Help in deciding MOSFET for 1kw SMPS

Discussion in 'The Projects Forum' started by RamaD, May 2, 2014.

  1. RamaD

    Thread Starter Active Member

    Dec 4, 2009
    254
    33
    I am designing a full bridge SMPS of 1kW, with input of 120V, at 42.5kHz.
    I have selected a choice of IRFP4668 or IRFB4127 MOSFETs, both 200V.

    IRFP4668 - 200V 130A
    Rds ON 0.0097 Ohm
    Trise+Fall 180ns
    TO247AC

    IRFB4127 - 200V 76A
    Rds ON 0.020 Ohm
    Trise+Fall 40ns
    TO220

    4 Nos. of IRFB4127 has a loss of about 15W, and IRFP4668 25W. I am inclined towards IRFB4127, due to efficiency and half the price, but is TO220. Since the Trise+fall is 40ns would there be any problems of fast rise and fall times? Is the choice of MOSFET alright?
    Thanks in advance.
     
  2. Papabravo

    Expert

    Feb 24, 2006
    10,137
    1,786
    Excuse me. If you have an AC source of 120V at 42.5kHz. Why do you need to build an SMPS. That term usually applies to DC Input/DC Output situations. It sounds like your post may violate the TOS for dangerous or inappropriate circuits such as mains connected circuits without a transformer. Post a schematic and tell us more about what you are trying to do.
     
  3. THE_RB

    AAC Fanatic!

    Feb 11, 2008
    5,435
    1,305
    I believe he has input of 120v AC, and the SMPS needs to be 1kW and run at 42.5kHz.

    I'm guessing his SMPS transformer is suited for 42.5kHz?

    Those FETs are a very poor choice. The 120v AC will rectify to maybe 170v DC, and a 200v rated FET is very close to the limit and will fail.

    Most FETs in offline SMPS are 600v or 800v rated. :)
     
  4. RamaD

    Thread Starter Active Member

    Dec 4, 2009
    254
    33
    @Papabravo
    @The RB
    Sorry, I was not clear in my descriptions.
    The input is 120V DC.
    The SMPS switching freq. is 42.5kHz.

    BTW, wont 200V rated MOSFET be ok for this? And even for 170V as the stress on the MOSFET is just Vdc for bridge circuits?

    The SMPS switching freq. is at 42.5kHz., I was greedy about a bit of efficiency. The switching losses are going up with switching frequency. So I decided to keep the frequency low and take advantage of low Rds On Mosfets, without the size constraints. Please comment.
     
  5. Papabravo

    Expert

    Feb 24, 2006
    10,137
    1,786
    You're starting to get the hang of complete and explicit specifications. Unfortunately we are not mind readers and still have no idea what you are trying to accomplish. You need to specify the output parameters and the input power available. Try again and include your proposed schematic.
     
  6. THE_RB

    AAC Fanatic!

    Feb 11, 2008
    5,435
    1,305
    Agreed! We need to know the output voltage and current. Also your SMPS "topology" ie if it is a buck or push-pull with a transformer etc.

    Re your question about the voltage, I think 200v FETs are much too low for a 120v DC input SMPS.

    What is your duty cycle and peak current in the FETs? How many FETs?

    1kW at 120v DC is roughly 9A and at 50% duty would be 9A average and 18A peak. Assuming you have a few FETs in parallel (normal for 1kW) you don't need a very low Rds-on FET. It's more important to have a high voltage FET.

    The world is full of offline switcher PSUs and there are a ton of good cheap high voltage switching FETs out there.
    :)
     
  7. RamaD

    Thread Starter Active Member

    Dec 4, 2009
    254
    33
    My plan is to build a Inverter with input from PV. The first stage is an SMPS with input from 120V DC from PV, whose output will be 350V DC. The second stage would be a PWM sine wave inverter, with the sine weighted PWM generated by a uC.
    The Open Circuit voltage from PV would be 168V max. The first stage SMPS topology is H-bridge, where the MOSFET stress would be only the Vdc, as the spike above that would get returned to the positive rail, and 200V would be sufficient, so my thinking! I was thinking about 75% duty cycle to keep the Peak current low, and not run it below 105V where the duty cycle would be just around 90%.
    I was only thinking of 1 MOSFET per arm, with a total power dissipiation for 4 MOSFETs of about 15W with IRFB4127s and 25W with IRFP4668s, so that works out to a max of 6W per MOSFET, which is manageable. Do you still recommend additional parallel MOSFET? I would need to add 4 for H Brige!
    The way I have calculated power dissipiation is, I*I*R losses + Switching losses,
    where switching losses is 1/2 * (Trise+Tfall) * Ipk * Vdc * f

    The IRFB4127 has fast rise and fall times (40ns total!), and my question was whether I am heading for an alien emi, dv/dt problems, and any other, which I would be glad to learn, or is that ok?
     
  8. Papabravo

    Expert

    Feb 24, 2006
    10,137
    1,786
    You still don't get it do you. Your verbal descriptions are meaningless gibberish without a schematic and without a clear definition of input power available.
     
  9. THE_RB

    AAC Fanatic!

    Feb 11, 2008
    5,435
    1,305
    Well if the 120v DC is from the solar array, what happens when load is reduced and your solar panels produce about 23v DC each? What does that 120v go up to?

    And even if you can guarantee the 120v DC will never exceed 120v, 200v FETs are asking for trouble. It's driving an inductive transformer and there will be spikes etc before diodes can commutate fully.

    I've pulled apart literally hundreds of SMPS's and never seen the max voltage of the switching transistors have so little margin, in that voltage range.

    The world is chock full of really good HV switching FETs now! Just get some 600v or 800v FETs like they use in other SMPS's.

    Or better still, buy a cheap Asian made 1kW inverter. You won't even be able to buy the parts for that cheap. And you are going to blow up a LOT of parts... ;)
     
Loading...