Welcome to our site!

Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

  • Welcome to our site! Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

Help for PWM controlling MOSFET for High Current LED

Status
Not open for further replies.

crutschow

Well-Known Member
Most Helpful Member
That circuit is fine since it has a transistor and source resistor current limiter for the MOSFET.
 

dhyaks

New Member
That circuit is fine since it has a transistor and source resistor current limiter for the MOSFET.

actually, can you explain to me, what the function of transistor beyond the MOSFET ? and relation between the PWM output ?
i mean how about if i'm not use that transistor, just Rsense below Source pin of FET, and gate still have the Output of PWM.
 

audioguru

Well-Known Member
Most Helpful Member
What is the value of your current sensing resistor then we can simply calculate the maximum peak current?
How did you measure the voltage and the current? A meter cannot measure high frequency pulses accurately.
 

audioguru

Well-Known Member
Most Helpful Member
The transistor turns on when its base-emitter voltage is about 0.65V (and its voltage requirement changes when the current and temperature changes).
The load current causes a voltage drop across the current sensing resistor. Ohm's Law calculates the resistor value. For 3A then the current sensing resistor should be 0.65V/3A= 0.22 ohms. But never use an LED at its maximum allowed current, use a 0.27 ohms resistor instead for an LED maximum peak current of 0.65V/0.27 ohms= 2.4A.

What value current sensing resistor did you use with the 35W halogen light bulb?
 

MrAl

Well-Known Member
Most Helpful Member
Hello there,

I highly recommend using a regular buck circuit, not a pulsing circuit. A pulsing (PWM without inductor and filter cap) circuit will pulse the LED on and off only and that does not provide for a true power conversion so for one thing the efficiency of the circuit will suffer because it's just like driving an LED with a variable resistor.
Second you also need to check the data sheet anyway to see if this LED can be pulsed at frequencies greater than about 1kHz or so as some LED's are not recommended for pulsing applications.
The above two points means you should really use a true buck circuit not a pulsing circuit.

Simple pulsing circuit example:
You have a 3v, 1.5 amp LED and a 6v power source and you pulse the LED on and off (with or without a series resistor, but you'd need a series resistor for proper LED operation), and the series internal and external resistance total is 1 ohm.
When the MOSFET turns on the current through the LED and the total series resistance is:
I1=(6-3)/1=3 amps
and when the MOSFET turns off the current is:
I2=0 amps.
To get an average current of 1.5 amps we'd have to pulse the MOSFET on and off with a duty cycle of 50 percent. We then would get the average current:
Iavg=(I1+I2)/2=1.5 amps

Doesnt seem like there is a problem right? Well, what happened to that extra 3v of the power source (it's 6v and the LED is only 3v)? Answer: it is dropped in the series resistance. We have 3v dropped and 1.5 amps average so we are wasting:
Pwasted=3*1.5=4.5 watts

and the power in the LED is:
Pled=3*1.5=4.5 watts

so guess what, that means our circuit is only 50 percent efficient! That's nasty by today's standards.

Using a buck circuit is a different story. The buck circuit provides for a true power conversion so the efficiency can be as high as 90 percent for a decent design, and 85 percent for just a reasonable design, and even 80 percent for an old type design.

So you see the buck is the way to go. Any other way is just a waste. The buck also does not require the LED itself to pulse.

Note that the color of some LEDs shifts slightly when operating under the rated current. LED's that can be pulsed may not suffer from this because they can be pulsed at the peak max current all the time. This only works with LEDs that can be pulsed however, and then there's the enormous waste of power.
 

crutschow

Well-Known Member
Most Helpful Member
actually, can you explain to me, what the function of transistor beyond the MOSFET ? and relation between the PWM output ?
i mean how about if i'm not use that transistor, just Rsense below Source pin of FET, and gate still have the Output of PWM.
As AG noted the transistor turns on when the current through the sense resistor creates a voltage drop across the resistor of about .65V. This reduces the gate voltage to the MOSFET and starts turn it off so that the current through the sense resistor doesn't go any higher, limiting the peak LED current.

