as low a value that keeps the oscillating from happening (at any time).
In this case, I consider it more important to keep the I/O pin that controls it from going into latch-up and to keep the stress on the BjT low.
220 Ohm won't hurt the controller in any case, but this is a starting point and any circuit, however simple, can (and should be) be experimented with if the proper tools (like an oscilloscope) is at hand.
The comparatively small losses due to the slow turn on won't matter, as we're introducing loss as the goal of the exercise.

Hmmm, Imagine the totem poll circuit driving a motor (the one lead), don't you usually place Schottky there.

Not at the gate side of the driven MOSFET, do you??
They have no purpose there.

I'm talking about these Schottky. Just before the gate resistor to protect the driving circuit, from the evil capacitive load. 

Before the gate resistor?
Do you mean directly at the I/O pin?
How on earth should a diode be able to negotiate a capacitive load?

The only diodes I use is drain-source, to protect against the inductive kick back from the motor.

With a 220Ohm resistor though, he may not need to use them.

Nobody needs to use the ones you mention and there's no need for a drain-source diode either, as he's driving an LED (i.e. no inductive kick-back).

(As a side note/FYI, if you write either "220 Ohm" or "220ohm", it's much easier not to read it as 2k2).

Mein Freund lerne ich Deutch, weil ich werde nach Bremen gehen. Und ich werde Robotik für sechs Monate dort studieren.
Erasmus, oh wunderbar, Erasmus!!! 

Hehe - you're better at electronics than at German   
(My "sister in law" is married to a German and lives in Germany about 150 km NE of Bremen and while I speak Danish to their kids, they usually answer me in German - and spoken German is not exactly my strong side )

So, how did you get on the Erasmus?

[TrickyNekro]

Hello,

"Usually" depends on where you look
Without a (small-ish) gate resistor (or a ferrite pearl), the MOSFET may start oscillating at HF in some instances.
For switching use, it should be as low a value that keeps the oscillating from happening (at any time).
In this case, I consider it more important to keep the I/O pin that controls it from going into latch-up and to keep the stress on the BjT low.
220 Ohm won't hurt the controller in any case, but this is a starting point and any circuit, however simple, can (and should be) be experimented with if the proper tools (like an oscilloscope) is at hand.
The comparatively small losses due to the slow turn on won't matter, as we're introducing loss as the goal of the exercise.

I've been hearing that " the slower, the better " a lot these days... Of course I don't argue about that, just asking.
Mosfets are voltage driven devices anyhow...


About the Schottky, See the above your post. I have uploaded an image with the circuit I'm referring too.
Generally, I learned about the Schottky from a friend that burned some pairs of transistors before adding them.
Knowing this, I recommend them

(As a side note/FYI, if you write either "220 Ohm" or "220ohm", it's much easier not to read it as 2k2).

Yup, generally for Ohm, here in Greece we have this symbol: Ω. The Omega. Which is a Greek letter.
But I think that many guys here had trouble seeing it (no language input) and I just replace it with Ohm.
And I usually don't know if I'm wrong or right to write either 220Ohm or 220 Ohm.... Cause I usually write 220Ω.

Mein Freund lerne ich Deutch, weil ich werde nach Bremen gehen. Und ich werde Robotik für sechs Monate dort studieren.
Erasmus, oh wunderbar, Erasmus!!! 

Hehe - you're better at electronics than at German   
(My "sister in law" is married to a German and lives in Germany about 150 km NE of Bremen and while I speak Danish to their kids, they usually answer me in German - and spoken German is not exactly my strong side )

So, how did you get on the Erasmus?

Man, I study physics in Aristotle University, third year now. So, I go there for my final (Praktikum, what so ever).
Erasmus is a European program about student exchange (yeah, I guess you knew that :p ), So I leave my university here,
for a semester and go to Bremen for a semester.
So, I did my paper, got my recommendations handed them to the Aristotle administration and they said... hmm Lefteris you seem okey dokish... you can go :p

And hell man, I've only been learning German for 2 months, that's as good as I can get in such short time
And I have something more than half a year, to go there. I guess I'll be better by then


Best Regards, Lefteris
Greece

[waltr]

The resistance unit, Ohm, is always capitalized since it is a proper noun named after the German physicist Georg Ohm. This goes for any of the other units is electronics (and other sciences).  TrickyNekro correctly capitalized Schottky and Ohm.

