Can't keep the smoke in!

Discussion in 'The Projects Forum' started by zerog, Apr 4, 2010.

  1. zerog

    Thread Starter New Member

    Mar 22, 2010
    5
    0
    Hey folks - after smoking my 4th power transistor, I thought I should turn and get some professional help.

    I'm trying to drive a biggish magnetic coil (that's the sketchy inductor in my schematic). The desired output is a 5-6A DC current with a maybe up to 1A 60Hz sinusoid.

    Everything works fine for about 3 minutes and then I blow the collector emitter junction, the smoke comes out, and I can't put it back in.

    So, where am I going wrong? The transistor (I'm using MJ15024G) is rated to 250W, it is mounted on an old Intel P2 heat sink, with a GOB of heat sink grease. The heat sink gets quite warm to touch, but still touchable so probably around 60C. The power supply is a beefy lab unit that can easily pump out 30amps at 40V.

    Any and all suggestions greatly appreciated
     
  2. bertus

    Administrator

    Apr 5, 2008
    15,645
    2,344
    Hello,

    Can you post a complete schematic, with the driving part?
    (what is at the input of the 2K resistor).
    Where is the back EMF diode?

    Greetings,
    Bertus
     
  3. rjenkins

    AAC Fanatic!

    Nov 6, 2005
    1,015
    69
    You definitely need a clamp or recovery circuit of some sort.

    When you switch off the supply to an inductor carryng current, the voltage across it goes high (theoretically infinite) with the polarity reversed. The transistor collector will be pulled to hundreds of volts, causing it to break down.

    A simple flywheel diode would give best protection but cause a slow turnoff.
    For an analog drive circuit, a diode & resistor in series may work better.
    Something like 10 or 22 Ohms plus a diode, across the coil so the resistor takes current if the collector goes positive of the supply.

    I would also add a resistor from base to emitter on each transistor. This ensures any leakage does not turn on a transistor and cause thermal runaway.
    Try 1K on the TIP and 100 Ohms on the MJ
     
  4. Wendy

    Moderator

    Mar 24, 2008
    20,765
    2,535
    Standard Darlingtons such as TIP120 already have this resistor incorporated. Given the gain of this circuit I think the OP has too many transistors, the CE drop is around 1.2V when the transistor is on. 6 Amps X 1.2V is 8 Watts, pretty hefty, but you are loosing a lot of potential with this kind of voltage drop.

    http://www.learn-c.com/tip120.pdf

    My advice, add a clamping diode and the nMOSFET as shown (I've seen units at my local provider that can handle 50Amps). R4 is 100Ω and as close to the gate as possible. Keep the input digital, 10V to turn on and 0V to turn off, and switch quickly. A 555 can drive this circuit well.
     
  5. Bychon

    Member

    Mar 12, 2010
    469
    41
    When you need a current gain of about 1000, 2 transistors is not too many. Still, a MOSFET will do that job with no helper transistor. The freewheeling diode is not just a good suggestion, it is absolutely necessary. The wattage imposed on these transistors is reasonable for them to survive. That's why voltage punch through is the prime suspect here.
     
  6. Wendy

    Moderator

    Mar 24, 2008
    20,765
    2,535
    It's not two transistors, the TIP120 is a Darlington. The OP has three transistors. I did post this info, as well as put up a datasheet on the PDF file.
     
  7. rjenkins

    AAC Fanatic!

    Nov 6, 2005
    1,015
    69
    Thanks Bill, I was half asleep when I did that post & didn't notice it was a darlington.
     
  8. zerog

    Thread Starter New Member

    Mar 22, 2010
    5
    0
    Hey folks
    Thanks all for the timely, and copious amount of advice!
    Let's take care of the questions in order:
    Bertus - that's essentially the whole schematic - the driver is a simple fnc gen, with a DC offset voltage and the 60Hz sinusoid.

    Others:
    I'll hunt down a flyback diode - this indeed may be the cause. I thought as I wasn't switching it on / off and slowly ramp my DC I was going to be OK without the diode. But come to think of it, I hear the fnc gen flip a relay as DC increases and I'm sure there's a brief disconnect.

    The 555/nMOSFET solution working in a class D type mode would work except I need the DC offset of 5-6A. Unless I am misunderstanding what you intended?

