Help with transistors - I need lower-voltage-drop ones

Discussion in 'The Projects Forum' started by beammy, Apr 16, 2016.

  1. beammy

    Thread Starter New Member

    Jun 21, 2015
    11
    1
    Hi,

    I have a project which controls an "array" of LEDs using a microcontroller. Because the MCU can't output nearly enough current, I use transistors to amplify the current.

    Currently I use 2222's (NPN) to switch between either letting current flow out of a group of LEDs' cathodes, or not. And I use 327's (PNP) to switch between either letting current flow into a group of LEDs' anodes, or not.

    This works great, but as my project is battery powered, I want to switch to use different transistors which perform the same function, but with the smallest voltage drop possible, to reduce power wastage. (I'm using a switching regulator for just the LEDs' supply voltage, so if I can reduce this, I'd get longer battery life.)

    Unfortunately, my knowledge of transistors is quite basic. Looking on Digi-Key, they don't list transistor voltage drop as a spec, so I'm not quite sure where to start. I'm embarrassed to say it, but I don't really yet understand why 2222's and 327's are commonly considered a good choice for controlling LEDs any more than any other transistor.

    I need each NPN transistor to supply up to 350mA peak, 35mA continuous. And I need each PNP to supply 100mA continuous.

    Does anyone have any recommendations, or pointers in terms of what other specs I should be looking out for?
     
  2. crutschow

    Expert

    Mar 14, 2008
    12,990
    3,226
    For really low drop and best efficiency you could use MOSFETs with a low ON resistance which require no gate current to turn on.
    You need "logic-level" types that have their ON resistance [Rds(on)] rated at a Vgs equal to or less than the μC output (supply) voltage as shown in the transistor data sheet specifications.
    This is not the MOSFET threshold voltage (where the transistor just starts to barely conduct) which is always much lower.
     
    Last edited: Apr 16, 2016
  3. beammy

    Thread Starter New Member

    Jun 21, 2015
    11
    1
    Thanks for that. I've done some looking on Digi-Key and some googling of terms I didn't know, and I'm starting to understand.

    To check that I've got it right: with a MOSFET, Vgs controls the amount of current that may flow from the drain to the source? So could I, using a resistive divider, fine-tune the voltage I apply as Vgs in order to fine-tune the amount of current that may flow from drain to source? (Is it in fact a constant current-limit that I'd be controlling, or more like a resistance that I'd be controlling?)

    So, it sounds like this would be a good replacement for my NPN transistor. However I'm still not quite sure what to do about replacing my PNP transistor, which I've been switching on by drawing some current out of the base. Its collector and emitter sit in the LEDs' positive power supply, at a higher voltage than my MCU can output, so replacing it with a MOSFET as we've talked about, I wouldn't be able to induce a positive Vgs. Is there another kind of MOSFET that I could use which I could instead switch on by pulling a voltage low?
     
  4. Bordodynov

    Active Member

    May 20, 2015
    637
    188
  5. Alec_t

    AAC Fanatic!

    Sep 17, 2013
    5,773
    1,103
    No. Use the MOSFET simply as a switch. Give the gate voltage all you've got (but less than the specified limit, ususally ~20V), to turn the MOSFET fully on. Control current with a series resistor in the LED current path.
     
  6. beammy

    Thread Starter New Member

    Jun 21, 2015
    11
    1
    Cool, that makes sense.

    I've answered my own earlier question and realised that there are N-channel and P-channel MOSFETS, which could be thought of as similar to NPN and PNP transistors.

    I'd actually prefer not to use a series resistor to control current to my LEDs, because I'm trying to minimise any voltage drops. In order to consistently set current with a series resistor, I'd sort of need to give the resistor a relatively large voltage drop (by raising my supply voltage), so that if there's some variation in how much voltage my LEDs use up (remember I'm using a matrix), they wouldn't have too much variation in brightness.

    In this regard, PNP BJTs were great - I could set Ic accurately and consistently by setting Ib (I hope I understood that correctly - it certainly seemed to work).

