Connect with us

looking for SMPS NFET...

Discussion in 'Electronic Components' started by Mike Deblis, Dec 2, 2003.

Scroll to continue with content
  1. Mike Deblis

    Mike Deblis Guest

    I'm trying to find a better NFET than an IRF740, preferably in a TO220 or
    similar package, for a HV PSU. 250V 50mA from 12V 2A in.

    What I'm trying to do is get a Qg around the 20nC mark, Vds > 250V, Id >2A
    with as low an Rdson as possible in order to minimise losses. If I can get
    Rdson really low (sub 0R100 if poss), the FET losses will drop but high Qg
    (>60nC) is a problem for the driver... Currently I'm hitting 81% efficient -
    I'd like to get closer to 90% if possible...

    The drive chip is a MAX1771. The inductor is 47uH 0R045 and 2.6A. Current
    limiting resistor is 0R050 which limits the max inductor current to 2A ATM.
    Board is carefully laid out as per data sheet recommendations.

    Thanks
     
  2. Robert Baer

    Robert Baer Guest

    Forget anything that Maxim claims to make; there are too many parts
    that are un-available, unless you aer willing to spends tens of
    thousands of dollars per part type and wait 6-26 minths...
    Many distributors refuse to cary Maxim due to this fact.

    As far as FETs go, try those made by ST Microelectronics:
    http://us.st.com/stonline
     

  3. MOSFETs rated for >250V with Rds(on) < 0.1 Ohm and with a Qg of around 20nC
    don't exist yet unfortunately.

    However, some improvement can probably be made over the IRF740. The IRF740A
    is a lower gate charge version of the IRF740 but is otherwise pretty much
    identical. Some performance advantage could likely be had using that
    device, however I might suggest the IRFB9N30A (available at Digikey) as an
    even better choice still:

    http://www.irf.com/product-info/datasheets/data/irfb9n30a.pdf

    The device is a 300V 0.45 Ohm 33nC max device.

    Be wary. Selecting a gigantic MOSFET rated for >250V and less than 0.1 ohms
    on resistance will not likely yield the best efficiency possible. Not only
    will such a beast have giant gate charge, but of additional interest it will
    also have relatively large Coss (output capacitance). Some of the switching
    loss is caused by slow transistions in the MOSFET itself (IE, periods of
    blocking relatively high voltage but passing high current still), but the
    MOSFET output capacitance can cause non-negligible loss contributions for
    high voltage MOSFET run at high frequency.

    Your MAX1771 uses a funky pulse frequency modulation technique but it can
    easily decide to run at up to 300kHz. That classifies as high frequency,
    especially when dealing with such large output voltages.

    The output capacitance loss can be estimated if you know the switching
    frequency, the maximum voltage the output capacitance gets charged to each
    cycle, and the actual output capacitance value at the voltage it will be
    seeing.

    The formula for estimating the output capacitance switching loss is:

    Power Loss = 0.5 * f * c * v^2

    Where f is the switching frequency, c is the MOSFET output capacitance in
    farads as measured in the datasheet at the drain-source voltage the MOSFET
    will be seeing (250V in your application), and v^2 is the drain-source
    voltage the MOSFET will be seeing squared. This formula assumes you are
    charging the output capacitance from an inductive device. Since you are
    using the MAX1771 I'm assuming you are using a plain ordinary boost
    converter topology. In this case when the MOSFET switches off, the drain
    voltage must rise from 0V (conducting) to 250V (off) in order for the output
    diode to turn on. During this period the inductor transfers some of its
    stored energy to the output capacitance losslessly. If the boost converter
    is operating in continous conduction mode then this energy stored in the
    output capacitance is lost when the MOSFET turns on for the next cycle (it
    discharges through the MOSFET channel and heats the die). In discontinuous
    conduction mode the inductor current drops to zero, but then reverses
    direction because the MOSFET output capacitance is at 250V (which is clearly
    larger than 12V supply). Then energy will flow back through the inductor
    and try to charge up the input capacitors on the 12V side somewhat.
    Eventually the output capacitance voltage will drop below 12V, but the
    inductor will keep the current flowing for awhile and thus it may pull the
    MOSFET output capacitance below ground and activate the MOSFET body diode.
    There will be some ringing, damped by the Q of the system and any snubbers
    if present, but in the end some of the energy may be returned to the supply,
    while other energy may get lost. If the MOSFET was switching a resistor
    instead of an inductor the above power loss forumula would not have the 0.5
    factor in front, but would be 1 instead since there would be energy loss for
    both charging and discharging the output capacitor then.

    Anyway lets take a look at your application. Lets assume either you are
    operating in continuous conduction mode or almost all of the energy in the
    output capacitance is lost so the above formula is fairly accurate. With
    the IRFB9N30A the datasheet claims the output capacitance is 52pF when
    measured at 240V Vds. 240V Vds is very near your 250V figure, so we will
    assume the 52pF is accurate (is practice it would be every so slighly
    smaller at
    the slighly higher voltage, but the difference is small). Now assume your
    MAX1771 is switching the device at the full 300kHz.

    Power loss = (0.5)*(300000)*(5.2x10^-11)*(250^2) = 0.49 Watts

    Half a watt is not to be sneered at. A bigger MOSFET with lower
    on-resistance will almost certainly have a larger output capacitance and
    subsequently higher output capacitance switching loss contribution (not to
    mention the other switching loss mechanisms) which will most likely more
    than offset any gains in lower Rds(on).


    What kind of diode are you using?

    My suspicion is you could probably eek our better performance by modifying
    your inductor (although considering the topology and your input output
    voltage requirements your efficiency is already pretty good). A boost
    converter with such a large input output voltage range is likely to have
    high AC losses in the inductor. These can be minimized by paying close
    attention to skin effect, proximity effect, and core eddy and hysteresis
    losses. But that is all another big topic in and of itself.
     
  4. Robert Baer

    Robert Baer Guest

    Go to the STM site i mentioned and look at the data sheets for their
    STP22NS25Z (250v, 0.15 ohms, Qg=120), STP16NS25 (250V, 0.28 ohms,
    Qg=59), STP8NS25 (250V, 0.45 ohms, Qg=37).
    Generally, the higher the voltage rating, the higher the Rds(on)
    found.
    Also, the lower the Rds(on), the higher the Qg found (for a given
    voltage rating).
     
  5. Mike Deblis

    Mike Deblis Guest

    ....
    ....

    I have ordered a few to try - should be here today...
    STTA206S - should be fast enough I hope.
    Inductor is an Epcos B82479 series - 47uH DCR 0R087, 2600mA.

    I really appreciate your detailed comments - I suspect that if I can get to
    85% that should be enough. Cutting the dissapation in the FET will help
    greatly - the inductor gets slightly warm - I'll check again when I get the
    FETs.

    I'm also going to see if I can increase the input voltage slightly as this
    will lower the current - every little bit helps.

    Thanks for the help so far,

    Mike
     
Ask a Question
Want to reply to this thread or ask your own question?
You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.
Electronics Point Logo
Continue to site
Quote of the day

-