    I'll also try to add a resistor to the BE Junction on the MJ.
    Thanks again all
     
  9. bertus

    Administrator

    Apr 5, 2008
    15,645
    2,344
    Hello,

    When driving it with a sinusoid signal the transistor will be in "linear" mode for quite some time.
    During this time it will dissipate a lot of heat.

    What is your goal with the magnetic coil?
    Do you want slow change in magnetic power?

    Greetings,
    Bertus
     
  10. zerog

    Thread Starter New Member

    Mar 22, 2010
    5
    0
    Unfortunately, I know I'm in pure class A operating like this. Extremely inefficient - but I need a slow (60Hz), and also relatively linear sweep of the magnetic field generated by the coil. I couldn't really think of another way to do this, but am open to suggestions while I await the opening hour of my local parts shop!
     
  11. bertus

    Administrator

    Apr 5, 2008
    15,645
    2,344
    Hello,

    Perhaps a subwoofer amplifier will do the trick:
    http://sound.westhost.com/project68.htm

    It may be overkill, but have a look.
    Lowering the voltage of the powersuplly will lower the output power.

    Bertus
     
  12. zerog

    Thread Starter New Member

    Mar 22, 2010
    5
    0
    I've investigated using big audio amps - unfortunately, the DC bias is in combination with the 60Hz sweep is what's killing me.

    I tried this combination (in the schematic) where the current source is my power supply, and the amp on the left was a 100W audio amp. C was something like 4mF - blew the output transistors on the audio amp. Shhhh, I haven't told anybody about that one yet!

    Anyhow, almost every audio amp I've looked at has coupling caps to kill the DC - and when you mess with those, you just shove them right back into class A again and you end up with the mess I've got!
     
  13. Wendy

    Moderator

    Mar 24, 2008
    20,765
    2,535
    OK, Linear.

    You want 6 Amps max.

    What is the power supply voltage?
     
  14. zerog

    Thread Starter New Member

    Mar 22, 2010
    5
    0
    Doesn't much matter to me - but right now I'm running @ 25V. As I mentioned in my first post, I can go up to 40V
     
  15. Wendy

    Moderator

    Mar 24, 2008
    20,765
    2,535
    OK, use just the TIP120, put a 0.1Ω 1W resistor on the emitter, and feed it a 0.6V waveform. What is feeding this transistor?

    [​IMG]

    The input resistance to the op amp is around 1KΩ (give or take), it is wired in a constant current configuration, so the current is self limiting. You can use qty 10 1Ω resistors in parallel if you can't get 0.1W, with the added benefit of being a lot more wattage.

    You might be able to go as low as 10VDC for the power supply, depending on the resistance of the coil. This will cool the transistor down considerably.
     
    Last edited: Apr 5, 2010
  16. ifixit

    Distinguished Member

    Nov 20, 2008
    638
    108
    Hi,

    You need a slightly different approch to your problem. These notes, and the attached circuit, are not ment to be an exact solution, but are to be taken as an example of how to proceed if you agree.

    First astablish a stable bias current of 5 Amps through the inductor. Use positive feedback from the collector to the base as shown with R3. Using feedback controls thermal runaway and helps to prevent back emf damage.

    The inductor resistance is 3.8 Ohms so this means a drop of 19 Volts, which leaves only 6 Volts for transistor headroom and the modulation voltage signal across the inductor. This may not be enough. You may need more than 25V supply depending on the inductance of the coil.

    Now inject the 60Hz modulation signal at the base via a decoupling cap and series resistance to control the exact amount of signal needed to get a 1 Amp p-p current delta in the inductor. Meter across R2, look for a p-p AC signal of 0.1V.

    Keep the headroom above the transistor collector just enough to do the job and not clip the signal. Too much wastes power in the transistor. The MJ15024G seems to be a good choice as its RθJC is only .7 °C/W. Dissapation should be around 30-40W, which is a Tj temp rise of .7 x 40 28 °C. Your heatsink choice should handle this easily.

    Hopefully, the inductor has an non-saturable core at 5 Amps of bias. What is the inductance anyway?

    Good Luck,
    Ifixit
     
Loading...