    So perhaps I should switch my old NPN to an N-channel MOSFET, but keep my old PNP to allow me to set the current accurately. Does this sound like the best option? Even if it is, though, I'm still struggling to determine which PNPs would have the lowest voltage drop - I can't actually find this info in any transistor datasheets. I'd be looking for a graph showing Vce at different Ic or Ib's, right? Or have I been looking for the wrong thing?
     
  7. MrChips

    Moderator

    Oct 2, 2009
    12,421
    3,356
    You are looking for the wrong thing.

    The only reason for reducing the voltage drop of the transistor is if you want to operate your circuit at the lowest possible battery voltage.

    The LED needs the current to be regulated. It doesn't matter whether you use a resistor, BJT or MOSFET to regulate the current.

    To regulate current you have to have excess voltage. The current is regulated by dropping excess voltage across a resistive device. By using a transistor to regulate the current, the transistor still becomes a resistor of equivalent resistance as a fixed resistor. Power will always be wasted in this resistor.

    A proper ideal current source has infinite series resistance.

    If you reduce the supply voltage, the value of the series resistor has to be reduced accordingly. When you get to the point where the series resistance is eliminated the LED current will be determined by the I-V characteristics of the LED itself (plus any other devices in series). Basically, Ohm's Law still has to apply.

    The other way to control the average current without wasting power is to use pulse-width modulation (PWM).
     
    absf likes this.
  8. dannyf

    Well-Known Member

    Sep 13, 2015
    1,771
    358
    The transistors you already have are more than capable of doing that.

    Going beyond that yields little incremental benefits.
     
  9. beammy

    Thread Starter New Member

    Jun 21, 2015
    11
    1
    The way that I've been limiting LED current up until now is with my BC327 PNP transistors. I feed them a fixed Ib, and they limit Ice, doing whatever voltage drop is necessary - but this stops working so well when the voltage drop gets small.

    I just did some tests so that I have some actual numbers. I feed a BC327 5.3mA Ib. It limits Ice to 80mA, doing whatever voltage drop is necessary. Depending on how much voltage I'm applying to the system (don't worry, I made sure Ib was constant), it can be doing a 0.8V drop or a 0.2V drop, and Ice is still approximately the same 80mA. This is really useful for setting LED current. However, somewhere between 0.2V and 0.1V drops, Ice starts to go down. At a 0.1V drop, it's down at 72mA.

    So, I can use a BC327 for current-setting with a voltage drop of only around 0.15V. That's actually not bad. I'm still hunting for better options though - for PNP BJTs. I've realised that this info can be divulged from transistor datasheets, though it can be kind of tricky because there are sort of 3 "dimensions" to the relevant graph - basically Ib, Ice, and Vce.

    Bordodynov, that ZTX690B looks good. Hope I can find an SMD version (I should have said, I'm looking for SMD).

    In any case, it looks like I'll be swapping my NPN BJT for a MOSFET.

    Thanks a lot for all the help.
     
  10. Alec_t

    AAC Fanatic!

    Sep 17, 2013
    5,773
    1,103
    The problem with that method, which you may not have found yet, is that the result is very temperature dependent.
     
    absf likes this.
  11. dannyf

    Well-Known Member

    Sep 13, 2015
    1,771
    358
    That may work for a one-off, but that's not the right way to design a circuit: it should be designed so that the characteristics of a circuit is dependent on its passive components. in your design, Ic is beta dependent, and beta depends on a host of parameters, like temperature, etc.

    You should have used the transistors as a switch - in that case, the current is dependent on the resistor in serial with the led.
     
    absf likes this.
  12. beammy

    Thread Starter New Member

    Jun 21, 2015
    11
    1
    Ah. Thanks, I didn't know that.

    That's a pain, because most LEDs' rated voltage drop according to their datasheet can vary by easily 30%. So, limiting current with a series resistor, I'd really want my supply voltage quite a bit higher so that the voltage over the resistor is consistent and the resistor can supply consistent current. So I'm going to be wasting a lot of voltage on resistors.

    Is there really no better way, other than PWM?
     
  13. dannyf

    Well-Known Member

    Sep 13, 2015
    1,771
    358
    For leds, 30% variation is nothing to naked eyes. better yet, 300% variation is nothing.

    That's the least of your worries.
     
Loading...