Transistor Behaviour

Discussion in 'General Electronics Chat' started by philipm, Jun 20, 2013.

  1. philipm

    Thread Starter Member

    Jun 27, 2012
    47
    3
    I've been working on a simple FET/NPN constant current driver. It should use an NPN transistor, but instead, I have accidentally ordered and soldered in a PNP.

    Please see attached image. The left circuit is how it should be. The right circuit is how it is on my board. Surprisingly it works and I only noticed when I went to reorder the components to build more!

    I can't quite get my head around the theory. Does the PNP play a part in the circuit? Or is the circuit equivalent to it not being there?

    If it's not playing a part, then I'm very surprised that the current flow through the LED (approx 2A instead of 700mA) didn't fry it!

    Many thanks.
     
  2. crutschow

    Expert

    Mar 14, 2008
    13,003
    3,232
    The PNP has the collector-emitter junction forward biased, so the circuit is similar to replacing the PNP with a forward biased diode between R8 and R1.
     
  3. #12

    Expert

    Nov 30, 2010
    16,278
    6,791
    Actually, the LED should be smoking.

    The original circuit is a constant current generator where the MOSFET is normally on and the NPN transistor uses the voltage across R3 to tell the MOSFET when to stop. With a PNP transistor, nothing is telling the MOSFET to stop. Do not try to replace the LED while it is still a liquid.
     
  4. philipm

    Thread Starter Member

    Jun 27, 2012
    47
    3
    #12 - yes that's my thoughts on it too. VCC in this case is an unprotected LiPo cell. The Rds(on) of the FET is tiny and R3 is 1 ohm. Every factor I consider points towards uncontrolled current inrush (of about 2A) resulting in LED failure. Yet, it's been quite happy to flash the LED for 100ms every second.

    The LED should be about around 140 lumen at the desired 700mA current level. I can't measure it accurately, but it appears less bright than the 8x15 lumen LED device I compared it against.

    Looking at it with a thermal camera, the FET is getting quite toasty (again another indication that the FET is not fully on). In fact the split appears about right with the LED sinking most heat, then the FET, then the resistor.

    I'm perplexed!
     
  5. WBahn

    Moderator

    Mar 31, 2012
    17,737
    4,789
    I don't see where you are getting your 2A figure from. Is that a measured value?

    The FET should be getting a Vgs of about a Si diode drop -- perhaps a bit more because the CB junction of the PNP may not act as a particularly good diode.

    Could you measure the voltages at each of the nodes when the LED is ON and post those?

    If you are only abusing it for 100ms each second, then on average it is only getting 200mA and that shouldn't present a problem from a thermal standpoint. The concern would be its ability to withstand the transient current for that 100ms.
     
  6. Ron H

    AAC Fanatic!

    Apr 14, 2005
    7,050
    657
    We need the component values and part numbers to fully analyze the circuit.
    I agree that the MOSFET should be off, with only about 0.7V Vgs. I can't see how the LED could be on.
     
  7. Ramussons

    Active Member

    May 3, 2013
    557
    92
    It's possible that the PNP transistor legs are wrongly connected. There does'nt seem to be any other way to (over)drive that Q5.

    Ramesh
     
  8. philipm

    Thread Starter Member

    Jun 27, 2012
    47
    3
    I've attached an image of the equivalent circuit (when, as you suggest the PNP is acting as a forward biased diode across the FET's G,S pins). This is starting to make a bit more sense.

    Here is the datasheet for the N-MOSFET
    And for the PNP

    The voltages applied when "on" are:
    VCC = 4.2v-3.7v approx (fairly fresh LiPo cell)
    UC = 3.0v (microcontroller powered by a 3.0v regulator)

    The 2A current I predicted is approximated using the LED voltage drop (2.1v) and series resistor (1 ohm) powered by about 4v (online LED calculator).
     
  9. Ron H

    AAC Fanatic!

    Apr 14, 2005
    7,050
    657
    See the attachment, which is snipped from the MOSFET datasheet.
    This means that, if Vgs<1.2V, the transistor will conduct less than 250uA.
    You will be lucky to get 2A, even if you use an NPN, with this MOSFET.
    Your calculated current is wrong. Id=Vbe/R3. Vbe≈0.7V, so Id≈0.7A.
     
  10. upopads

    Active Member

    Dec 18, 2007
    42
    0
    Guys what does "UC" designate here?
     
  11. upopads

    Active Member

    Dec 18, 2007
    42
    0
    Nevermind!
     
  12. upopads

    Active Member

    Dec 18, 2007
    42
    0
    Very humorous results here, thanks for the analysis ron and cruts.
     
  13. philipm

    Thread Starter Member

    Jun 27, 2012
    47
    3
    Hi Ron,
    Yes, the 700mA is the target current. 2A is what it would be if the MOSFET was just fully on (in effect it is then just an LED and resistor).

    250uA is definitely way off what is happening in reality.

    Could the Base-Emitter be playing some sort of role here too? At a guess some sort of NP (base-emitter) breakdown, but at a lower voltage due to the PN (collector-base) conducting? Is that possible?
     
  14. WBahn

    Moderator

    Mar 31, 2012
    17,737
    4,789
    UC = μC = microcontroller.
     
  15. Ron H

    AAC Fanatic!

    Apr 14, 2005
    7,050
    657
    BJT BE breakdown is almost always >5V, regardless of part number.
    I think your PNP is bad, or connected incorrectly.
    I would just forget the PNP and put in an NPN as planned.
     
  16. WBahn

    Moderator

    Mar 31, 2012
    17,737
    4,789
    It could be a lot of things, but to make any kind of a reasonable guess we really need some actual measurements. Turn the thing on long enough (a second or so?) to get a good voltage reading at one point. Then turn it off and let it cool for a few seconds. Then turn it on to get a good voltage reading at another point. Repeat until you can post actual voltage readings at all of the nodes.
     
  17. GopherT

    AAC Fanatic!

    Nov 23, 2012
    6,026
    3,790
    Micro controller.

    More correctly than UC is, uController or, more correctly, μController.
     
  18. Ron H

    AAC Fanatic!

    Apr 14, 2005
    7,050
    657
    If source and drain are swapped, you will get about 1.5 - 1.6 Amps, due to the fact that the intrinsic body diode will conduct.
     
  19. philipm

    Thread Starter Member

    Jun 27, 2012
    47
    3
    The experimental results are in. Measured voltages are indicated on the attached diagram (when the LED is on). Somehow, we're seeing just over 500mA constant current (0.52V across the 1 ohm resistor(R1))!

    I've rechecked the datasheets versus actual board layout. I've also checked for shorts. All is correct.

    The PNPs should arrive Monday/Tuesday, but I'm still keen to work out what is going on!
     
  20. Ron H

    AAC Fanatic!

    Apr 14, 2005
    7,050
    657
    Your PNP is working like an NPN. What is the part number?
    THIS IS IMPORTANT!
     
Loading...