High Voltage MOSFET Drivers

Discussion in 'The Projects Forum' started by ams0178, Aug 5, 2008.

  1. ams0178

    Thread Starter New Member

    Aug 5, 2008
    8
    0
    Hello Everyone,

    I came upon AllAboutCircuits while trying to research info about my latest project. There was a similar topic to my problem addressed in bigbigblue's thread, http://forum.allaboutcircuits.com/showthread.php?t=2262. I believe my application and problem is different than his.

    To give everyone a little bit of background on the project I'm hoping to build an AC Power generator that can produce 90-260VAC, 45-65Hz, and deliver 2-4Amps to a load. That's very high level and the project details haven't all been worked out and as I've been working the project becomes more defined. Look at http://www.avrfreaks.net/index.php?name=PNphpBB2&file=viewtopic&t=64487
    to see a history of my progress and what issues I've thought about. The project centers around an Atmega2560, which is why I was posting on the AVRFreaks web site.

    My most recent problem is taking a low level pwm signal from my microcontroller to drive high power mosfets in a bridge configuration. I've been working with this issue for over a month now and been getting great use out of my safety glasses. I've tested my sinwave pwm scheme at low power levels and been able to produce a +/-15V sin wave using a full H-Bridge. I'm using FAN73832 Fairchild half bridge drivers for each side of the H-Bridge. At high power I'm using a rectified AC voltage (170VDC) for the top of the H-Bridge. I've been using the SPP04N50C3 MOSFETs from Infineon. At this point I'd simply like to have a working prototype that drives HV through an H-Bridge, regardless of what my specs are. I'm constantly going through the driver chips (which have a rating of up to a 600V rail) and as a result go through MOSFETS as well.

    I believe the issue is with the bootstrap capacitor and the sizing on that. I realize that this affects the pwm frequency and duty cycle. At times I simply test a 50% duty cycle to get away from the more complicated sinwave pwm scheme. Could anyone help me with the bootstrap design? Or does anyone know of any discrete MOSFET driver designs I could use with high voltage rather than an IC solution? I've seen the designs on this website, http://www.innovatia.com/Design_Center/High-Side Drivers.htm,
    but they are for low power and I haven't seen any elsewhere. I'm really stuck at how to drive my H-Bridge under the 170VDC rail. Any help would greatly be appreciated.

    Thanks, Alan
     
  2. Tahmid

    Active Member

    Jul 2, 2008
    344
    25
    Hi Alan,
    The problem you are facing with full bridge circuit, is facing or faced by almost all, who tried or trying to use full bridge configuration for converting high voltage dc to 50/60 Hz sine AC. I also faced this problem and could overcome with lot of effort and losing almost half a century mosfets and drivers.
    Before I provide suggestion to you, please tell me the followings.

    How you tried to drive the mosfets? Simultaneously firing both highside and lowside mosfet or keeping highside one on for full half cycle and pwm-ed the low side one with high frequency and doing the other pair in the same way in the opposite phase. What is your modulating frequency?

    Have you faceed the same problem with the test with 50% duty cycle pulses? what was the freq. in that? How you have fired the mosfets in 50% duty cycle test?
    Please let me know the informations I asked for and then I hope I can help you.
    Please let me kindly know the modalities of your sine routine also if possible.

    With thanks.
     
  3. ams0178

    Thread Starter New Member

    Aug 5, 2008
    8
    0
    Hello Tahmid,

    I'm hoping to answer all of your questions. Thanks for the help.

    I'm driving the MOSFETs with half bridge drivers, so I'm using two half bridge drivers from Fairchild, FAN 73832. I've actually run the full bridge under a 15V test supply with my sinwave pwm scheme. The pwm frequency is 62.5KHz and I constantly vary the duty cycle through a look up table in software. The software scheme is correct. The H-Bridge driver chips have a dead time control by placing a resistance off of one of the pins to ground. I've also tried testing a 62.5KHz, 50% duty cycle through a single half bridge. It's actually the half bridge driver chips that keep blowing up and not the MOSFETS, which is why I think it is a bootstrap circuitry issue.

    Please let me know if you have any other questions or comments.

    Thanks, Alan
     
  4. ams0178

    Thread Starter New Member

    Aug 5, 2008
    8
    0
    Here is a copy of my schematic with some text notes.

    Alan
     
  5. Tahmid

    Active Member

    Jul 2, 2008
    344
    25
    Dear Alan,
    I am studying your schematic. Tomorrow I hope to answer.
    Thanks.
     
