fry mosfet after 30 mins

Discussion in 'The Projects Forum' started by sq-aristo, Jun 12, 2013.

  1. sq-aristo

    Thread Starter New Member

    Jun 1, 2013
    20
    1
    hi guys,
    i have made a bicycle exercise trainer with an arduino, lcd screen and a tractor alternator.
    i have stripped the alternator so i am sure i can control the current to the rotor precisely.
    i have to arduino sending a pwm signal to the control circuit that is adjustable via the lcd and keypad. frequency is default (490hz) i think.
    the control circuit consists of an optocoupler, a pnp transistor and a ir320 n channel mosfet.
    After it working perfectly for a few days i gave it a hard run for half and hour at maybe 60% duty cycle and i noticed the alternator locked solid (rotor had full current 6 amps) also the mosfet was too hot to touch.
    any simple things that might have caused this?

    circuit is here: http://sphotos-a.ak.fbcdn.net/hphotos-ak-ash4/p480x480/375585_10151645022846308_744151784_n.jpg
     
    Last edited: Jun 12, 2013
  2. sq-aristo

    Thread Starter New Member

    Jun 1, 2013
    20
    1
    im wondering if the mosfet body diodes are overheating at to load i had on them. I have had the same circuit run for an hour a while ago but my legs are stronger now and i was running about 40-50% instead of about 20%
    i have some byw29e-200 on order i hope these will fix the problem.
    any enlightenment of ideas would be greatly appreciated
     
  3. crutschow

    Expert

    Mar 14, 2008
    13,000
    3,229
    The MOSFET appears to be upside down. The source should go to ground.

    You also need to add a power rectifier across the field connections (cathode to plus) to carry the inductive rotor field current when the MOSFET is off.

    In normal operation the MOSFET body diode should then carry no current.
     
    sq-aristo likes this.
  4. ScottWang

    Moderator

    Aug 23, 2012
    4,853
    767
    Exchange the DS pins.
     
  5. sq-aristo

    Thread Starter New Member

    Jun 1, 2013
    20
    1
    i might have drawn the schematic wrong the mosfet is definitely wired correctly.
    by power rectifier you mean diode?
     
  6. sq-aristo

    Thread Starter New Member

    Jun 1, 2013
    20
    1
    can you please explain how the lack of diode has made the mosfet melt? (figuratively)
    i assumed that being an overkill mosfet this would be the smallest of my problems
     
  7. sq-aristo

    Thread Starter New Member

    Jun 1, 2013
    20
    1
    i did draw the mosfet around the wrong way. sorry.
    the circuit other than that is correct and like i sad worked correctly for a considerable amount out time before failure, ant reasons for failure would be great!
     
    Last edited: Jun 12, 2013
  8. richard.cs

    Member

    Mar 3, 2012
    162
    31
    Working for 30 minutes before failure certainly looks like a thermal problem, however crutshow is correct in that a diode across the rotor is needed. It provides a path for the current to flow when the mosfet is off and without it the mosfet may be damaged by overvoltage.

    Out of interest how are you loading the stator is the alternator? It's not particularly relevanty to your problem just nice to know.
     
  9. mik3

    Senior Member

    Feb 4, 2008
    4,846
    63
    The MOSFET died due to overheating.

    It turns on fast due to the BJT driving its gate but it turns off very slowly because the only path to discharge the gate is R3. You need to provide a low resistance path to turn it off. Better to use a low side MOSFET driver IC.

    The fact it did not fail due to the inductive kick back of the motor winding is because of the slow turn off.
     
    sq-aristo and absf like this.
  10. wayneh

    Expert

    Sep 9, 2010
    12,100
    3,034
    Oh crud, I completely buggered my edit.

    Anyway, I had basically agreed with mik3 because the gate turn-off time was calculated at maybe 5% of the PWM cycle time. I didn't realize that the snubber diode suggested back in #3 was still was not in place. It's a more likely cause as crutschow reiterates in #11.
     