[TrickyNekro]

The resistance unit, Ohm, is always capitalized since it is a proper noun named after the German physicist Georg Ohm. This goes for any of the other units is electronics (and other sciences).  TrickyNekro correctly capitalized Schottky and Ohm.

The correct symbol is the Greek letter omega. " Ω ".... But that would be a very large problem cause everyone would need a Greek keyboard just for some letters...

So in English, we simply write Ohm... ;-)

Capital or not, isn't the case.... The symbolism is relative to anyone...

Is I say in a book that I write that "faegfag" is for Ohm then "faegfag" means Ohm in that book....

BUT it's generally accepted that we symbolize Ohm with the Greek omega.

[kl22]

Thanks a lot TrickyNekro for the schematic!!

that should be perfect for me to get started/finish it

The only thing I can't understand is where does the LEDs go?

UPDATE:
Neverming, R4

Couple of questions:
1. So I should be able to drive two 3W LEDs with this circuit? And how would I hooked it up (series or parallel?)?
2. Can I lower the VCC to either 6 or 8.4V(2 cell lipo), since my 12V battery is dedicated to the motor?
3. Does the circuit schematic you provided, provide the required volts and current to R4 (3.2-3.7V 200-1400mA per LED)?

[kl22]

What do you think about this:

http://www.instructables.com/id/Circuits-for-using-High-Power-LED-s/step8/a-little-micro-makes-all-the-difference/

From what I gather the following are the parts for it:

R1: approximately 100k-ohm resistor (such as: Yageo CFR-25JB series)
R3: R3 = 0.5 / 0.7 = 0.71 ohms. closest standard resistor is 0.75 ohms. Current set resistor.
Q1: small NPN transistor (such as: Fairchild 2N5088BU)
Q2: large N-channel FET (such as: Fairchild FQP50N06L)
Z1: use a 4.7 or 5.1 volt zener diode (such as: 1N4732A or 1N4733A)

[TrickyNekro]

Hmm, why to use a power FET that also needs a negative voltage at the gate to be driven (negative in respect to the Source)
and don't use a power transistor...

Much much easier.

Secondly, never connect a power source directly at a zener. NEVER! That's a rule, always place a current limiting transistor.

You can build a very very nice current source with a microcontroller, a power transistor, two resistors and one capacitor.
If you need I'll upload a circuit.


About the voltage consideration you had. What is the forward voltage of your leds? Knowing this I can answer

Best Regards, Lefteris
Greece

[kl22]

You can build a very very nice current source with a microcontroller, a power transistor, two resistors and one capacitor.
If you need I'll upload a circuit.

Yes please, could you provide a circuit. If you are able to create the schematic for me, please keep in mind there are two 3W LEDs that need to be connected.

What is the forward voltage of your leds? Knowing this I can answer

Is there anyway I can find this out? This is what it said on the package it says 3.2-3.7V 200-1400mA. And I did some research online and it seems that is the forward voltage.

[TrickyNekro]

So that's your forward voltage, no problem...

The bad news, is that your gonna need something more than a simple power transistor (my bad)
But I'll try make it as simple as I can. Basically I'll try using a LM317 as it's very cheap and easy to find
no worries...

Give me some time, I'll have it ready ;-)

[Soeren]

Hi,

The resistance unit, Ohm, is always capitalized since it is a proper noun named after the German physicist Georg Ohm. This goes for any of the other units is electronics (and other sciences).  TrickyNekro correctly capitalized Schottky and Ohm.

Sure, and when dealing with symbolic notation in Ohms Law, you use Either "U" (in Europe) or "E" (in the US) and not "V" together with "I" and "P", but even so...
But the point here was just to clarify the difference between zero and the capital letter "O".

Strictly speaking, it's not correct to use eg. 2k2 for 2.2kOhm either, but this too is to avoid misreadings and vanishing decimal points through photocopying photocopies and since you use a point in the US, while we use a comma for decimals, the 2k2 notation is more universal, sanctioned by the standards institutes around the world or not.
And there's plenty of similar stuff going on.
 
I think good old George would have us excused for doing it (not that I do it myself) and it's certainly not as bad as writing mF for µF, which can ultimately lead to mishaps and accidents, while I don't think anyone would doubt that "Ohm" and "ohm" represents the same thing.

It's just about making things more readable.

[Soeren]

Hi,

What do you think about this:

http://www.instructables.com/id/Circuits-for-using-High-Power-LED-s/step8/a-little-micro-makes-all-the-difference/

That's exactly what I was talking about (minus R1 and the zener diode).