  6. Tahmid

    Active Member

    Jul 2, 2008
    344
    25
    HI alan,
    I have studied your schematic and found almost alright. Just you don't require to put 100k resistors and just connect shutdown pin directly with ground with out the 100k resistor. Probably the bootstrapping process is not function properly and it is shorted out. But one mystry. If it is shorted out, then your mosfets should be shorted out but you said that mosfets remain intact. So, check your physical connections properly. Even than if it is not alright, then your firing process/sine table should be checked. You cannot rely on testing with low voltage, because all the problem starts from >100 volts. How you are firing the high side and low side pair mosfets and also other pair? Probably that is the key. If you let me know, I can think about it. Thanks.
     
  7. ams0178

    Thread Starter New Member

    Aug 5, 2008
    8
    0
    Tahmid,

    Thanks for the input. You're right I don't actually "need" the 100k resistors. The half bridge driver chips actually have a smaller deadtime before switching on the high to low or vice versa so that both sides of the bridge aren't on. The datasheet states that adding a resistor to ground from that pin will increase the deadtime (only up to 220k). I tried testing another bootstrap configuration yesterday that included a larger electrolytic capacitor in parallel with a smaller ceramic capacitor as recommend by an application note, http://www.irf.com/technical-info/designtp/dt04-4.pdf . This is one of the more comprehensive ap notes on high voltage gate drivers that I've seen. During testing I only tried a half bridge configuration and 50% duty cycle. First testing at a 17.5V rail (the same as the chip supply) worked fine and then connecting a rectified AC rail, 170VDC, the chip blew.

    I feel as though I've gone over my circuit repeatedly in the last month. I agree, it appears that there could be a wrong connection. But the low voltage pwm signals are driven through to the MOSFETs, so I would think that if it is connected wrong then something would happen at low voltage.

    At this point I just want to run one half of the bridge with high voltage and run a 50% duty cycle through it. I just received another driver chip from digikey that I'm going to try, the IR2104. Ap note is here http://www.irf.com/product-info/datasheets/data/ir2104.pdf. I'm starting to think its something small that I'm overlooking in my setup environment. I just don't know.

    Any other pointers/comments are greatly appreciated.

    Thanks, Alan
     
  8. Tahmid

    Active Member

    Jul 2, 2008
    344
    25
    Hi Alan,
    Your bootstrapping capacitor which is of .22uF is alright and those should be of film or tantalum or at least ceramic, not electrolytic. Your carrier frequency is 62.5 khz and hence the bootstrapping capacitor has to create virtual ground for upper side mosfets for 8u sec and the capacitors you provided are completely perfect. Bootstrapping is not your problem. I have sorted out your problem.
    You have used FAN 73832 half bridge drivers, which are good drivers but they can not be used for sine wave inverter. Because, they are having only one input and so your microcontroller is providing one signal to each driver and driver is providing 2 outputs. For sine wave generation, you require 2 inputs for each driver, so that 4 signals can be utilized independently from microcontroller for that purpose. You have to drive one highside and one lowside mosfets with 2 pwm-ed independent signals from microcontroller for first half cycle and then other highside mosfet and low side mosfet in reverse phase with 2 signals from your microcontroller. Either you use bipolar spwm or unipolar spwm-ed topology for generating sinusoidal waveform, you require 2 separate pwm-ed signals for 1st half cycle and 2 separate pwm-ed signals for next half cycle and in reverse phase. I hope I could make you understand.
    Reschedule your sine table and use 4 signals for 4 mosfets and use proper low pass lc filter and you will achieve what you want. It is not easy but can be achievable with lot of hard work.
    Thanks.
     
    Last edited: Aug 7, 2008
  9. ams0178

    Thread Starter New Member

    Aug 5, 2008
    8
    0
    Tahmid,

    I understand what you are saying, but I have already achieved a sinwave through my scheme at +/-15V with these drivers and the MOSFET H-Bridge and filtering the output. My microcontoller sends two pwm signals that are directly out of phase from each other. One signal goes to one driver chip and the other signal to the other. This causes one chip's input to be high which drives its high side on and the low side off, while the other side to be off which drives this chip's high side off and the low side on. Sinwave generation can be obtained with these drivers which I have seen on my oscilloscope. And I understand the driving scheme you propose, but it complicates the software that would be necessary as well as timing. The drivers I'm using will guarantee that both MOSFETs on the same side will never conduct at the same time.

    Thanks for the input. I feel like I have a rebuttal to all of your suggestions, but please keep the comments and recommendations coming as it stimulates my thinking and might lead me to a resolution.

    Alan
     
  10. Tahmid

    Active Member

    Jul 2, 2008
    344
    25
    Hi,
    Alan, you should try what you have proposed and continue your efforts. If you are successful, please let me know, as I could not do the way you are trying. Good Luck!

    Thanks.
     
