Did I Pick the Wrong MOSFET? LTSpice Simulation Polyphase Boost Circuit

Discussion in 'General Electronics Chat' started by Major Tom MIB, Feb 28, 2011.

  1. Major Tom MIB

    Thread Starter Member

    Nov 28, 2010
    31
    0
    Each phase works by itself, but they require a diode to work together--Did I miss the boat on how to connect p-channel mosfets that are driven in different phases? Part indicates reverse break down voltage of 12 and the circuit simulates to ~8v. Didn't find much on the web, but the circuit is the 'basic' boost model 'tuned' for a narrow voltage. An educational project.

    --- Bill of Materials ---

    Ref. Mfg. Part No. Description
    C1 -- -- capacitor, 1mF
    D1-6 -- D diode
    L1-6 Vishay IHLP4040DZ01-0R19 inductor, 190nH, 90A pk
    R1 -- -- resistor, 100m
    U1,3-6 International Rectifier irf1324s-7ppbf integrated circuit
    U2,8-12 International Rectifier irf7410 integrated circuit
     
  2. tom66

    Senior Member

    May 9, 2009
    2,613
    213
    I'm no expert on buck converters but should the two drains of the fets be tied together? Shouldn't it be source and drain? Or is that because you have mixed P and N types. Most synchronous converters I've seen use only N type and they drive them with alternating waveforms.
     
  3. Major Tom MIB

    Thread Starter Member

    Nov 28, 2010
    31
    0
    The theory was to use the MOSFETs as a ~perfect switch and diode--which appeared to work in the 'basic' circuit.

    Initially the 90amp rated 1uH inductor gets shunted to ground for 20us + the 20us start-up for 40us total at which time the current measured ~70A (~80%). Then it's connected to the load for 1us + 1us for the (rise+fall time)/2. Then back to ground for 10us to bring it back up to 70 A. The PWM timing was based on the n-channel (low side) mosfet, but that 1us rise/fall time is a pretty big buffer to avoid having both the ground and load 'connected' at the same time, so I didn't see any problems using that timing inverted for the p-channel (high side.)

    The discharge was limited to ~18V to keep it 'well' under the MOSFETs' 24V limit. I selected an arbitrary load of .1 ohms at 12.25V to simulate a 1500 watt electric heater as installed in an RV/marine/automotive environment (10.5V to 13.8V/1100 to 1900 watts.) Next projects will be the buck from 24v to 12v and then the sepic/buck-boost for 12v to 12v.

    The above circuit running from 1.2v draws 1.6KA (1380w) at startup and 1.15KA steady state while converting to 8.25v (.1v ripple) 80A (660w) or~50% efficient :( Needs more work :) But if you put the converter inside the heater it reaches the target watt range! :)

    So, why did I need diodes?

    [Edit] The lower than expected voltage should just be as simple as adjusting the single phase timing for the 6 phase model after I check to make sure I'm within operational limits of the parts I've chosen. I want to design PWM source with feedback for duty cycle and the provide n-phase inverter/delay signals. It might just be a good computer programming exercise to create a design program to configure a micro controller based dc-dc converter given ideal or real world components.

    [More] There's a diode in this LTC4358 - 5A Ideal Diode--guess it's needed AND it's connected in a different way Source to Gain. Worth a try. That was a no-go :(
     
    Last edited: Feb 28, 2011
  4. Ron H

    AAC Fanatic!

    Apr 14, 2005
    7,050
    657
    I can't make sense of your partial schematic. Why do you have it spread across a football field? And why are you using those generic MOSFET symbols that do not show whether they are N- or P-channel? LTspice has separate symbols for N-channel and P-channel MOSFETs. They might make your schematic more intelligible to us.
     
  5. Major Tom MIB

    Thread Starter Member

    Nov 28, 2010
    31
    0
    I have a small monitor and the large scale was to make the text in the bitmap legible--guess it backfired.
    LTSpice library didn't have the International Rectifier parts I selected. IRC provide them as '.sub' files and LTSpice gave the option automatically build them. I'll rework them to the standard and provide a better scaled circuit diagram.

    Attached are the .asc files for the single and six phase projects. The link for the IRC library is above, but I'll try to revise them and get them reposted soon. Sorry, it didn't look that bad on my screen.
     
  6. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    One big problem is that you have the P-ch MOSFET gate voltage referenced to ground instead of the source terminal. This kind of thing is specified in the datasheets as Vgs.

    With both N-ch and P-ch MOSFETs, when Vgs=0, the MOSFET is off.
    In order to turn on an N-ch MOSFET, Vgs needs to be a positive value, 10v for standard and 5v for logic level MOSFETs.
    For P-ch MOSFETs, Vgs needs to be negative.

    Also, your rise and fall times were specified at 1uS, with an ON-time of zero. This means your MOSFETs are spending a good bit of time in their linear region, generating heat.
     
  7. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    In-between a zillion other things, I took a look at the datasheets for those MOSFETS.
    N-ch: IRF1324S-7PPbF
    P-ch: IRF7410
    While the N-ch MOSFET is rated for Rds(on)=0.8m and Id=240A, the P-ch is only rated for Id=-16A. That's quite a mismatch.

    You're going to have a hard time finding a P-ch MOSFET that'll be a decent match for an N-ch MOSFET, as if Vdss, Rds(on), and Id are equal, the gate charge will be about 2.5 times as high as the N-ch MOSFET. That's just the way it works; hole flow is not as efficient as electron flow. This is also the reason why high-side drivers have become popular; with them, you can use N-ch MOSFETs for both the low and the high side, so you have a near-perfect match.

    You'll need to include some "dead time" from the low side turning off and the high side turning on, and vice versa. Otherwise, you'll short the output cap to ground.
     
  8. Ron H

    AAC Fanatic!

    Apr 14, 2005
    7,050
    657
    Here is how I assign a subckt to a MOSFET symbol. The MOSFET symbols by default are meant to represent devices modeled in .MODEL statements, but with a simple change, they can represent subcircuits. The steps are shown below.
     
Loading...