If you don't have the transistor then the peak current will go much higher before it is limited and possibly zap your LED.

You seem to have difficulty understanding that the peak LED current must be limited in a PWM driver (which isn't needed for an incandescent bulb since they inherently limit the current). Is it clear now why the LED current must be limited?
 

audioguru

Well-Known Member
Most Helpful Member
The transistor and the low voltage (0.65V) current sensing resistor allow the power supply voltage to be variable and the maximum current remains constant. Then when the Mosfet turns on it becomes a variable power resistor to regulate the maximum current.
 

dhyaks

New Member
What is the value of your current sensing resistor then we can simply calculate the maximum peak current?
How did you measure the voltage and the current? A meter cannot measure high frequency pulses accurately.

The transistor turns on when its base-emitter voltage is about 0.65V (and its voltage requirement changes when the current and temperature changes).
The load current causes a voltage drop across the current sensing resistor. Ohm's Law calculates the resistor value. For 3A then the current sensing resistor should be 0.65V/3A= 0.22 ohms. But never use an LED at its maximum allowed current, use a 0.27 ohms resistor instead for an LED maximum peak current of 0.65V/0.27 ohms= 2.4A.

What value current sensing resistor did you use with the 35W halogen light bulb?

I'm just measure at the output voltage on Pin 3 of 555, with analog meter, and measure the load on point in my schematic in page 1,
And when i put Halogen load the Sensing Resistor is 0,1 Ohm / 5 Watt.

Hello there,

I highly recommend using a regular buck circuit, not a pulsing circuit. A pulsing (PWM without inductor and filter cap) circuit will pulse the LED on and off only and that does not provide for a true power conversion so for one thing the efficiency of the circuit will suffer because it's just like driving an LED with a variable resistor.
Second you also need to check the data sheet anyway to see if this LED can be pulsed at frequencies greater than about 1kHz or so as some LED's are not recommended for pulsing applications.
The above two points means you should really use a true buck circuit not a pulsing circuit.

Simple pulsing circuit example:
You have a 3v, 1.5 amp LED and a 6v power source and you pulse the LED on and off (with or without a series resistor, but you'd need a series resistor for proper LED operation), and the series internal and external resistance total is 1 ohm.
When the MOSFET turns on the current through the LED and the total series resistance is:
I1=(6-3)/1=3 amps
and when the MOSFET turns off the current is:
I2=0 amps.
To get an average current of 1.5 amps we'd have to pulse the MOSFET on and off with a duty cycle of 50 percent. We then would get the average current:
Iavg=(I1+I2)/2=1.5 amps

Doesnt seem like there is a problem right? Well, what happened to that extra 3v of the power source (it's 6v and the LED is only 3v)? Answer: it is dropped in the series resistance. We have 3v dropped and 1.5 amps average so we are wasting:
Pwasted=3*1.5=4.5 watts

and the power in the LED is:
Pled=3*1.5=4.5 watts

so guess what, that means our circuit is only 50 percent efficient! That's nasty by today's standards.

Using a buck circuit is a different story. The buck circuit provides for a true power conversion so the efficiency can be as high as 90 percent for a decent design, and 85 percent for just a reasonable design, and even 80 percent for an old type design.

So you see the buck is the way to go. Any other way is just a waste. The buck also does not require the LED itself to pulse.

Note that the color of some LEDs shifts slightly when operating under the rated current. LED's that can be pulsed may not suffer from this because they can be pulsed at the peak max current all the time. This only works with LEDs that can be pulsed however, and then there's the enormous waste of power.

