Buck/Booster circuit for LED gets very HOT.

Discussion in 'The Projects Forum' started by Alex Jurtan, Aug 10, 2015.

  1. Alex Jurtan

    Thread Starter New Member

    Jul 14, 2015
    14
    0
    Hi everyone,
    I have built a circuit for a flashlight that can be used with any common size/amount of batteries. (1xAA, 2xAA, 3xAA, 1xAAA ...etc)
    The Buck/Booster circuit has an input voltage from 0.8V-5.5V and gives a constant output of 3.3V.
    For the IC I am using the TPS61201 and for the LED I am using a Lumiled 5630 4000K.

    My Problem:
    I have tested this circuit on a breadboard, but never noticed that the LED gets really hot.
    I have several ideas what to change, but not sure if any of these would work out:
    - Change the resistor of the LED (probably would get very dimmed)
    - Design a copper surface on the PCB around the LED. Like a heatsink. (Not sure if this small scale would be enough. PCB 16mmx16mm)
    - Integrate some kind of current regulator. (Maybe the LED is getting too much current?)

    Any help or advice is greatly appreciated! Here is the Schematic and my PCB attached.

    Regards
    Alex

    LED TPS61201 SCHEMATIC.JPG IMG_3760.JPG
     
  2. pwdixon

    Member

    Oct 11, 2012
    488
    56
    Does the LED on your breadboard version get hot?
    If not what's the difference in the two circuits? Or is the LED less bright on your breadboard version?
     
  3. pwdixon

    Member

    Oct 11, 2012
    488
    56
    You could easily be over-running the LED, the spec sheet is running at 100mA.

    You are defining the LED using a 2 Ohm resistor so depending on the LED forward voltage which could be as low as 2.9V you could be doing 0.4V/2Ohm ie 200mA.

    The absolute max current is quoted as 150mA.

    200mA is 2.9V x 200mA of power in the LED ie 0.58W.

    Perhaps your original LED had a higher forward voltage and didn't exceed the max current?

    By-the-way your schematic shows the LED the wrong way round.
     
  4. blocco a spirale

    AAC Fanatic!

    Jun 18, 2008
    1,440
    368
    Last edited: Aug 10, 2015
  5. MikeML

    AAC Fanatic!

    Oct 2, 2009
    5,450
    1,066
    Blocco beat me to it. LEDs should be driven with an IC constant-current regulator; not a constant-voltage regulator and a resistor. They are not incandescent lamps!!!

    All high-power LEDs will get hot enough that they they need to be mounted on a heat-sink. Sometimes, the heatsink can be extra copper on a PCB, sometimes they are bolted to a block of aluminum.
     
  6. Alex Jurtan

    Thread Starter New Member

    Jul 14, 2015
    14
    0
    Thanks to everyone for helping out!

    This is my first electronic project and I had a couple of hick-ups on the way. Such as having the LED the wrong way around, but I noticed that during building the circuit. :)

    So as I understand, I need a constant-current regulator. I do not fully understand the APPLICATION REPORT, Blocco has send.
    Can I modify my board and use the same chip TPS61201 on it? Connecting the LED to FB with a resistor?
     
  7. pwdixon

    Member

    Oct 11, 2012
    488
    56
    I would say you can't use the same chip as the one you have has a fixed output voltage of 3.3V, the one in the application note takes the output to whatever voltage will result in FB reaching a reference value independent of the voltage across the LED itself. In the application note the FB voltage (and hence the ref voltage) and the resistor define the current in the LED. The sad thing is that had you chosen a chip with a higher output voltage then variations in LED voltage would not have been such a factor and your original circuit would have worked albeit that the brightness might have varied a bit.
     
  8. blocco a spirale

    AAC Fanatic!

    Jun 18, 2008
    1,440
    368
    It looks like you will need to use the TPS61200 and the circuit in fig.2

    Connect the top end of R2 to FB and omit R5 and R6. R2 sets the current.
     
  9. Alex Jurtan

    Thread Starter New Member

    Jul 14, 2015
    14
    0
    Ok thank you.
    The problem I have now is that I want to make this circuit work with different amount of batteries. The TPS61200 has not a fixed output.
    How could I control the output? Or does the output depend on the input voltage.
    That was the main reason why I chose the TPS61201, because it is able to use one battery or up to three batteries and give the same output voltage.
     
  10. MrAl

    Well-Known Member

    Jun 17, 2014
    2,438
    492
    Hello there,

    You can use a fixed output voltage to drive an LED but only if you provide the right dropping resistor.

    But first, the power dissipation requires a commitment to physical board space or heatsink size. There's no way to cheat on this part unless you use a fan which i dont think you want. If your LED draws 100ma and the forward voltage is 3v, then that's 0.3 watts and using the rule of thumb of 1 square inch of surface area per watt per 60 degree C temperature rise, you'll need at least 0.3 square inches of surface area to ensure only a 60 degree C rise. Twice that would be even better, like 0.5 to 0.6 square inches. If you dont have at least 0.3 square inches then you need a heat sink or fan or both.

