IGBT Failures

Discussion in 'The Projects Forum' started by NoSmoke, Nov 13, 2011.

  1. NoSmoke

    Thread Starter New Member

    Nov 13, 2011
    5
    0
    Hello all, new to forum and fairly new to electronics.

    I am attempting to build a load controller for a windmill with a 3ph alternator, the objective being to unload the alternator at low wind speeds (so the windmill won't stall when the wind picks up) and to increase or decrease the load later as the wind speed increases or decreases. I have designed a circuit that rectifies the incoming AC and feeds it through an IGBT to the load. The IGBT is controlled by a Picaxe micro via a driver chip and PWM from the Picaxe. My test setup derrives input power from the 120VAC mains. I am running the PWM at 500Hz over duty cycles from 0 to 100%.

    The circuit seems to work fine once underway but the problem is the IGBT sometimes fails on power up (when plugging load power into the mains). What happens when it fails is that the load (incadescent lamps of 900W total) immediately come on full brightness and the IGBT is found to be shorted internally. I have placed diodes and caps across the load and IGBT C-E terminals to try to dampen voltage spikes. An external 5V zener is there to protect the gate.

    A schematic of the circuit is attached. I would appreciate any comments or suggestions of how to avoid this problem.
     
  2. praondevou

    AAC Fanatic!

    Jul 9, 2011
    2,936
    488
    Your electrolytic capacitor symbols are inverted.

    Incandescent light bulbs have a much lower resistance when they are cold than what you would calculate by their power / voltage rating. Measure the resistance of your load to have an idea of the inrush current you may get.
    At the moment you connect the power supply, are you powering up the PWM circuit as well or is it already energized?

    I couldn't find a datasheet. Is it the STGB35N35LZ? Can you post the datasheet?

    You have a 6A gate driver. Lower the 1k resistor to e.g. 10R. The zener diode value should be higher than the gate pulse voltage and lower than the maximum Vge of the IBGT. We really need a datasheet but I doubt the IGBT is supposed to be driven with only a 5V gate signal.

    EDIT: Ok, I got the datasheet. It has a clamp circuit on the gate and an internal gate resistor. So you can remove the 1k gate resistor and the zener. Gate threshold voltage is also extremely low. 5V drive should be ok.
     
    Last edited: Nov 13, 2011
  3. NoSmoke

    Thread Starter New Member

    Nov 13, 2011
    5
    0
    Thanks for the fast response praondevou.

    Good point about the low light bulb cold resistance. I hadn't considered that. When the load power is connected, the Picaxe is powered up but in reset mode so no PWM is going to the IGBT. The lamps should therefore remain off until the PWM is initiated. If I can get past that stage, when the PWM is initiated, everything seems to work fine over the entire range of duty cycle but I guess it's possible I am still damaging the IGBT such that it fails next time it is powered up (?).

    I was unsure as to what gate resistor would be required (a 1K was recommended by others on another forum for a similar IGBT so I just went with that). I'll remove it.

    I am also going to put a resistor & diode across the large filter cap to limit charge current surge.

    Oh, I'll also change the electrolytic cap symbols. I was unfamiliar with that notation (it came from the DesignSpark schematic/PCB program) and didn't know which terminal was which.

    Thank you again for the assistance.
     
  4. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    You're indicating a 900W load; since the supply is 120V, that's 7.5A current.

    That's pretty close to the typical Vce(sat)=1.15 for Ic of 10A, but it could be as high as 1.7V. That's anywhere from ~8.6W to 12.75W power dissipation in the IGBT, which is a fair amount of heat to get rid of.

    And like praondevou mentioned, for that 1st few mS when power is first applied across the load, it'll look almost like a dead short.

    You might try starting off using a very short ON duty cycle for a few hundred mS, and gradually ramping it up over a second or three until you get to the desired intensity. That should also help extend the life of the bulbs.
     
  5. NoSmoke

    Thread Starter New Member

    Nov 13, 2011
    5
    0
    Sgt, thanks for the comments. The particular IGBT I'm using is rated at 40A so I am well below but, as you say, the first few ms will be a lot higher than normal. I will program the Picaxe to start out with a zero duty cycle and ramp up for a bit as you suggest.

    I'm not sure though how that relates to my more immediate problem, that being the IGBT shorting out (C-E) on load power connect and there is no PWM signal being sent. Might it be that I am damaging the IGBT (from the aforementioned strong surge currents) during times when it appears to be working normally such that the damage causes a failure on the next power-up(s)? Another thing I could try is using a electric heater for a load which should not exhibit such a low start-up resistance (rather than light bulbs).
     
  6. praondevou

    AAC Fanatic!

    Jul 9, 2011
    2,936
    488
    I don't think that it was "slightly" damaged before and then the next time you turn it on it has a shorted CE.
    In high power applications there is often a negative gate voltage applied when the IGBT is OFF to make sure that it stays OFF.
    I wonder if the 1k gate resistor might have been a problem. The drivers low impedance is increased because of this resistor (looking from the IGBTs gate)

    Did you try to power up the circuit again after removing the resistor?
     
  7. NoSmoke

    Thread Starter New Member

    Nov 13, 2011
    5
    0
    Interesting - I hope that's the cause of it. Haven't tried it yet - waiting for my new batch of IGBTs to arrive from Digikey (should be tomorrow).

    I'll post the results...........
     
  8. NoSmoke

    Thread Starter New Member

    Nov 13, 2011
    5
    0
    Looks like it works now! Removed the 1000 Ohm gate resistor (& zener) but the IGBT still failed at 900W load even with slow PWM start. Then replaced the 900W of light bulbs with a 1200W heater - now it appears to work fine.

    I measured the light bulb resistance of about 1 Ohm (which would give a surge of 150A with the 150Vs DC at the filter cap!). The electric heater OTOH, has a cold resistance of about 13 Ohms.

    The only other thing I need now is a larger heat-sink (I hope).

    Thanks much for all the help!
     
Loading...