Thanks for your suggestion, can you show the the sample of "SIMPLE REGULAR BUCK CONVERTER" that can make it D.I.Y ?
because i need a simple but durable circuit to turn the LED on for little long time journey. (amybe Between 4-8 Hours).
 

dhyaks

New Member
Field Report of WEDNESDAY, 17 - 09 - 2014

Have borrowed XML-T6 LED from my friend, and going to try my circuit on.
at first time plug the LED on "LOAD" cable, then adjust the trimpot, with meter show on 2A (i place the Sensing Resistor at 0,3R / 5W)
fisrt 1 Minute is going steady at 2A, but after that the Current goes down to 1,4A, and Heatsink of MOSFET is too Hot.
after 5 Minutes and it can reach over than 100'C temperature.
1ioe1f.jpg


29yr56r.jpg
2qujjtw.jpg


so what's wrong then? may my Heatsink is too little? or i just get wrong circuit again? aaaaaaaaaaaaaah life is too difficult. :( can anybody give a better circuit than mine?
 

dhyaks

New Member
and How about the "TOTEM POLE" adding circuit to drive the MOSFET ?
like this schematic : (sorry for bad camera)
16862r7.jpg


i'm surfing and get this link : "totem pole circuit"
and one more question, : can i put the "LOAD" L-C-D component after Source pin on MOSFET? so not put it in Drain but after Source Pin like above schematic? with still using an N-MOSFET?
 

crutschow

Well-Known Member
Most Helpful Member
............................................................
i'm surfing and get this link : "totem pole circuit"
and one more question, : can i put the "LOAD" L-C-D component after Source pin on MOSFET? so not put it in Drain but after Source Pin like above schematic? with still using an N-MOSFET?[/QUOTE]No. A standard N-MOSFET requires a gate voltage 10V higher than the supply voltage in that circuit to fully turn-on.

You could however use a P-MOSFET with the source to the power supply and the drain being the output.

Note that circuit has nothing to limit the LED current.
 

dhyaks

New Member
............................................................
i'm surfing and get this link : "totem pole circuit"
and one more question, : can i put the "LOAD" L-C-D component after Source pin on MOSFET? so not put it in Drain but after Source Pin like above schematic? with still using an N-MOSFET?

No. A standard N-MOSFET requires a gate voltage 10V higher than the supply voltage in that circuit to fully turn-on.

You could however use a P-MOSFET with the source to the power supply and the drain being the output.

Note that circuit has nothing to limit the LED current.[/QUOTE]

So i can keep my first circuit to play the MOSFET (that load and L-C component is put on Drain pin)
now how can my circuit is getting too HOT while i was put the Rsense (say Current limit) is no more than 3A full load.?
 

audioguru

Well-Known Member
Most Helpful Member
Please write the power supply voltage on every schematic you draw.
On your "totem pole" circuit I assume the supply is +12.5V but the diode reduces it to about 11.5V.
The Mosfet needs a gate-source voltage of 10V to fully turn on. Maybe your Mosfet is more sensitive than most.

Why is your current sensing resistor the strange value of 0.3 ohms? 0.27 and 0.33 ohms are standard values. Your ohm meter measures the resistance of its leads plus a resistor it is measuring. If the resistor is actually 0.3 ohms and it has an average current of 3A then it heats with 2.7W and too much heat might increase its value (reducing the current).
Why does the circuit have a current sensing resistor when that resistor is doing nothing? It could be used to measure the current since the current is the voltage across it divided by its value. Your current meter also has a current sensing resistor inside that actually reduces the supply voltage to the circuit.

Your analog meter measures the average current, not the peak current. The average includes the off time of the PWM so if the PWM has equal on and off times and you measured 3A then the peak current is 6A. Maybe the duty-cycle of the PWM is 1/4 on and 3/4 off then with an average of 3A the peak current is 12A.
6A x 0.3 ohms= 1.8V. The maximum output from the totem pole NPN emitter-follower is 11.5V - about 1V= 10.5V. With a peak current of 6A then the LED forward voltage might be 4.0V.

