Help me choose a transistor for this circuit

Discussion in 'The Projects Forum' started by campeck, Sep 7, 2010.

  1. campeck

    Thread Starter Active Member

    Sep 5, 2009
    194
    3
    This is another project I am making for my movie making friends.
    It is a high power RGB fader. This should put out up to 1,200 lm when displaying white.
    The RC network before the opamps will be charged via a Basic Stamp command and the opamp outputs go to the transistors. I could just pick an NPN of a high enough current value to work but was wondering if there is anyway to do better than that? What other things should I consider? Heat needs to stay as low as possible. The switch will ground the RC network through a resistor allowing the User to fade all LEDs out. And the way I write the program should allow the LEDs to fade back on once released. Anyone know where to get a 3pdt switch?

    Thanks guys!

    [​IMG]
    [​IMG]
     
    Last edited: Sep 8, 2010
  2. M1EUF

    New Member

    Sep 1, 2010
    6
    0
    I would personally use a Mosfet in a TO-220 case, it will be a bit of an overkill for this type of circuit but the losses would be less due to the low resistance between source and drain when turned fully on. Look for an IRF series device capable of handling 30V. You will probably struggle to find anything rated at less than 5A. I would go for the IRF510 as I have plenty in the junkbox. If you do decide to use a bipolar transistor, I would choose something like TIP31, TIP41, 2N3053, etc. Even better, just have a look through your junk box/scrap equipment for transistors and 'suck it and see'; I reckon it would be difficult to find a transistor that DOESN'T work in this application.
    Since your controlling with a basic stamp, why not use PWM (pulse width modulation) to dim the LEDs and get higher efficiency?

    Regards, John
     
  3. campeck

    Thread Starter Active Member

    Sep 5, 2009
    194
    3
    Will the Mosfet slowly turn on with increasing voltage from the opamps? I like the idea of low voltage drop. Especially since I forgot to subtract it in my LED resistor calculation. lol


    Because I need to check switches and buttons at the same time. This way I can put a voltage on the RC network and then move on to checking other inputs.

    Thanks!
     
  4. M1EUF

    New Member

    Sep 1, 2010
    6
    0
    'Will the Mosfet slowly turn on with increasing voltage from the opamps?'

    Yes- think of the Mosfet as a voltage controlled variable resistor. With around 10V on the gate it will be fully on, but with lower voltages it will be somewhere between of and fully on. You could experiment by placing a 10k variable resistor across a 12V power supply, wiper of the pot to the gate terminal, and vary the resistance to see how your LEDs look. When feeding with the op amps you could try different RC values to get the timing that you want.

    Regards, John
     
  5. campeck

    Thread Starter Active Member

    Sep 5, 2009
    194
    3
    You say 10v...
    The basic stamp will only output 5v. Of course I could use a transistor to turn the mosfet on if need be. But I just thought of something...

    The cap in the RC network gets charged to an exact voltage by the BasicStamp2 depending on the command I send. Lets say 3.2v. The reason for the opamps is because they are Hi-Z and while I go off in the program and do other things the cap needs to stay charged to 3.2v until I can come back and recharge it. (lets say a safe zone of 5sec.)

    But If I am using Mosfets do I need the opamps? Mosfets are Hi-Z aren't they?
    If the Mosfet will be fully turned on at 5v and be Hi-z and keep the cap charged then I can simplify my parts count.

    Thanks!
     
  6. Audioguru

    New Member

    Dec 20, 2007
    9,411
    896
    Then use a modern "logic level" mosfet instead of the ancient IRF510 one. It fully turns on when its gate is 4.5V or more.

    The cap in the RC network gets charged to an exact voltage by the BasicStamp2 depending on the command I send. Lets say 3.2v. The reason for the opamps is because they are Hi-Z and while I go off in the program and do other things the cap needs to stay charged to 3.2v until I can come back and recharge it. (lets say a safe zone of 5sec.)

    with logic-level Mosfets then the opamps do not do anything useful. The opamps won't work properly anyway unless they work properly without an additional negative supply voltage.
     
