PWM problem

Discussion in 'General Electronics Chat' started by masa6614, Sep 18, 2010.

  1. masa6614

    Thread Starter Active Member

    Nov 9, 2008
    49
    0
    Hello,
    I'm trying to PWM an n-channel fet as a highside switch. As shown in the drawing, I'm using a powerex BG1A isolated high side driver board with two mosfets in parallel, ixys ixfn56n90p. i drew my layout and attached it, the board is meant for IGBTs, so its connections are gate, emitter, and collector. I just connected the emitter connections to the fet's sources and the collector connection to the drain.
    It outputs a +15V/-8V on the gate with respect to the 'source' to turn the fets on and off. It works when at 100% duty cycle, the fets turn fully on and the load sees 20V across it. However, when i try to PWM (20kHz, 10kHz, 2kHz) with varrying duty cycles from 10%-50%, the fets stay fully on. And the load sees a constant 20V. I probe the gate with respect to load ground and I see the PWM just fine as I want it, with voltages switching from 35V high to 12V low, which is the +15/8V gate drive that should be turning the fets on and off, yet they're remaining fully on.
    I was only able to get the fets to somewhat switch off, yet not fully all the way to 0V, when i lowered the switching frequency down to like 200Hz. That seems way too slow when the gate driver datasheet says its capable up to 20kHz. Is there something I might be doing wrong? Thanks.

    http://ixdev.ixys.com/DataSheet/DS100066(IXFN56N90P).pdf
    http://www.pwrx.com/pwrx/docs/db_00004c.bg1a.pdf
    http://www.pwrx.com/pwrx/docs/vla504_01.pdf
     
    Last edited: Sep 18, 2010
  2. timrobbins

    Active Member

    Aug 29, 2009
    318
    16
    That two expensive FETs to fry when playing around!

    Firstly, confirm performance with just a simple cheap FET - sya an IRF130 or something equally mundane.

    Then check what voltage levels you have on the gate. The positive swing exceeds the max rating (+30V) - so you have your first problem.

    For testing, you could use a 1k series resistor with 22V back-to-back zeners across gate-source for protection.

    Again try with just a simple resistor as load. You are probing the load resistance only when checking for PWM operation - and the load you were using was just simple resistance?

    The only issue I can see with the IGBT driver (I don't have much experience with them) - is the short-circuit detection function - the Vds must be less than 15V, which it should be when in the 'on' state.

    The driver is only meant for 20kHz max - so don't get your hopes up toooo high.

    Ciao, Tim
     
  3. eblc1388

    Senior Member

    Nov 28, 2008
    1,542
    102
    That's measured w.r.t. load ground. The voltage on the load is nearly +20V and this makes the gate-source voltage only +15V.

    To masa6614:

    You have a perfectly legitimate technical question so you should get in touch with the supplier and ask them for the reason of the slow performance of the driver.
     
    Last edited: Sep 19, 2010
  4. timrobbins

    Active Member

    Aug 29, 2009
    318
    16
    Sorry for the kurt phrasing - I meant:
    Then check what voltage levels you have on the gate, with respect to source. The positive swing you indicate from gate-source exceeds the max rating (+30V) of the FETs - so you have your first problem in using that IGBT driver.

    Can you confirm that you probed the voltage with the CRO across the load resistance when checking for PWM operation, and only saw a constant level? Can you confirm that the load you were using was just a simple resistor?

    Tim
     
  5. masa6614

    Thread Starter Active Member

    Nov 9, 2008
    49
    0
    i hooked up a 10k resistor to load ground, so it isnt drawing much current, i just wanted to see the load switching. I cant really probe the gate to source voltage without shorting the source to load ground due to the connections with earth ground on the scope. but i think it is a good idea to ask the supplier whats the deal and try testing on cheaper components. Thanks for the suggestions.
    also, i know they've got 18V back to back zeners on their driver board between the gate-source, though there is no resistance between them and the totem pole outputs of the driver. I shorted what should have been the gate resistor location, so i can put two gate resistors for the parallel parts. and if that short circuit desaturation current is being detected, it would turn off the output, which i tested by just disconnecting the connection to the drain, my problem seems to be the output is just staying fully on.
     
    Last edited: Sep 19, 2010
  6. timrobbins

    Active Member

    Aug 29, 2009
    318
    16
    Yes those gate zener's are fine - I didn't see them on first pass.

    Have you got link J1 in? I woulod guess that i should be out.

    Have you shorted the fault input - your circuit schem doesn't show you haven't ?

    Tim
     
  7. eblc1388

    Senior Member

    Nov 28, 2008
    1,542
    102
    You should repeat the test with a load of 1A or more, just to be sure.

    The 10KΩ load is too high to give a meaningful result.
     
  8. masa6614

    Thread Starter Active Member

    Nov 9, 2008
    49
    0
    Okay, figured out the problem. Also connected to the outputs was a programmable load unit, even though it wasnt enabled, it was somehow screwing things up and affecting the output. As soon as i disconnected it, things started working as expected. Probably should have put that in the picture too, but i kinda forgot it was hooked up. Got a nice clean square wave up through 25kHz now.
     
Loading...