The source of the Mosfet will be at 4.0V + 1.8V due to the 6A peak current in the sensing resistor so the peak gate-source voltage is 10.5V - 4.0V - 1.8V= 4.7V.
The heating of the Mosfet is the voltage across it times the current in it. The voltage across it is 11.5V - 4.0V - 1.8V= 5.7V. It might be turned on for half the time by the PWM then its average voltage is 2.85V. The average current is 3A then the heating in the Mosfet is 2.85V x 3A= 8.55W. Your little heatsink might cool 3W if it is outside the box it is in.
 
Last edited:

crutschow

Well-Known Member
Most Helpful Member
................................................
so what's wrong then? may my Heatsink is too little? or i just get wrong circuit again? aaaaaaaaaaaaaah life is too difficult. :( can anybody give a better circuit than mine?
The simplest high efficiency circuit would be one with an IC controller similar to what Ron showed in Post #7.

Below is the simulation of a Hysteretic constant-current design using a commonly available LM339 comparator. The circuit should have an efficiency of around 85% or better with proper components. The inductor needs to carry 10A or greater with low resistance, the P-MOSFET should have an ON resistance of no more than about .02 ohms, and the diode should be a Schottky type rated for at least 10A.

The circuit is shown delivering 9A average to the LED. This can be adjusted by pot U3.

Edit: Not shown are the necessary decoupling capacitors of 0.1uF ceramic and 10uF electrolytic from the MOSFET source directly to the ground plane. With 9A current you need a board with a ground plane for reliable operation (done't even think about doing it on a solder-less plug board).

Edit 2: The hysteretic circuit below shows no current overshoot on startup (although voltage output versions of such circuits do).

LED Driver.gif
 
Last edited:

MrAl

Well-Known Member
Most Helpful Member
I'm just measure at the output voltage on Pin 3 of 555, with analog meter, and measure the load on point in my schematic in page 1,
And when i put Halogen load the Sensing Resistor is 0,1 Ohm / 5 Watt.



Thanks for your suggestion, can you show the the sample of "SIMPLE REGULAR BUCK CONVERTER" that can make it D.I.Y ?
because i need a simple but durable circuit to turn the LED on for little long time journey. (amybe Between 4-8 Hours).


Hi,

Well several IC chips come to mind, in particular, the LM2576 and i think the LM2596 chips. These are cheap enough but guess what...you done even have to build a circuit you can buy the whole circuit on sites like Amazon for maybe 3 dollars (USD). For that you can get a small circuit board that converts any higher voltage to 3v by just adjusting the pot on board. Maybe a small series resistor addition and your good to go.
There are other board sold too like with current adjustment.

Also, if you build your own circuit with a constant duty cycle (like with a 555 ic) then you still get a buck but now you get a buck with a slowly decreasing output current as the battery runs down so you can see that the battery is running down and should be replaced at some point. These kinds of circuits are easy to build because they dont need feedback.

One caution however with home made buck circuits:
When buck circuits start up they can deliver a huge current surge to the load (LED) until the circuit stabilizes (because there is no feedback for a while). Thus you have to be careful and have some way to deal with this. Usually the idea is to use a soft start circuit that gradually ramps up the load current over time.

I understand where you are coming from with this circuit a little better too now. You want to run the LED from a battery of higher voltage and get either full brightness, half brightness, and lower brightness for long lasting battery life. What works nice here is a multi position switch to change modes.
I built a light like this a long time ago and just recently upgraded it to an XML2 LED also. The switch i used was a low current 4 position switch (one position for 'off') and i am going to use a separate switch for the high 2 amp mode.
One difference though, i use an 18650 Lithium Ion cell and that means i dont need a buck circuit or any other circuit, just the switch and various resistors to limit current. In this case the series resistor isnt too much of a problem because the voltage of the cell is close to the voltage of the LED, and with low current settings well it's just low current so who cares as the battery life is very good. I think i have it set for 10ma at the lowest setting, which gives unbelievable battery life for emergencies when the power might be off for an extended time. This came in really handy with Hurricane Sandy when we lost power for 4 days in a row.
 

dhyaks

New Member
Please write the power supply voltage on every schematic you draw.
On your "totem pole" circuit I assume the supply is +12.5V but the diode reduces it to about 11.5V.
The Mosfet needs a gate-source voltage of 10V to fully turn on. Maybe your Mosfet is more sensitive than most.