From what I gather the following are the parts for it:

R1: approximately 100k-ohm resistor (such as: Yageo CFR-25JB series)
R3: R3 = 0.5 / 0.7 = 0.71 ohms. closest standard resistor is 0.75 ohms. Current set resistor.
Q1: small NPN transistor (such as: Fairchild 2N5088BU)
Q2: large N-channel FET (such as: Fairchild FQP50N06L)
Z1: use a 4.7 or 5.1 volt zener diode (such as: 1N4732A or 1N4733A)

Loose R1 and Z1.
You need a resistor between I/O pin and gate (the 220R I mentioned).
Q1 can be just about any small signal transistor (BC547, 2N3904 etc.)
Q2 could be rated lower, but will do fine - it will need a heatsink!
R3 should be ~0.65V/I_LED
(0.65V is the approx. base-emitter voltage drop for an average BjT).
The collector of the BjT goes directly to the gate of the MOSFET.

You can still use a series connected diode or a couple (depending on your drive voltage) if you like and this will require less heatsink area, as less power will then be dissipated in the MOSFET.

[kl22]

Hey TrickyNekro,

Give me some time, I'll have it ready ;-)

No Rush! Whenever you can and when you have time! Thank you!

That's exactly what I was talking about (minus R1 and the zener diode).

Sorry, I'm a little electronics illiterate.



Hi Soeren,

Loose R1 and Z1.
You need a resistor between I/O pin and gate (the 220R I mentioned).
Q1 can be just about any small signal transistor (BC547, 2N3904 etc.)
Q2 could be rated lower, but will do fine - it will need a heatsink!
R3 should be ~0.65V/I_LED
(0.65V is the approx. base-emitter voltage drop for an average BjT).

Couple of questions:
1. Does the circuit I attached make sense assuming I only want 1000mA of current going to the LEDs?
2. Is the value of Rc correct? Checked up on DigiKey and the closest to 0.65ohm was 0.68ohm
3. What should the watt ratings for Rg and Rc be? I'm assuming Rg=1/4Watt and Rc=5W

The collector of the BjT goes directly to the gate of the MOSFET.

As for this, should I use the solid line as it was originally connected or use the dotted line that I put in?

Thanks in advance!

[Soeren]

Hi,

Couple of questions:
1. Does the circuit I attached make sense assuming I only want 1000mA of current going to the LEDs?

Yes, it's a constant current circuit and can be set as low as you like.

2. Is the value of Rc correct? Checked up on DigiKey and the closest to 0.65ohm was 0.68ohm

It is this resistor that sets the current.
The 0.65V is an average and it can be between ~0.6V to ~0.7V with the current passing through it.
If the base-emitter drop is 0.65V, the LED current will be around 956mA, but can be from ~882mA to ~1029mA (not a very large difference in light output).

3. What should the watt ratings for Rg and Rc be? I'm assuming Rg=1/4Watt and Rc=5W

Rg (I assume you mean the 220 Ohm gate resistor) need just be a regular 0.25W resistor.
The current sense resistor dissipates slightly under 3/4W, but I have chosen a 2W resistor in the circuit below.

As for this, should I use the solid line as it was originally connected or use the dotted line that I put in?

Here's what it should look like:


I you place the two LEDs in series, you need to raise the voltage to the circuit to slightly above 8V (to say 10V).
If you want to drive them in parallel, put a resistor in series with each LED before parallelling. Their value should be 0R33 to 0R47 and rated for 1W (or higher) and their purpose is to counter for inequalities in the LED's - you may run them without the equalizing resistors, as modern white LED's have a quite large vltage drop range, but you may experience a little difference in their light output (if they're from the same production batch, there will probably not be any difference though).

[kl22]

Wow that's perfect! Thanks a lot Soeren!

I noticed that you wrote 5V input, does this mean I cannot use my 2cell LiPo (~7.4V)?

[Soeren]

Hi,

You can use whatever voltage (up to whatever's the MOSFETs level), but don't go higher than absolutely needed, as this just means more heat to get rid of.
The "+5V" is just a leftover, as I reused an old BjT based circuit made for +5V and just changed values and transistor.
Q1 can be any small signal NPN.

[kl22]

Perfect!!!

Thanks Soeren! I'll post back once I get the circuit made, currently ordering the parts from digikey.

[TrickyNekro]

Hmm, yup I'm sorry I didn't post earlier but go with Soeren solution.

Much simpler than most circuits, I'll be trying to figure out.