Discrete Class D Amp For Fun

Discussion in 'The Projects Forum' started by iimagine, Sep 21, 2013.

  1. iimagine

    Thread Starter Active Member

    Dec 20, 2010
    129
    9
    Assuming that my 7 transistors triangle wave generator works as expected (As I am still waiting for 2N2369's and 2N5771's to arrive), I need your help on completing this amp for fun and educational purpose because I am not quite sure if this is right

    Thanks in advance
     
  2. iimagine

    Thread Starter Active Member

    Dec 20, 2010
    129
    9
    Here is the .asc file
     
  3. GopherT

    AAC Fanatic!

    Nov 23, 2012
    6,061
    3,823
    Triangle wave, input signal, Pwm output, reconstituted input signal with amplification. Looks good to me.
     
    iimagine likes this.
  4. tindel

    Active Member

    Sep 16, 2012
    568
    193
    Full disclosure: Audio is not my forte...

    A few points...

    Most amplifiers require feedback to remain stable... it looks like your gain at 20kHz is only about 2.7... so maybe it doesn't matter much with this small of gain.

    It appears that your input will have 1.4V DC offset... how do you plan on achieving this?

    You have a a DC offset on your speaker... making the average output power be 1W @ 20kHz... I'm not sure that most speakers can handle that.
     
    iimagine likes this.
  5. iimagine

    Thread Starter Active Member

    Dec 20, 2010
    129
    9
    It appears that your input will have 1.4V DC offset... how do you plan on achieving this?

    I have completely forgot about this :D, thanks for pointing it out. This can be solve by adding a cap and two resistors. I have updated my circuit to include this.

    You have a a DC offset on your speaker... making the average output power be 1W @ 20kHz... I'm not sure that most speakers can handle that.

    I need help with this, I'm not understanding at all, can you please elaborate? Is there any reading material on this matter you can point me to?

    Thank you for helping
     
  6. #12

    Expert

    Nov 30, 2010
    16,318
    6,818
    It's simpler than it sounds. Speakers don't like DC because it displaces the cone and heats the windings. A 60 watt speaker might not care, but a 3 watt speaker would suffer.
     
    iimagine likes this.
  7. Experimentonomen

    Member

    Feb 16, 2011
    331
    46
    Simple solution, add a cap in series with the speaker, around 470-1000µF should do.
     
    RichardO and iimagine like this.
  8. RichardO

    Well-Known Member

    May 4, 2013
    1,235
    384
    I ran the Spice simulation and noticed that the drive voltage to the FET's does not go to ground. I have not analyzed why this is happening.

    Also, the circuit would be more efficient if the FET's were run common source instead of as source followers. This does, however, need different FET's than the ones you have in the simulation. For this to work, the drive has to go close to both the power rail and ground and the FET gate thresholds must be around 1 to 2 volts.

    One final comment. You may want to get rid of the large coupling cap to the speaker. You can do this by driving the ends of the speaker 180 degrees out of phase using 2 output stages. This is commonly called a bridge amp drive. An added benefit of the bridge amp is you can supply 4 times the power to the speaker with the same power supply voltage.
     
    iimagine likes this.
  9. Experimentonomen

    Member

    Feb 16, 2011
    331
    46
    Or using two supplies with a midpoint centertap, though you loose the automatic dc protection you get from the output cap.

    Also the reason your circuit doesent pull all way to ground is that it is more or less just your average class b amplifier circuit, wheres what you want is a comparator.

    Look at the philips UcD class d circuit for inspiration.
     
    SgtWookie and iimagine like this.
  10. iimagine

    Thread Starter Active Member

    Dec 20, 2010
    129
    9
    I have been looking at the Phillips UcD circuit but could not get anything out of it. Its a little too complicated for me :confused:

    Could you provide me with a comparator circuit for this please? Thanks
     
  11. Experimentonomen

    Member

    Feb 16, 2011
    331
    46
    I already did, the philips UcD one. Or use a ic like the LM311 or the MAX913/TL1016.
     
  12. iimagine

    Thread Starter Active Member

    Dec 20, 2010
    129
    9
    I am a little confused, are you saying that I should use the philips UcD comparator or LM311 for the comparison of the triangle and the input signal?
     
  13. Experimentonomen

    Member

    Feb 16, 2011
    331
    46
    Yes! There is no other way to do it.
     
  14. iimagine

    Thread Starter Active Member

    Dec 20, 2010
    129
    9
    I managed to drive the gate at ~0.4-5v. The output is much better now, although I dont understand how it can be slightly above 5V. Please advice

    Thanks :)

    I'm working on driving the ends of the speaker 180 degrees out of phase using 2 output stages as richardO suggested
     
  15. Experimentonomen

    Member

    Feb 16, 2011
    331
    46
    Possibly resonance of the output lc filter.
     
  16. iimagine

    Thread Starter Active Member

    Dec 20, 2010
    129
    9
    Thank you Experimentonament for sticking around and help

    Added phase shift but now facing problem with ripple. Please advice
     
  17. Experimentonomen

    Member

    Feb 16, 2011
    331
    46
    I built a little bridged class d amp once using two UC3710 igbt gate driver ics and a LM311 comparator ic.

    These gate driver ic's in TO220-5 package makes nice output stages for a low power class d.

    As for a more practical class d amp, i built this: https://www.youtube.com/watch?v=asC3zPaB9mY

    Uses a MAX913 comparator, a IR2010 halfbridge gate driver and IRF540 mosfets for the output stage.
     
  18. iimagine

    Thread Starter Active Member

    Dec 20, 2010
    129
    9
    Very nice! Can you share those schematics? I would like to build one to study
     
  19. iimagine

    Thread Starter Active Member

    Dec 20, 2010
    129
    9
    Solved that spiky ripple problem! :D

    I'm only doing this to study, learn and to see if there is any other approach, alternative way of doing things :). Any one have any ideas or suggestions please share. Although I do not think many here cares about this little project of mine too much too bother, but its alright I can always turn to Experimentonomen, RichardO and #12 for help :D.... I think
     
  20. Experimentonomen

    Member

    Feb 16, 2011
    331
    46
    iimagine likes this.
Loading...