  7. campeck

    Thread Starter Active Member

    Sep 5, 2009
    194
    3
    So you are saying with logic level mosfets I need no opamps? good...

    Does the LM324N need negative supply? Because I have built this before using small transistors and 70ma LEDs and that opamp.

    http://www.youtube.com/user/campeck944#p/u/3/ASHEzlgiUKU

    I just need to make it BIGGER!!

    Thanks for the help guys. I have never used a Mosfet in a circuit before so i am sure more questions will arise. Like....do they need gate resistors? I would assume not...
     
  8. retched

    AAC Fanatic!

    Dec 5, 2009
    5,201
    312
    Boy is your wife gonna be pissed! You just broke her favorite LED diffuser!

    ;)

    [​IMG]

    And yes, using logic level MOSFETs will allow you greater control over without using an extra FET driver.
     
  9. campeck

    Thread Starter Active Member

    Sep 5, 2009
    194
    3
  10. Audioguru

    New Member

    Dec 20, 2007
    9,411
    896
    Then you don't need and don't want opamps.

    The LM324 and its smaller sister the LM358 are the slowest opamps ever made. They will work without a negative supply but they are too darn slow so the ramping they cause in the Mosfets creates a lot of heat. There are many much faster opamps that also work from a low voltage supply without a negative supply like the MC34071, mC34072 and MC34074.
     
  11. campeck

    Thread Starter Active Member

    Sep 5, 2009
    194
    3

    Well the very nature of this project requires slow ramping. So things are going to get hot. It doesn't help that this is in a studio where it is already super hot and muggy...I would use PWM to increase efficiency if I wasn't scared of the flicker being caught on camera and the extra difficulty.

    Because I could have the BS2 control a digipot controlling the duty cycle of a 555 oscillator which would control the brightness...
     
  12. Wendy

    Moderator

    Mar 24, 2008
    20,766
    2,536
    Actually, not. BJT are linear, but MOSFETs are generally not used in their middle area, they work best either on or off. The heat generated in their not so linear state is considerable, and will likely burn em out.

    PWM is a much better way to go, it is more power efficient by far with MOSFETs, and you can get the same effect you are looking for, including the slow ramping.

    LEDs, 555s, Flashers, and Light Chasers
    Chapter 5

    The reduced heat and power savings are worth the extra circuitry.
     
  13. retched

    AAC Fanatic!

    Dec 5, 2009
    5,201
    312
    If this is a filmed event, you must be sure that your PWM frequency is not detectable by the camera.

    There are many "flicker frequencies" and the only way to know for sure if the camera will see it, is to test it with that particular setup.
     
  14. Wendy

    Moderator

    Mar 24, 2008
    20,766
    2,536
    Actually just go very high speed, say around 5Khz or faster. It will not work with slow op amps such as the LM324, but there are many parts (such as the LM393 or LM339 comparators) that will handle it.

    Is there any way you can test it? It just occurred to me that really fast film may catch the peaks in the lighting, which is the same as full lighting, and I'm off base.

    If you need to go with a linear component, I would not use a MOSFET, as they are not linear like a BJT is. You could go with feedback to an op amp with a MOSFET to make it more linear, but BJTs also have better case styles for heat dissipation, which is a consideration.

    It's been a long day, and I'm pretty tired. When is your deadline?
     
    Last edited: Sep 9, 2010
  15. Audioguru

    New Member

    Dec 20, 2007
    9,411
    896
    Sorry, I didn't notice that you have a series resistor charging or discharging a capacitor at the input of the opamps. Then the LEDs are supposed to slowly ramp bright or slowly ramp dim.

    But your transistor drivers have no negative feedback and might abruptly turn fully on and turn fully off because they are common-emitter switches.
    A transistor ramps very well when it is an emitter-follower (the load is connected from the emitter to ground). But an emitter-follower has no voltage gain and then the opamps must have gain so that the 5V from the micro-controller is amplified to almost 12V for the LEDs.
     
