Problem with inverter

Discussion in 'General Electronics Chat' started by leafar, Jun 17, 2012.

  1. leafar

    Thread Starter New Member

    Jun 3, 2011
    13
    0
    First sorry for mistakes english is not my first language. I implemented a full bridge inverter with spwm using a dspic 30f4011 (that has a pwm module), the drivers ir2110 and the igbt's irg4pc50ud, to control the light intensity of a incandescent bulb. The images show the circuit (i show various images for a better seeing of components):

    Comlete circuit:
    http://s10.postimage.org/hik5na3bd/completo.png

    Image for stage from output of dspic to driver's input:
    http://s14.postimage.org/y1owiucep/control.png

    Image for igbt's stage:
    http://s18.postimage.org/o5q2kjekp/potencia.png (the gate resistor value is 27 ohms)

    For isolate the control stage and the power stage i used four 6n136 optocouplers. The pwm signal is inverted because of optocouplers so i used a 7404 to recover the original signal. I used 5 volts for logic part and 18 volts as source for the output signal of the ir2110 . The power rail has 170 volts (120 ac rectified and filtrated).

    However, the thing is that when i tested the circuit it just worked for like 15 minutes and then it seems there was a Shoot-Through Current. Two igbt from the same arm got damage. I think that both went on at the same time but i don't know why since i previously set the maximum dead time in the dspic, saw the gate signals in the oscilloscope and they look good with the dead time that i set and the form was ok. Any suggestion about the problem?
     
    Last edited: Jun 17, 2012
  2. shortbus

    AAC Fanatic!

    Sep 30, 2009
    4,004
    1,525
    How big of a heatsink did you have on your IGBT's? If the IGBT thermal limit is surpassed they will be destroyed.
     
  3. leafar

    Thread Starter New Member

    Jun 3, 2011
    13
    0
  4. shortbus

    AAC Fanatic!

    Sep 30, 2009
    4,004
    1,525
    Really not very big but probably OK. Could you post a photo of your board? Is is built on a breadboard or PCB?
     
  5. takao21203

    Distinguished Member

    Apr 28, 2012
    3,577
    463
    Did you use thermal grease?

    Also yes if Ampere limit at higher voltage is exceeded, MOSFETs, IGBT etc. will likely fuse immediately.

    Have you considered to use VGA cooler for this? They are small and have thermal capability of about 40W, and more for larger one's.

    The black sink you show has max. capability of 10W.

    Do you use any kind of NTC inrush current limiting, or fuse?
     
  6. leafar

    Thread Starter New Member

    Jun 3, 2011
    13
    0
    The circuit is built in a pcb. I dont' have a camera rigth now but i'll post a photo later.

    Sorry for (probably) trivial questions:

    I didn't use thermal grease. Is it very neccesary?

    mm i think i'll use VGA cooler but do i have to put one for each igbt?

    No i don't use any kind of NTC inrush current limiting, and fuse, just one in the main rail.
     
  7. shortbus

    AAC Fanatic!

    Sep 30, 2009
    4,004
    1,525
    The thermal grease is needed. the surfaces of both the device and heatsink are not completely flat, even though they look like it. The thermal grease is used to make a complete thermal connection between the two parts.
     
    spark8217 likes this.
  8. praondevou

    AAC Fanatic!

    Jul 9, 2011
    2,936
    488
    18V gate voltage is quite near the maximum rating of the IGBT. Why not use a voltage a little lower, e.g. 15V?

    It may be a dissipation problem as was suggested before.
    One thing you need to know about the 2110 is that it's high side can latch up, i.e. the output remains on even if the input signal says otherwise.
    This can happen if the voltage on VS gets lower than the voltage on COM. Proper layout is important here. Have a look at this application note which I posted a few times already. http://www.irf.com/technical-info/appnotes/an-978.pdf
    Section 5 page 9.
     
  9. leafar

    Thread Starter New Member

    Jun 3, 2011
    13
    0
    Hi guys. I'm still stuck... i changed the gate voltage to 15. Before make other test with 170 Vdc i just dismount all and tried to make a simple test in a protoboard with low power rail voltage (20 volts) and a load (resistor) of 22 ohms with only a half bridge. But there is a very odd problem. The gate signal doesn't seem to turn off the igbt's at the correct time. I even used a drive with negative bias like the one used in pag. 21 of this document http://www.irf.com/technical-info/appnotes/an-978.pdf but the problem remains. The strangest thing is that it seems that the low side affects the high side and vice versa so both igbts turn on fine with gate signal but turns off with the change of state of the other. Is very very odd and honestly i don't know what to do.
     
  10. leafar

    Thread Starter New Member

    Jun 3, 2011
    13
    0
    Here are the photos of high side Vds vs Vgs,

    [​IMG]

    Uploaded with ImageShack.us

    and low side Vds vs Vgs:

    [​IMG]

    Uploaded with ImageShack.us

    In both cases the bottom signal is the gate-source signal. You can see that there is a delay or lag of 4.2 uS between the off signal of gate and the Vds signals. The Vds signal must goes up when the gate signal goes down almost inmediatley right? And as you can see that is not happening. Any idea of what could be the problem?
     
  11. mik3

    Senior Member

    Feb 4, 2008
    4,846
    63
    The IR2120 has not an internal dead time. You need to provide a 1us (at least) dead time in software.

    Also, use turn off resistors in series with the gate diodes to reduce voltage spikes.
     
  12. leafar

    Thread Starter New Member

    Jun 3, 2011
    13
    0
    Yes i provided a dead time of 1.5 us in the dspic. I follow you advice about turn off resistors but the problem remains
     
Loading...