Why is your current sensing resistor the strange value of 0.3 ohms? 0.27 and 0.33 ohms are standard values. Your ohm meter measures the resistance of its leads plus a resistor it is measuring. If the resistor is actually 0.3 ohms and it has an average current of 3A then it heats with 2.7W and too much heat might increase its value (reducing the current).
Why does the circuit have a current sensing resistor when that resistor is doing nothing? It could be used to measure the current since the current is the voltage across it divided by its value. Your current meter also has a current sensing resistor inside that actually reduces the supply voltage to the circuit.

Your analog meter measures the average current, not the peak current. The average includes the off time of the PWM so if the PWM has equal on and off times and you measured 3A then the peak current is 6A. Maybe the duty-cycle of the PWM is 1/4 on and 3/4 off then with an average of 3A the peak current is 12A.
6A x 0.3 ohms= 1.8V. The maximum output from the totem pole NPN emitter-follower is 11.5V - about 1V= 10.5V. With a peak current of 6A then the LED forward voltage might be 4.0V.

The source of the Mosfet will be at 4.0V + 1.8V due to the 6A peak current in the sensing resistor so the peak gate-source voltage is 10.5V - 4.0V - 1.8V= 4.7V.
The heating of the Mosfet is the voltage across it times the current in it. The voltage across it is 11.5V - 4.0V - 1.8V= 5.7V. It might be turned on for half the time by the PWM then its average voltage is 2.85V. The average current is 3A then the heating in the Mosfet is 2.85V x 3A= 8.55W. Your little heatsink might cool 3W if it is outside the box it is in.
thank you so much for the explain, i ithink i must take a rest first for being crazy with this circuit hahahaha! :D
but if you want, you can give suggestion for this problem like a simple & robust circuit that not use any "weird" component, because in my country i can't fine any difficult IC/P-MOSFET. i don't know how the market is not selling that thing, like an ordinary NE555, LM393, LM358 is ready but if you suggest any IC like LM2596 or either "difficult" component i will hard to try it. ;)


The simplest high efficiency circuit would be one with an IC controller similar to what Ron showed in Post #7.

Below is the simulation of a Hysteretic constant-current design using a commonly available LM339 comparator. The circuit should have an efficiency of around 85% or better with proper components. The inductor needs to carry 10A or greater with low resistance, the P-MOSFET should have an ON resistance of no more than about .02 ohms, and the diode should be a Schottky type rated for at least 10A.

The circuit is shown delivering 9A average to the LED. This can be adjusted by pot U3.

Edit: Not shown are the necessary decoupling capacitors of 0.1uF ceramic and 10uF electrolytic from the MOSFET source directly to the ground plane. With 9A current you need a board with a ground plane for reliable operation (done't even think about doing it on a solder-less plug board).

Edit 2: The hysteretic circuit below shows no current overshoot on startup (although voltage output versions of such circuits do).

View attachment 88224
so i will try to make your circuit, but i will change the base IC with LM393 comparator, because it will hard to find a "weird" series of IC like built up Buck IC or other things, just make an Ordinary circuit, and get the usually Component that in southeast asia selling that on Market. thank you so much... :)


Hi,

Well several IC chips come to mind, in particular, the LM2576 and i think the LM2596 chips. These are cheap enough but guess what...you done even have to build a circuit you can buy the whole circuit on sites like Amazon for maybe 3 dollars (USD). For that you can get a small circuit board that converts any higher voltage to 3v by just adjusting the pot on board. Maybe a small series resistor addition and your good to go.
There are other board sold too like with current adjustment.

