SMPS increasing PWM IC gate pulse output

Discussion in 'General Electronics Chat' started by seanbein, Feb 27, 2012.

  1. seanbein

    Thread Starter New Member

    Mar 16, 2010
    21
    0
    Hi all,

    I have tuned my PWM IC UC3842 at D~0.5 and f~50khz.
    It seems that the output of the PWM IC has about 20v if tested alone, however, when tested (attached)to the gate of a MOSFET it is weird to find that the output is at most about5~10v only. Can anyone explain why?If so, how do i increase it? Also, the waveform is no longer a square pulse wave but it seems it is puncturing in a very distored form like a sawtooth basically.
     
  2. w2aew

    Member

    Jan 3, 2012
    219
    64
    It sounds as if the PWM IC doesn't have a low enough output impedance to drive the gate capacitance of the MOSFET.
     
  3. magnet18

    Senior Member

    Dec 22, 2010
    1,232
    124
    try looking for some high-speed low-capacitance MOSFET's
     
  4. seanbein

    Thread Starter New Member

    Mar 16, 2010
    21
    0
    http://www.vishay.com/docs/91124/91124.pdf
    is 980pf low enough?im suspecting the mosfet has been damaged,no voltage across the drain-ground and source-ground and my ac supply transformer connected to a circuit breaker keeps tripping WHENEVER i connect the MOSFET to the PWM IC now.What do you think?

    ALSO, I would like to enquire what are the possible causes of the sense resistor rated 1.5R,8W being burnt at the source of MOSFET, I am trying to find out the value of sense resistor required for duty cycle control
     
  5. magnet18

    Senior Member

    Dec 22, 2010
    1,232
    124
    i cant back it up with math, but I would guess yes

    have you tried different fets? probably either the fet or the IC, but its hard to tell with out a schematic

    over current might be it if the mosfet is indeed blown
     
  6. praondevou

    AAC Fanatic!

    Jul 9, 2011
    2,936
    488
    What MOSFET DID you use?

    Did you have the sawtooth waveform before or after the FET burnt?
    Can you post a complete circuit diagram of what you built?

    Is the supply voltage of the UC3842 20V?
    20V gate voltage is at the limit of what most MOSFETs can handle.
     
  7. seanbein

    Thread Starter New Member

    Mar 16, 2010
    21
    0
    I used a fqp6n60c mosfet.
    I have the "sawtooth like" with very high distortion right after i connect the mosfet to the circuit.When MOSFET is removed, it can be seen that the waveform of output gate is actually veyr high.

    The supply voltage of uc3842 is about 14.3~15v
    http://www.freeimagehosting.net/id58c
     
  8. praondevou

    AAC Fanatic!

    Jul 9, 2011
    2,936
    488
    16k5 for the startup resistor looks like quite low to me. How did you establish this value? This resistor provides current to startup the IC, after that the IC is supplied by the additional feedback winding. I remember having used 56k to 100k for 125VDC, have to check it though.

    If as indicated in your diagram the feedback winding is open for the test you did than the current to charge/discharge the gate of the FET will have come from the 10uF capacitor and eventually also through the 16k5 resistor. I think what you are seeing is quite possible if the feedback winding is not connected.
     
  9. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    You show that you are using a 1N4007 diode from the transformer low voltage tap; that diode is far too slow to use. You will need to use a Schottky or ultra-fast diode instead.

    The FQP6N60C/FQPF6N60C has an Rds(on) of 2 Ohms when Vgs=10v; that's quite high. Unless you are planning on having >450V on the 220uF filter cap, you really should consider a MOSFET with a lower Vdss; this will help a great deal to reduce the Rds(on).

    You don't show what your input AC voltage is, so we can't really determine if your 16.5k value is appropriate - or is it that your input is 14.3-15V AC? It looks like you are testing it with 14.3v to 15v on Vcc.

    You don't show your circuit layout.

    The 1.5 Ohm current sense resistor needs to be a non-inductive type.
    The ground side of the current sense resistor needs to be close to the UC3842 ground, and the bypass caps for the UC3842 need to have leads that are as short as possible between the IC's Vcc and GND terminals.

    I suspect that you are trying to breadboard the circuit, which simply won't work very well. The traces between the components need to be as short and wide as possible to reduce parasitic inductance. A breadboard-ed circuit that looks like a plate of spaghetti will have a tremendous amount of parasitic inductance.

    50kHz might not sound like much, but you need a 500kHz bandwidth or better to pass a reasonable approximation of a square wave; this is because an ideal square wave is composed of the fundamental frequency, plus ALL of the odd harmonics of the fundamental frequency - which implies a requirement for unlimited bandwidth.
     
  10. Ralph33000

    New Member

    Feb 28, 2012
    3
    0
    Keep an eye on gate capacitance AND gate charge at various gate voltages. They are closely related but, when a data sheet specs, e.g., 870pf gate capacitance, you need the graph to see at what gate voltage that applies.

    Also cheaper parts only give a gate capacitance spec in a table; better parts give a graph to go along with that.

    I have been bitten before by SMPS chips that didn't spec the gate drive current. You may well need over 2 A gate drive for 980 pf gate at 50 KHz.

    Keep in mind too that before TI bought Benchmark, Benchmark was notorious for playing tricks in their data sheets; they would give a representative schematic in a data sheet which would give problems. Then the FAE would say, "oh, didn't you know that you have to include this circuitry as well ....". And the board real-estate would grow blowing your estimates.
     
  11. seanbein

    Thread Starter New Member

    Mar 16, 2010
    21
    0
    Thank you all for your help,
    I have tried everything mentioned, and yes I am using a breadboard, the result was:
    I got a 44v output at the secondary when I turn on for 10 seconds.The waveform for gate output and corresponding MOSFET D-S was close to theoretical waveform. Then once off, there was still some voltage inside and then for the next 10 seconds the MOSFET was damaged.
    *I tried it without snubber, I am thinking it could be due to the absence of snubber components, however, if it is SO, WHY DIDNT THE MOSFET get damaged within the 10 seconds while in ON state?

    Thank you all for your attention
     
  12. twister007

    New Member

    Feb 29, 2012
    10
    0
    I static tested some mosfets and they wouldn't turn on very good with a resistor on the source. They seem to work better with the source grounded. But I know nothing of SMPS. Would a current sense on the output side work better?
    How much modulation will that chip do? I need something that will modulate the pulse from 0 to 100% for a solar panel voltage regulator. Do you think this chip will do that? But it would have to control the voltage, not the current.
    I would also check the voltage across the current sense resistor to make sure you're not overloading that transistor. Just turn it on for a second, not 10 seconds! Try it without drive, and then with drive. If it is overloaded, either the transformer is wrong or the pulse width is wrong, or your heat sink is wrong, or your gate return resitor is wrong. Let us know what happens!
     
    Last edited: Mar 2, 2012
Loading...