SMPS shorts and kills mosfets

Discussion in 'The Projects Forum' started by Juhni, Feb 24, 2013.

  1. Juhni

    Thread Starter New Member

    Feb 22, 2013
    4
    0
    I've been building this uzzors2k's SMPS:
    http://uzzors2k.4hv.org/index.php?page=benchsmps1
    There has been already 3 pairs of dead mosfets so little advice would be definitely needed!

    I haven't connected the GDT to the power section yet, so IRFP450s' gates are not connected to anything. For 250V 3.3µF and 250V 3µF capacitors I'm using these:
    http://www.ebay.com/itm/Yellow-3-3u...al_Components_Supplies_ET&hash=item53f28549fb

    I do not know much about mosfets but according my test with two IRFP450 connected to each other like in SMPS's schematics, if there is voltage on connection of source and drain, the mosfets will open. I noticed this by using my linear power source.

    The strange phenomena that concerns me is that if the mosfets are not connected the circuit and I power it on, there's about 150 volts on that source-drain connection point, though it starts to decrease when I measure it towards negative rail. But that part of circuit is DC and according my knowledge DC won't go through capacitors. But because this DC is rectified from mains, could it be that ripple voltage is causing this voltage? And by plugging voltage meter to it it would draw that minimal current slowly because of voltage meters high impedance?

    The ripple is about 8 volts and 50Hz measured by oscilloscope:
    [​IMG]
    1volt/DIV and 5ms/DIV

    So are my capacitors right type or what should I do? Bleed the source-drain current to negative rail? I have only one try after which I'll have to buy more mosfets!
     
    Last edited: Feb 24, 2013
  2. praondevou

    AAC Fanatic!

    Jul 9, 2011
    2,936
    488
    Never do that. Either you connect to the correct voltage source or you connect the gate directly with the source.

    If you want to test your circuit except the FETs do the latter.

    Leaving gates open will lead to completely unexpected switching behaviour.
     
  3. panic mode

    Senior Member

    Oct 10, 2011
    1,321
    304
    are you sure that outputs of pulse transformer are wired correctly (GDT)? are you sure you follow dot convention? do you know what would happen if one of the output windings is connected backwards (both dots going to gate or both dots going to source)?
     
  4. takao21203

    Distinguished Member

    Apr 28, 2012
    3,577
    463
    The gate drive is extremly bad, as these cores can saturate easily. Additional components are neccessary.

    Such a gate drive only should be used for lower voltages, where you have more reserves for the MOSFET.

    What are the ratings for this MOSFET?
     
  5. Juhni

    Thread Starter New Member

    Feb 22, 2013
    4
    0
    Well about those dots... It affects the phase, I probably designed the PCB that both dots are going to the gates, which is wrong I assume? How they should be connected?

    This is what comes from GDT, I know it isn't pure square wave but I though that it could work, I would make a better GDT later:
    EDIT: Or maybe I should just make a better one to begin with...
    [​IMG]
    5 volts/DIV and 5µs/DIV

    Are those dots read like:
    [​IMG]
     
    Last edited: Feb 24, 2013
  6. takao21203

    Distinguished Member

    Apr 28, 2012
    3,577
    463
    Have you some day tried to fire a digital camera (with flash) close to such a transformer?

    I mean when the circuit is working normally.

    I got some circuits with the regulator IC actually turning off.

    These are not proper gate driving circuits even if they may work. If they are not sensitive to EMI they are a little better but you need to test for that. Or you may get a nasty surprise some day.
     
  7. panic mode

    Senior Member

    Oct 10, 2011
    1,321
    304
    well, they should be connected according to schematic. so only one MOSFET is triggered at any time. if both dots are connected to gates (or both dots are connected to source), then MOSFETS will fire at the same moment (same phase).

    this means that they are going to create short circuit for your DC supply and MOSFETs will be toasted instantly. :p
    the 230VAC supply with 8A rectifier plus 400uF capacitor can provide plenty of energy, way more than two MOSFETs can handle.

    see, in normal operation only one of them conducts and in either case load is series combination of 3.3uF/250VAC capacitor and primary side of the "power transformer". even if transistor was to stay on for a while longer, capacitor would limit energy transferred.

    but when both mosfets are fired at the same time, all energy only goes through MOSFETs. they may be able to handle 14A (at low voltage) or 500V (at low current) but in this case both voltage and current are large and this is way above their power rating (190W) so they go "POOF!".
     
    Last edited: Feb 24, 2013
  8. JMac3108

    Active Member

    Aug 16, 2010
    349
    66
    Yes, NEVER do that! The MOSFET gates will almost always float such that the MOSFETs turn on continuously.
     
  9. takao21203

    Distinguished Member

    Apr 28, 2012
    3,577
    463
    Get a ballast from an old CFL fixture. It will allow the caps to charge to full DC, but as soon as you draw current, the voltage will go down.

    This can save your MOSFETs until you removed all bugs.

    Or a lightbulb etc.
     
  10. Juhni

    Thread Starter New Member

    Feb 22, 2013
    4
    0
    Okay I tested the circuit with TL494 and applyed max 24 DC volts after the rectifier, everything seemed to work. I checked the MOSFETs and the SMPS's output voltage with oscilloscope and they seemed to be all right. And when drawing 0,75A and 24V from the linear power supply, the SMPS gave 4,6A and 6.5V. (The power transformer isn't winded exactly to 1:2.4.)

    And then I moved to the AC mains: it worked with TL494 turned off but after I turned it on, automatic fuse on mains (about 4 amps) blew and the MOSFETS were dead, again.

    Damn I'll have to buy new ones and it takes couple of weeks them to arrive..

    I have this 2000VA transformer 230V-24V 4A and adjustable transformer which can change voltage from 17 volts to 280 volts when using the 24V output. Can I use this transformer to slowly level up the voltage so next time the mosfets wouldn't burn? I'm wondering would a circuit designed to work with 230V work with lower AC volts and 50Hz?

    Thank god I have that 10 amp autofuse, which I have adjusted somewhere 4A (it wouldn't go lower). At least the mosfets haven't blown on my face yet.

    So I am in a need of little advice now! Some things just have to be learnt the hard way. Oh I'm waiting for my graduation after which I can learn these things in university, not only by myself :D

    EDIT: OH, there was one old IRFP450 in my mosfet box after searching replacements for testing: now I'll stick with linear PS to powet up this mosfer eater. But can that adjustable transformer be used to test this with lower AC volts?
     
    Last edited: Feb 25, 2013
  11. Juhni

    Thread Starter New Member

    Feb 22, 2013
    4
    0
    Can't edit the last post second time?
    Well after using linear PSU to power up the SMPS and max 1A I found that there is exacly half of the operating voltage on source-drain connection, is this bad? Though the SMPS work now again: 24V on rectifier's negative anf positive line and current is 0,75A and the SMPS output is 6,5V and 4,6A.
     
Loading...