  16. campeck

    Thread Starter Active Member

    Sep 5, 2009
    194
    3
    Well if I was going to use Mosfets in a PWM scheme I would not use the RC network and opamps. Those would be replaced with a digipot and 555 with the output going to the Mosfets.

    Yes I could test it. Not sure what camera they are using but they will work with me to achieve the desired results.

    Feedback to an opamp with a Mosfet? Not sure what you mean.
    If I need to stay linear then I will use a BJT. Which brings me back to my original question...Any suggestions on a BJT?

    So I should switch the LEDs from the collector to emitter? ok.
    And you say when connected like that my 5v signal wont switch 12v? Why not? I thought that's what transistors are for!
    You mention negative feedback to the transistors. Never done that before. How does that work?

    If the opamps need gain to drive the transistors ( why again? ) could I not just set them up at 2 gain and get 10v when I input 5v? I could hook them up to the 12v supply instead of 5v if need be. I forgot the supply in the schematic. (it was going to be 5v)

    And the only reason they are going to ramp is because of programming. My program will put a voltage on the cap through the series resistor which the opamp will buffer, then send that to the transistor turning the leds on. The ramping will come when I continuously go back to the cap and put increasingly higher voltages on it. Otherwise the circuit will just sit there at the last voltage. (until the cap is discharged through the opamp and leakage current.)
     
  17. Audioguru

    New Member

    Dec 20, 2007
    9,411
    896
    The opamps cannot produce enough current to drive transistors that have a load as high as 700mA. So you need darlington transistors like the TIP120.

    An emitter-follower was recommended as a driver for slow ramps, not as an on-off switch. It has no voltage gain. Instead it even has some voltage loss. If the opamp feeds +5V to a transistor emitter-follower then the emitter is only +4.0V to +4.4V. A darlington transistor has a 1.5V loss so an input of +5V produces an output of only +3.5V. But you need +12V.

    An emitter-follower already has 100% negative feedback.

    When an LM324 opamp has a +12V supply, has a gain of 2.5 and an input of +5V then its output is +10.7V when it has no load. The darlington transistors add their 1.5V loss so the LEDs get only +9.2V instead of +12V and they might not even produce any light.
    The LM324 opamps and darlingtons need a 15V supply to drive the darlington emitter-followers and produce an output as high as +12.2V.

    The D to A converter in your micro-controller will hold the voltage so the resistor and capacitor is not needed. But if you disconnect the D to A converter so it does not hold the voltage then the positive-going input bias current of the LM324 opamp (it has PNP input transistors) will continue charging the capacitor to higher and higher voltages. Then the LEDs slowly get brighter and brighter.
     
  18. campeck

    Thread Starter Active Member

    Sep 5, 2009
    194
    3
    Sounds like a plan. Sucks for the voltage drop though. Looks like my +rail just got bumped up to 15v

    What if i use a common emitter like I have now, but put an emitter resistor on it? Doesn't that increase linearity? never done it before though.

    Oh...

    See above about the emitter resistor hopefully solving this problem.

    The BS2 doesn't have this capability.
    As for the opamp charging the cap. I didn't know that. I will have to make sure when I set the output of the BS2 to Hi-z that the LED's don't start turning on again.

    Ok...

    So It looks Like I have 3 options.

    Use a good single supply opamp into a TIP type transistor as a common emitter with emitter resistor and turn up my positive rail voltage.

    Use a good single supply opamp into a TIP type transistor as a common collector and turn up the positive rail.

    Use a PWM scheme with logic level MOSFETS and test with the camera. (added benefit of being able to use more current with the LED's)

    Am I seeing this correctly?
     
  19. Audioguru

    New Member

    Dec 20, 2007
    9,411
    896
    You must not allow the input of an LM324 opamp to be at a Hi-z to prevent its positive-going input current from making the input voltage float higher and higher.

    I sketched common-emitter and emitter-followers.
     
  20. campeck

    Thread Starter Active Member

    Sep 5, 2009
    194
    3
    Dang. All these limitations and new information and circuit configurations I have never built before is really making me lean toward PWM. I will start experimenting tonight.

    Thank for everyones help. I am sure I will need more of it. :cool:
     
Loading...