    To use a fixed output voltage regulator it is necessary to maintain the proper amount of headroom and choose the resistor value such that it can handle every LED forward voltage that the circuit will ever encounter. This variation comes from two main sources:
    1. The characteristic LED voltage.
    2. The change in voltage with temperature.
    If the LED voltage is 3v and you use a 3.3v fixed source and you want 100ma, then you need a resistor value:
    R=(3.3-3)/0.1=0.3/0.1=3 ohms

    If you might see a variation in forward voltage of -0.1v then the current will rise:
    I=(3.3-2.9)/3=0.4/3=133ma

    So you see we picked up an extra 33 percent increase in current due to the possible variation of the LED voltage. To remedy this, we increase the resistor value slightly to 3.5 ohms. Now we get slightly less current during normal operation but if the forward voltage goes down by 0.1v now we only see a 14 percent rise in current.
    So we choose a resistor value that is between the required value for the normal current and the required value for the current if the forward voltage dropped a little.
    This is why it is better to use a constant current drive, because then the LED always gets the right current.

    So in summary, there are two main issues you have to deal with:
    1. The rise in temperature due to the power dissipation and the available surface area.
    2. The change in LED forward voltage due to the characteristic voltage range and the temperature change.

    Because of #2 above you may have to hand select the resistor value for every LED you install because that resistor value has to be chosen to keep the LED from burning out yet not be too underdriven. Of course you also have to have a way to keep the resistor exposed to free air flow so that it stays cool too.

    The amount of voltage headroom is also important to think about because you can not have the same headroom voltage as the variation of LED voltage will be. For example, if the LED is 3v and you use a 3.1v source and there is a chance that the LED voltage can rise to 3.1v or down to 2.9v (plus or minus 0.1 volts) then you will see the LED go either dim or burn out with any seemingly good resistor choice. You should have at least three times the expected variation of LED voltage or better. You dont want to go too high either though or else the circuit becomes very inefficient.
     
    Last edited: Aug 11, 2015
  11. blocco a spirale

    AAC Fanatic!

    Jun 18, 2008
    1,440
    368
    To quote the application note: "The circuit can work for a wide input voltage from 0.9 V to 5 V".

    The output current is automatically adjusted (within limits) to maintain 0.5V across R2 independent of input voltage.
     
  12. Alex Jurtan

    Thread Starter New Member

    Jul 14, 2015
    14
    0

    Thank you for this detailed information! But you mentioned using different LEDs. I plan on using the same LED, the only thing that changes are the batteries. It is for a flexible flashlight that can change in size and depending on that, the amount of batteries you put in.
    The heat is definitely a problem and your rule of thumb shows that I need a proper heatsink. For my design it is not ideal because I want to keep it as small as possible.


    Yes I have seen that it has a wide range of input voltage. But how do I make sure that the LED always gets 3.1V@100ma? Because the input will change from 1.5V to 4.5V, due to my flexible flashlight construction.
     
  13. blocco a spirale

    AAC Fanatic!

    Jun 18, 2008
    1,440
    368
    You only need to supply a fixed current to the LED, the forward voltage will take care of itself.

    For 100mA; R2=5Ω
     
  14. Alex Jurtan

    Thread Starter New Member

    Jul 14, 2015
    14
    0
    Hi,
    I made a new SCHEMATIC with the TPS61200 and the same LED. I have some questions. (Sorry for any simple or repeated questions, just want to make sure I understand this and it would work)

    1. Would this Schematic work with an input from 0.9V-5V and have a constant bright LED? No matter what input voltage? This is the part that keeps me puzzled. How can it know how much voltage the LED needs?
    2. Do I need a resistor in front of the LED as well?
    3. How do I know if the values of capacitors and inductor are correct?
    4. The LED has a forward current of max 150mA. So maybe I could change the resistor to make it around 125mA. BLOCCO how did you calculate this?

    Thanks everyone for helping. Hope I am not asking too much.

    LED TPS61200 SCHEMATIC.JPG
     
  15. MikeML

    AAC Fanatic!

    Oct 2, 2009
    5,450
    1,066
    You still don't understand. You need a circuit that feeds the LED 100mA, PERIOD!; the voltage will be whatever the voltage needs to be. Some LEDs you pick out of the bin have a Vf of 2.9V, some may have a Vf of 3.5V. A particular LED that has a Vf of 3.1V at 25degC might have a Vf of 2.9V at 60degC. Got it?
     
  16. blocco a spirale

    AAC Fanatic!

    Jun 18, 2008
    1,440
    368
    Trust the application note, it addresses exactly the thing you are trying to achieve. TI have done all the design work for you.

    R2 = 0.5/iLED
     
  17. MrAl

    Well-Known Member

    Jun 17, 2014
    2,438
    492

    Hello again,

    You say you are using the *same* LED but your first post shows five PC boards with *five* LEDs. So you seem to be using five (or more) LEDs. If this is true, then all that i had said before still applies. It does not matter that they are all the same part number because LEDs vary from part to part, even with the same exact part number.
    If you really are using only one LED period then that's different. You just need to adjust for that one LED. But if you are doing several boards then you need to adjust for every individual LED unless you use a constant current driver and then the driver will take care of the housekeeping.

    Also, i see from some other posts that you may not yet grasp the concept of driving an LED in different ways. An LED is not like a light bulb where you choose the voltage to match the bulb and all is well and good. For say a 12v bulb you apply 12 volts and you are done with it. But with an LED, you apply 100ma and you are done with it. See the difference? For one you apply a constant voltage (12v) and the other you apply a constant current (100ma). You never need to nor are even able to set both current and voltage simultaneously; you set one, then let the other be what it may be as the device nicely goes to it's normal operating level.

    The voltage of the LEDs may vary slightly during normal operation and over the life of the LED. The voltage may be slightly different for different LEDs of the same part number too. This means a constant current driver will always be able to drive it properly. If however you chose to use a constant voltage driver with dropping resistor, then you need to think about the design a little carefully by planning ahead for what might happen to the LED characteristic voltage over time.
     
Loading...