Good MOSFET selection

Discussion in 'The Projects Forum' started by gdylp2004, Jan 9, 2012.

  1. gdylp2004

    Thread Starter Member

    Dec 2, 2011
    Hi guys,

    I thought of upgrading my buck converter's high side MOSFET. Previously, I was using IRF640NPbF (datasheet:

    Do anyone know if the following are the main criteria for searching a good mosfet (please let me know if I've missed some important considerations out):
    1. Rds(on) - Commonly know for power dissipation since high Rds(on) represents higher voltage drop, hence more power loss
    2. Gate-charge - Lesser gate charge the better? Since the charge, hence time required for the MOSFET to turn on and off directly related to the gate charge?
    If I am choosing this Toshiba MOSFET (datasheet: with a Rds(on) of only 38mΩ (10x reduction compared to IRF640N), is it a good choice?

    If yes, what about the gate-charge? For a Gate-source voltage about 12V & Vds of ~100V, IRF640N has ~50nC but the Toshiba FET has ~150nC! (3X difference). Does this mean the Toshiba FET is not better than the IRF640? Or I can't made such comparison because the test Id is not the same, then what should we do?

  2. jimkeith

    Active Member

    Oct 26, 2011
    It all depends upon the switching frequency--to reduce switching losses, keep the frequency down, reduce gate charge capacitance or increase drive current. Conduction losses may be reduced via lower Rdson, but this is not a function of switching frequency.

    All this stuff is a trade-off between switching losses and conduction losses--impossible to state on the surface if one is better than the other--the lower the gate charge, the better it will work at high frequencies. On the other hand, the IFR640 is an old design and has higher gate charge per amp than modern designs--as technology improves, gate charge continues to go down.

    Your circuit may not be that critical and my be able to handle the increased gate charge capacitance OK--try it and see what happens. Or it might smoke...

    Sorry I bowed out of your previous thread--it was apparent that the others who were taking over had superior smarts.
    gdylp2004 likes this.
  3. gdylp2004

    Thread Starter Member

    Dec 2, 2011
    Thanks Jim, no problem for that. Instead, thank you for your assistance in the early stage.

    Infact, do you still remember that you've mentioned that the non-sync buck that I'm building may not have a provision for the Cbst to discharge (no return GND path), I've actually experienced that whenever my o/p is open. Haha, got really fustrated why the V(gs) appears and disappear momentarily, but after some thought and some good recap from my previous thread, I've found out that it was actually the O.C @ the o/p prob.

    Now the other problem I've faced is the transient spikes which has a pk-pk value of about 50V! I've asked my research senior for the way to solve it, and he told me to either parallel another o/p cap to suppress the transient, but this doesnt work. He has also asked me to try to increase Rgate from 5.6ohms to 22ohms. But that either doesn't reduce the transient but instead burnt my IRF640 as I believe the discharge time for the MOSFET was too long after introducing 22ohms, and hence allowing the FET to be in the linear region too long, hence too hot and got burnt.

    I wonder what else can I do to suppress the transient. Really at wits end. :(
  4. gdylp2004

    Thread Starter Member

    Dec 2, 2011
    Hi guys,

    I've changed my nMOS from IRF640NPbF to IRFB4227PbF as intended. Was glad because initially, the nMOS was having a fever of >160 °C after running for only a min or so, but now the replaced MOSFET is only at 155 °C (steady state ~15mins). Also, the new MOSFET has a higher max. operating temperature range is 175 °C compared to the former's max. 150 °C.

    However, the V(gs) waveform seems to tell me that the charging of the gate terminal of the FET is overdamped by looking at the knee of the rising edge. I've removed 5.6 Ω in hope to improve the dv/dt response but to no avail.

    Does anyone has clue why removing Rgate has not much effect, and are there any tricks that I can do to improve this? Or I would need to re-select a nMOS with a lower gate capacitance (which means choosing another nMOS with lower gate charge).

    Thank you.
  5. crutschow


    Mar 14, 2008
    The driver you selected has a drive capability of 250mA. The IRFB4227PbF transistor has a gate charge of 70nC. That gives a calculated rise time of 70nC/.25A = 0.28μs, about what the scope photo shows.

    To get a faster turn-on you will need a transistor with a lower gate charge, or a driver with a high output current capability.
  6. joeyd999

    AAC Fanatic!

    Jun 6, 2011
    It will also be very helpful if you keep this discussion on *one* thread. All these issues are intertwined, and it makes it difficult for those of us trying to help you when we need to bounce back and forth between threads!
  7. gdylp2004

    Thread Starter Member

    Dec 2, 2011
    Sorry, I create a separate thread because I would like to ask a more specific and precise question, and also, sometimes I do not get any replies at all due to a title mismatch. I've experienced that whenever I try asking questions in my main design thread.
  8. gdylp2004

    Thread Starter Member

    Dec 2, 2011
    If I would to estimate its discharge time using your method I=dq/dt, from
    the IR2117 datasheet, its "sink" capability is 400mA as written as Io-.

  9. shortbus

    AAC Fanatic!

    Sep 30, 2009
    Most gate drivers sink more current than they source. Usually twice as much sink current. This is because the gate is basically a capacitor and it charges faster than it discharges.
  10. gdylp2004

    Thread Starter Member

    Dec 2, 2011
    On a second look, the rise time looks more like a 1us than 0.28us. Did u refer to the 63.2% instead of 100%?
  11. gdylp2004

    Thread Starter Member

    Dec 2, 2011

    I would like to know is there a distinct difference between IRFB4620 and IRFB5620.


    I've literally compared each and every parameters and realised almost most of them are the same yet the applications as suggested by the individual datasheet are different. One is for SMPS while the other is a Class D amplifier?

    What I would like to know is if they are interchangeable?

    Thank you!
  12. SuperHerman

    New Member

    Jan 12, 2010
    i just into the same question, have also searched the Datasheets for a significant difference which makes the IRFB5620 better for Class-D,

    Matter of factly it has more QRR and consequently a longer reverse recovery time of the internal Body Diode.

    At the top of the Datasheet they state, that these are improved parameters where they are worse than with the irfb4620,

    If anyone has an idea, where the real difference is, i too would be very interested.

  13. praondevou

    AAC Fanatic!

    Jul 9, 2011
    ??? ONLY 155°C ?

    Even if you wanted to run it at this temperature I'm sure you did the test at room temperature, right? So that means if the ambient temperature rises so will the MOSFET temperature.

    I would very much recommend to use a heatsink (or a better one if you already have one) and forced ventilation.