Also, if you build your own circuit with a constant duty cycle (like with a 555 ic) then you still get a buck but now you get a buck with a slowly decreasing output current as the battery runs down so you can see that the battery is running down and should be replaced at some point. These kinds of circuits are easy to build because they dont need feedback.

One caution however with home made buck circuits:
When buck circuits start up they can deliver a huge current surge to the load (LED) until the circuit stabilizes (because there is no feedback for a while). Thus you have to be careful and have some way to deal with this. Usually the idea is to use a soft start circuit that gradually ramps up the load current over time.

I understand where you are coming from with this circuit a little better too now. You want to run the LED from a battery of higher voltage and get either full brightness, half brightness, and lower brightness for long lasting battery life. What works nice here is a multi position switch to change modes.
I built a light like this a long time ago and just recently upgraded it to an XML2 LED also. The switch i used was a low current 4 position switch (one position for 'off') and i am going to use a separate switch for the high 2 amp mode.
One difference though, i use an 18650 Lithium Ion cell and that means i dont need a buck circuit or any other circuit, just the switch and various resistors to limit current. In this case the series resistor isnt too much of a problem because the voltage of the cell is close to the voltage of the LED, and with low current settings well it's just low current so who cares as the battery life is very good. I think i have it set for 10ma at the lowest setting, which gives unbelievable battery life for emergencies when the power might be off for an extended time. This came in really handy with Hurricane Sandy when we lost power for 4 days in a row.

actually i can buy it. but there is no "challenge" for it, Do It Yourself is more preferred, beside we can modify it, we can learn from the circuit.
and i'm not using an ordinary battery like 18650 or 14500 but i have done it with Lead Acid battery or VLRA battery (12V) with high Amperage inside it.
 

MrAl

Well-Known Member
Most Helpful Member
Hi,

Ok, no problem, i was offering suggestions to make life simpler, but if you are doing this as part of your hobby or learning goal then that's a different story. In this case we have a hundred ways to go about it :)
 

dhyaks

New Member
Hi,

Ok, no problem, i was offering suggestions to make life simpler, but if you are doing this as part of your hobby or learning goal then that's a different story. In this case we have a hundred ways to go about it :)
Mr.Al
can you give me a suggestion circuit too for my project that i can assembly it ? better if i can't meet difficult IC/P-MOSFET that in my country is hard to find. thank a lot.. because lot of peoples outside there that asking same question to this project but can find the efficient circuit.. :)
regard.
 

crutschow

Well-Known Member
Most Helpful Member
...................................
so i will try to make your circuit, but i will change the base IC with LM393 comparator, because it will hard to find a "weird" series of IC like built up Buck IC or other things, just make an Ordinary circuit, and get the usually Component that in southeast asia selling that on Market.
.............................
The LM393 is a dual version of the quad LM339 so they should both work identically in the circuit.
 

dhyaks

New Member
The LM393 is a dual version of the quad LM339 so they should both work identically in the circuit.
Hi crutschow..
Last night i have try your suggested circuit,, choose the best component that i have on , IRF9540, 1% of resistors, high grade transistor and make a choke and measure it with LCR meter. and assemble it on matrix pcb.
after have done for all. trying to load a Halogen lamp first. (why halogen? because it's have none polarity than LED so i think it more save to try the circuit before LED load)
and can adjust the brightness from the POT. after 5 minutes i get 60'C on MOSFET heatsink.
now i turn it off and change the load into a LED (XML-t6)
with Rsense 0,22 Ohm / 5 watt and Vsource is 12,5V
but only 3 minutes after the LED goes on, do you know.? the temperature is just same as mine too. above 100'C on Heatsink.
so... i can say : "theory is can't same as your practism." "too much theory without Do It Yourself and you have nothing."
sorry i can't take a picture of your based circuit last night. my ipod is empty and i don't want to burn up my other LED too.

ANYBODY CAN SHOW ME THE BETTER WAY TO MAKE THIS LITTLE THING? :)
 
Status
Not open for further replies.

Latest threads

EE World Online Articles

Loading
Top