MOSFET heating up too much ????

Discussion in 'The Projects Forum' started by dlpweb3, Jan 28, 2012.

  1. dlpweb3

    Thread Starter New Member

    Jan 5, 2012
    16
    0
    I have a very simple MOSFET switch I'm using to turn on an emergency night light. The circuit works OK - very simple after all - but the MOSFETS heat up way more than I expected.

    The MOSFETS are IRFP4310ZPb - and I use TWO of them in parallel to drive a little 50W, 12VDC lamp. Each of the MOSFETS is rated at 134A, so I was expecting these two not be barely warm. Besides, the light isn't even on 100% - it's run at about a 50% duty cycle (blinks).

    Instead, the things are up to about 135 degrees (each has a little heat sink - I'm using ones designed for a TO220 package, not the TO247 of the actual fets, but the cooling of these is rated almost the same. Using thermal grease).

    This is a simple circuit - apply 11VDC to the gate, 12VDc source, 50W load.

    Isn't this WAY too much heat for this 50W load???

    (thx in advance)
     
  2. PaulEE

    Member

    Dec 23, 2011
    423
    32
    Is your lamp on top, or on the bottom? Can we see a schematic? The Rds ON resistance is spec'd in the milliohms for this part fully on. That doesn't equate to much power at a few amps...
     
  3. praondevou

    AAC Fanatic!

    Jul 9, 2011
    2,936
    488
    Yes, something is connected the wrong way. These transistors have 6mOhm max RDSon. Since you are switching very slowly forget about the switching losses.
     
  4. hgmjr

    Moderator

    Jan 28, 2005
    9,030
    214
    Post a schematic so that we can see how you connected the two mosfets.


    hgmjr
     
  5. Audioguru

    New Member

    Dec 20, 2007
    9,411
    896
    A 50W light bulb draws much more power when it is not completely hot. Yours is blinking so it is cool most of the time and drawing a lot more current than you think (maybe 10 times as much current).

    Mosfets oscillate when the inductance of the gate wire resonates with the high gate capacitance. Usually a 10 ohm to 47 ohm resistor is connected in series with the gate at the gate pin to eliminate the oscillation.
     
  6. hgmjr

    Moderator

    Jan 28, 2005
    9,030
    214
    Audioguru has a good point. When faced with this problem in the past, we used what are referred to as keep-alive resistors. These are resistors the value of which are chosen to keep the filaments just on the verge of lighting. Thus the filaments are kept warm and are less prone to induce the high startup currents mentioned.

    hgmjr
     
  7. praondevou

    AAC Fanatic!

    Jul 9, 2011
    2,936
    488
    Ok, let's say 500W instaed of 50W. That's 42A at 12V, 21A for each MOSFET if they are in parallel.

    2.6W for each MOSFET. He must be using really small heatsinks to get 135°C...
     
  8. PaulEE

    Member

    Dec 23, 2011
    423
    32
    3 watts is more heat than it sounds like... :)
     
  9. dlpweb3

    Thread Starter New Member

    Jan 5, 2012
    16
    0
    Here is the circuit for the mosfet switch. not much here.
     
  10. praondevou

    AAC Fanatic!

    Jul 9, 2011
    2,936
    488
    That looks to me as if gate charge and discharge times are too long, while it's charging/discharging the MOSFET will stay for a too long time in the linear region, it's neither on nor off, power dissipation increases.

    You should be using a better gate drive circuitry or a dedicated gate driver IC.

    Goto digikey and look up http://search.digikey.com/us/en/cat...drivers-external-switch/2556427?k=gate driver
     
  11. dlpweb3

    Thread Starter New Member

    Jan 5, 2012
    16
    0
    Hi Praon. Thanks for taking a look.

    Should this G D IC REPLACE the 2N222, go between that and the Mosfets, or ... what?
     
  12. dlpweb3

    Thread Starter New Member

    Jan 5, 2012
    16
    0
    And about the current with this light.

    I measured the resistance of the bulb at between 0.3 and 0.4 ohms. That puts the current far higher than could be (35 amps or so at 12 VDC), and I know this entire device is fused at not more than 20 amps - per memory a lot less than that.

    How the heck can THAT be ???
     
  13. praondevou

    AAC Fanatic!

    Jul 9, 2011
    2,936
    488
    Yes a gate driver IC replaces the npn transistor and resistors. You will need two gate resistors for the MOSFETs though, low values, i.e. 10Ohm.

    The light bulb resistance is much lower when the filament is cold. This was explained earlier in this thread. So everytime a light bulb is switched on there will be a high inrush current.
     
  14. thatoneguy

    AAC Fanatic!

    Feb 19, 2009
    6,357
    718
    Fuses usually need about 150% of their rated current for over 250 mS to blow. Slow Blow fuses are even more lenient. You'll find that many transistors sacrifice their lives to save the fuse. :D

    Try to switch the gate on and off faster, use the NPN as a switch rather than an amplifier. You want the full 12V to feed the MOSFET Gates.
     
  15. crutschow

    Expert

    Mar 14, 2008
    13,028
    3,237
    A standard incandescent bulb conducts high current for about 50ms and takes about 350ms to achieve full brightness (see this). So the high current occurs for a short period, which is not enough to blow a fuse.
     
  16. dlpweb3

    Thread Starter New Member

    Jan 5, 2012
    16
    0
    Thanks, Crutshow. What a wellspring of knowledge!

    What wouod it take to 'preheat' the incandescent filament - and does the gate driver IC driver 'praondevou' suggested do this? Would an in-series resistor AudioGuru & HgmJr spoke of inhibit this? It's not at all clear to me how an in-series resistor would do anything other than decrease the gate voltage.

    Are they, perhaps, speaking about a resistor between line voltage and the gate "just enough" to preheat but not enough to illuminate (i.e., in parallel with the pulse input)?
     
  17. praondevou

    AAC Fanatic!

    Jul 9, 2011
    2,936
    488

    For this simple design you can do one or the other.

    Either you put a resistor in parallel with the MOSFETs (drain-source), the resistor will then ALWAYS be in series with the light bulb, FET on or off. The resistor value is chosen to keep the filament preheated but not glowing.

    OR you do something to turn the MOSFET on and off much faster. To achieve this you need to charge /discharge the gate fast, with a low output impedance gate driver or appropriate circuitry.
    Your gate is charged trough two several kOhms resistor, it takes therefore forever.
     
  18. dlpweb3

    Thread Starter New Member

    Jan 5, 2012
    16
    0
    Hi ThatOneGuy

    To use this NPN as a switch - simply connect emitter to the Mosfet gate (collector to line voltage) with the pulse driving the npn gate? No resistors?
     
  19. thatoneguy

    AAC Fanatic!

    Feb 19, 2009
    6,357
    718
    You'll need a base resistor to limit current to 1/10th of the C-E Current, and possibly a small emitter resistor to avoid smoking the transistor.

    Look at the design of BJT H-Bridges to get an idea of how to use them for fast switching.
     
  20. Audioguru

    New Member

    Dec 20, 2007
    9,411
    896
    Of course the Mosfets get extremely hot and the light bulb is dim.

    Your Mosfets are connected as source-followers instead of as efficient switches.
    The light bulb should be between the drain pin and the +12V supply.

    The gate voltage of the Mosfets must be at least +10V higher than their source voltage for them to completely turn on but since their gate voltage is only +12V then their source voltage is much less and the Mosfets have a huge voltage across them and the light bulbs have a reduced voltage.

    Connect a SINGLE Mosfet like this:
     
Loading...