Buck simulation

Discussion in 'The Projects Forum' started by renatomassa88, Jun 15, 2011.

  1. renatomassa88

    Thread Starter New Member

    Jun 12, 2011
    8
    0
    Hi, I'm having a problem with this simulation in Proteus.
    I already tried a lot of different components but still doesn't
    work. I also search why isn't working, including a topic in this
    forum,but it stays the same. The output should be 11,25V,
    considering 25V in input and a duty cicle 45%.
    The result in a graphic is aroudn 5V and in the real time
    simulation keeps rising. Can anybody give a hand?
    Thanks
     
    • buck.jpg
      buck.jpg
      File size:
      16.5 KB
      Views:
      140
  2. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    You don't have any feedback method for controlling the output voltage.
    How will it know when to stop?

    I can't tell for certain by the schematic, but I'll have to assume that the signal you are feeding to the MOSFET gate (Q1(G)) is referenced to ground, which won't work - and that is why you start out with ~5v on the output. That signal needs to be referenced to the MOSFETs' source terminal.

    The IRL540 is a logic level MOSFET, meaning that it is considered OFF when Vgs (the voltage on the gate referenced to the source terminal) is less than the threshold voltage (specified as Vgs(th) in the datasheet)
    Datasheet: http://www.vishay.com/docs/91300/91300.pdf
    So, if Vgs<=1v, the MOSFET is off.
    To get the Vgs specification for when the MOSFET is considered turned on, look at the Rds(on) specification; this one has two Rds(on) for two Vgs's.

    Referencing the gate voltage to GND for a high-side MOSFET is a VERY common error for people new to MOSFETs. Gate voltage always has to be referenced to the source terminal.

    Another problem that you will run into is the start-up current through the inductor. If you simply start feeding the gate a 45% duty cycle square wave, a plot of the output voltage over time will look like you cracked a whip. You'll need to start off at a much lower duty cycle, and work your way up to 45%.

    Your load resistor, R1, is 1000k, or 1 MEG. That is a very light load. If your MOSFET Rds(on) is 77 milliOhms when it's turned on, how do you think that R1 will drain enough current to keep the output at ~45% of Q1(D)?

    It won't. After the "crack the whip" oscillations from L1 are over, the MOSFET will charge C1 up to nearly Q1(D).
     
  3. ifixit

    Distinguished Member

    Nov 20, 2008
    638
    108
    Also...

    You are operating the IRL540 N-Channel MOSFET in source-follower mode. This means the MOSFET is never fully turned on unless you drive the gate with with 30V with respect to ground. NOT a good design.

    Use a P-Channel MOSFET instead and connect the source to the 25V supply. Now drive the gate with -5 volt PWM applied with respect to the source. Start with a 100Ω minimum load and see how it works.

    Change D1 to a Schottky type diode such as a 1N5819.

    Can't help much more without information from you regarding, L1 and load requirements.

    Good Luck,
    Ifixit
     
  4. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    I think I covered that when I mentioned the PWM signal needed to be referenced to the MOSFET source terminal instead of ground.

    The trouble with P-ch MOSFETs is the relatively high gate charge compared to a same-rated N-ch MOSFET, not to mention that there are far fewer P-ch MOSFETs on the market than N-ch.

    A Schottky or a fast recovery diode is a must. The 1N400x series are too slow for use above about 1kHz.
     
  5. renatomassa88

    Thread Starter New Member

    Jun 12, 2011
    8
    0
    I don`t know if I can reference the signal to the source`s terminal
    in this program (proteus). As to the start-up current , can`t I put
    a resistor to limit this current?
    Actually I just want to simulate this circuit, for now, without feedback
    (open loop) and see the output`s regulation changing the dutycycle.
    Is there another software that is better to do it?
    Thanks.
     
  6. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    Try Linear Technology's LTSpice.
    Google "LTSpice download".
    It's very good and it's free. The schematic interface is just a bit quirky, but not bad once you get used to it.

    There is a support group for LTSpice on Yahoo! Groups:
    http://tech.groups.yahoo.com/group/LTspice/
    You'll need to register for a free Yahoo! ID, and request membership in the group; approval may take a day or three.
     
  7. ifixit

    Distinguished Member

    Nov 20, 2008
    638
    108
    I second that idea from Sgtwookie to try LTspice. There will be a learning curve.

    Here a source file (.asc) of the circuit you posted to get you started with experimenting. To limit the start-up current I have reduced C1 to 100uF.

    I have included a plot settings file (.plt) to get you started with the waveform viewer.

    After you load LTspice, place my 2 files together in a suitable directory and double click the .asc file, or open it from within LTspice.

    Good Luck and have fun.
    Ifixit
     
  8. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    And attached is the same simulation, only using an N-ch MOSFET instead of a P-ch MOSFET.
     
  9. renatomassa88

    Thread Starter New Member

    Jun 12, 2011
    8
    0
    I downloaded the software and seems to work,
    but shoudn`t the circuit with 20% duty cicle
    (ton=10us,T=50us) give an output of 5V with
    25V input?
    (Vo=D*Vi)
    I also tried in the multisim software, do you
    know that one? In this one I can reference
    the signal to the source`s terminal. But for
    some reason the output isn`t right. You think
    this circuit doesn`t work in an open loop analysis?
    I attached the circuit and some editions from
    the components.
    Thanks so far.
     
Loading...