    Last edited: Jun 12, 2013
  11. crutschow

    Expert

    Mar 14, 2008
    13,000
    3,229
    Yes, a power diode. Without a diode the current going through the field winding generates an inductive spike that breaks down the MOSFET substrate diode when the MOSFET turns off and creates momentary high dissipation. It has to dissipate all the energy in the winding inductance each cycle. That's likely the cause of your observed failure. Thus the need for a diode across the winding.
     
    sq-aristo likes this.
  12. sq-aristo

    Thread Starter New Member

    Jun 1, 2013
    20
    1
    Thanks for all the replies. A driver ic is out of the question for this project. I will fit a snubber diode first thing today, I assumed the body diode in the MOSFET would be fine. As for the low resistance turn off, could r3 be decreased to 1k or even 560ohms or would I need another transistor for turn off. Peak efficiency isn't needed I just want it to work.
    Loading the stator is achieved by using a stuffed battery that absorbs the charge but even after half hour isn't over 12.8v.
     
  13. richard.cs

    Member

    Mar 3, 2012
    162
    31
    The body diode is in the wrong place to work properly. If you reduce R3 will will speed up the turn off time but increase power dissipation in the resistor and the driving transistor. If this is acceptable try it, but see how hot the mosfet gets. Make sure you are confident its dissipation is reasonable under all conditions.

    I suppose the stuffed battery works but that energy is going somewhere. Either warming the battery or producing explosive hydrogen gas - probably best to check what's really happening there. A resistive load is more conventional here, for example a bank of car headlight bulbs.

    The other thing that works really well is replacing the stator with a thick-walled aluminium tube turned to the same inner diameter - it works directly as an eddy current brake (bear in mind that means it needs to dissipate all the heat safely, but it can run fairly hot without damage and you still have the alternator fan I assume).
     
    sq-aristo likes this.
  14. sq-aristo

    Thread Starter New Member

    Jun 1, 2013
    20
    1
    ok i used the diode and i think it fixed the problem but i measured the output of the transistor and it was taking too long to turn off. i think the only reason the mosfet wasn't getting hot was because it is overkill for the project.
    I have another project coming after this one where they wont be overkill so I may as well get this right now.
    I replaced r3 in the original diagram with a 1k resistor and the transistor was heating up too much so i have done a quick redesign.
    Two questions;
    Do you think it will work?
    Should i leave pin 6 (the output base) on the optocoupler floating?

    btw i will use a few bulbs in series to load the stator like suggested to save blowing myself up :D

    Here is the messy schematic
    http://sphotos-g.ak.fbcdn.net/hphotos-ak-ash4/s403x403/1002080_10151646739601308_1008607710_n.jpg
     
  15. sq-aristo

    Thread Starter New Member

    Jun 1, 2013
    20
    1
    oops just popped 2 transistors haha i will put them in opposite positions now
    pnp top npn bottom
     
    Last edited: Jun 12, 2013
  16. sq-aristo

    Thread Starter New Member

    Jun 1, 2013
    20
    1
    I apologize for posting so much without replies but I have spent the day getting this working. I fiddled with the two above posts for a while and i couldn't get the wave form to pull down more than 5V so i got angry pulled the circuit to bits and started from scratch with new parts.
    I built the first circuit that I posted after all i had it working before. I noticed the mosfet got warm after about 10 mins even though the snub diode was in place. Oscilloscope showed slow drop time so i changed R3 to a 1k and it worked better. Tried a 560ohm and it got hot a little quick so i used 2 1k in parallel.
    Tested the circuit for an hour and everything is great, a little hot but still touchable.
    thanks everyone for your help.
     
  17. richard.cs

    Member

    Mar 3, 2012
    162
    31
    In your circuit of post #14 you spotted the error with the two transistors swapped but you could use another small npn transistor instead of the optocoupler. I know you've already solved it another way but just for future reference.
     
    sq-aristo likes this.
  18. shortbus

    AAC Fanatic!

    Sep 30, 2009
    4,006
    1,527
    Instead of messing with an opto-coupler and transistors, wouldn't it be easier to use a mosfet gate driver? The gate driver gives the correct level shift for the gate and isolation in one single package.
     
  19. bertus

    Administrator

    Apr 5, 2008
    15,647
    2,346
    sq-aristo likes this.
  20. sq-aristo

    Thread Starter New Member

    Jun 1, 2013
    20
    1
    The reason for not using a driver is because it takes too long to get one to where I am. I think I will use gate drivers for my next project (dual h bridge) but I just wanted to get this going with the parts I had laying around.
    Bertus that was a good read thanks
     
    shortbus likes this.
Loading...