Simple BJT-based SPDT?

Discussion in 'General Electronics Chat' started by xox, Oct 11, 2017.

  1. xox

    Thread Starter Member

    Sep 8, 2017
    307
    63
    This is an emulated SPDT switch I came up with using MOSFETS. Does the design look okay for the application (in this case, powering two LED's)?

    mosfet_SPDT.png

    Anyway, I was wanting to duplicate the functionality using BJT's, but so far all of my ideas have failed pretty miserably. Worse still, an online search only yielded a mishmash of complicated circuits. I'm guessing the reason for such an increase in complexity is mainly due the fact that BJT's are current-controlled versus the MOSFET's voltage-driven operation. Does that mean that there's no way to do this with, say, two BJT's?
     
  2. crutschow

    Expert

    Mar 14, 2008
    19,754
    5,525
    Yes, that tends to make the biasing more complicated.
    Here's the LTspice simulation of a circuit using two PNP BJT's.
    It adds two diodes to properly bias the second transistor.

    upload_2017-10-11_1-51-35.png
     
    xox likes this.
  3. xox

    Thread Starter Member

    Sep 8, 2017
    307
    63
    Impressive! Never thought of diodes being used like that either. Very elegant solution. :)
     
  4. xox

    Thread Starter Member

    Sep 8, 2017
    307
    63
    Another interesting thing about your design is that the current draw of the BJT circuit is much lower than one might expect. In fact, it's practically the same as the MOSFET switch.

    SPDT_pair.png


    Typically you hear that the bipolar is much less efficient than the field-effect transistor. Good to know that isn't always true!
     
    Last edited: Oct 11, 2017
  5. crutschow

    Expert

    Mar 14, 2008
    19,754
    5,525
    Well you do have to add the base current into the calculation, which is typically about 10% of the maximum collector current.
    And the fully ON saturation voltage of a BJT can be around 100mV, but a low on-resistance MOSFET can be only a few mV.

    But why did you increase the top BJT's base resistor from 3kΩ to 15kΩ?
    15K is too large to insure full saturation of the transistor (the rule of thumb is that the base current should be about 10% of the collector current for complete saturation).
     
    xox likes this.
  6. xox

    Thread Starter Member

    Sep 8, 2017
    307
    63
    I see, so for an LED or other similarly low-current applications it doesn't necessarily amount to much, but otherwise MOSFETS would be a better choice.

    Oh snap, I didn't realize it could adversely affect the operation like that (just trying to save a few milliamps). Thanks for the admonishment.
     
  7. crutschow

    Expert

    Mar 14, 2008
    19,754
    5,525
    You saved about a mA.

    If you want to minimize the base current you could probably get by with 5% base current at those low collector currents.
    That would give a maximum value for the top base resistor of (5-.7-.7) / 12mA(.05) = 6kΩ
    and a bottom base resistor of (5-.7) / 12mA(.05) = 7.17kΩ
     
    bug13 likes this.
  8. xox

    Thread Starter Member

    Sep 8, 2017
    307
    63

    Okay, so just out of curiosity, does the transistor's datasheet provide any information that could be used to calculate a more precise resistor value?
     
  9. crutschow

    Expert

    Mar 14, 2008
    19,754
    5,525
    More precise than what?
    The data sheet shows the conditions for the measurement of the ON saturation voltage and that's typically done at a base current that is 10% of the collector current, so that's the accepted value if you want minimum saturation voltage.
    They also give curves of "typical" saturation voltages but you don't want to use typical values in a good design.
     
  10. xox

    Thread Starter Member

    Sep 8, 2017
    307
    63
    Alright, so if I understand you correctly, "typical" saturation voltages are to be avoided simply because they could easily place you "just" into cutoff or saturation depending on variations due to manufacturing imperfections, changes in temperature, etc, and thus the 10% rule pretty much guarantees that you'll be in the right state?
     
  11. dl324

    AAC Fanatic!

    Mar 30, 2015
    7,083
    1,628
    You do it because you're not guaranteed to get a typical device, so a conservative designer would use the appropriate min/max value for critical parameters.
     
    xox likes this.
  12. xox

    Thread Starter Member

    Sep 8, 2017
    307
    63
    Right, got it!
     
Loading...