  11. ams0178

    Thread Starter New Member

    Aug 5, 2008
    8
    0
    Tahmid,

    If you're interested here is a response to a question by IR about using the IR2104 chip that I just ordered from digikey, http://irf.custhelp.com/cgi-bin/irf...D1oYWxmIGJyaWRnZSBkcml2ZXI*&p_li=&p_topview=1

    In reading IR's answers to some of the other H-Bridge questions I believe it may be a control issue and how I initially power up my setup. I've been running my microcontroller first with pwm outputs as well as VCC to the driver chips. I then apply the rectified voltage. I'm thinking that I need to make sure the pwm input is first low due to allow the bootstrap to charge. Or does the bootstrap charge as long as VCC is applied and there is no DC rail?

    Alan
     
  12. ams0178

    Thread Starter New Member

    Aug 5, 2008
    8
    0
    Hello Everyone,

    I believe I found an issue with the half bridge driver chip, FAN73832, http://www.fairchildsemi.com/ds/FA/FAN73832.pdf.
    In the attatched oscilloscope photos you can see the issue. The photos show two signals corresponding to one half bridge output's two gate drive signals. The upper signal is the high side MOSFET's gate drive and the lower signal is the low side MOSFET's gate drive. I was running the driver chip at 18V with SPP04N50C3 MOSFETs at a DC rail of 18V. The input to the driver is a 62.5KHz pwm signal at 50% duty cycle.

    The problem appears to be a lack of fully turning off the high MOSFET's gate drive. Taking a look at the photos when the high turns off and the low turns on, the high gate drive only decreases halfway (10V/division for both photos). The amount of time between the high side turning off and the low side turning on is 1us, which is supposed to be the dead time according to the 100kOhm resistor on the DT/SD pin on the driver. Does anyone know why the high side gate drive isn't turning fully off? I believe this could be causing my problems when I try an AC rectified rail. The gate drives are tied directly to the driver and no resistor or diodes in place. Would that make a difference? I have one photo zoomed in and one further out.

    Help is much appreciated.

    Alan
     
  13. RmACK

    Active Member

    Nov 23, 2007
    54
    0
    I think that may just be due to your dead time. Hard to explain.

    Remember the source of the high side is floating, and when top mosfet is on the source is at a high voltage. It may not necessarily fall to zero when the fet is off if the lower fet is also off & your output is unloaded so you may find your gate = source voltage but not = zero until lower fet turns on.

    Are you able to so a math function on your scope to subtract one trace from the other? If so put one channel on the gate, one on the source and set it to subtract source from gate andyou should be able to tell if the gate drive is ok or not.
     
  14. Tahmid

    Active Member

    Jul 2, 2008
    344
    25
    Hi,
    If the frequency is>20khz, then the mosfet drivers like ir2110 can be used safely as these highside and low side drivers are having internal modality for creating floating/virtual ground for high side mosfets. Only problem is the synchronization of triggering between highside and low side mosfet pair. Moreover in sinewave generation, long deadtime is not required, just enough but minimum gap should be maintained so that cross conduction is prevented.1usec deadtime is completely alright.
    Problem of Alan may be for the drivers whose bootstraping process is not working properly and hence the upperside mosfets are not turning off fully. This can happen maybe for the problem in code for which the drivers/mosfets are not getting proper and syncronized signal. Check again and again.
    Thanks.
     
  15. Nando37

    New Member

    Aug 15, 2008
    3
    0
    You are trying to build a DC/AC Inverter with variable frequency and variable output AC voltage

    You are trying to design a variable output UPS or a DC/AC inverter, now common in the alternative energy industry.

    I have not read all the messages, too long, but you will need to vary the duty cycle from almost 0 % ( 5% ) to about 95 % to keep the design stable, also the scheme to produce the sine wave needs to be chosen, either a sine wave reference or the equivalent.

    The output to attain a clean sine wave needs an Integrator circuit made by a choke and capacitor to filter the high frequency and to leave the low frequency AC.

    the half freqency switching using one half of the bridge the second half to produce the PWM.

    You need to make sure that the Fet driving cricuits have enough OFF time between switching to avoid cross current in the half bridge set up, do not assume that they do NOT since th FETS do have different ON/OFF times.

    The maximum voltage you can generate seems to be 170 - 40 = 130 volts peak, AC, since the choke or inductor needs to have a certain voltage to give you the sine wave under maximum current -- so you need to increase the 170 volts to 200 for 164 AC Volts peak, for 117 Volts RMS AC so for higher AC voltages, you can calculate the proper supply voltage to the AC changes.

    Now your basic problem, blowing drivers and MosFets

    Your low voltage power supply is too high, limit it to around 13 volts.

    Driver : Pin 6 connected to the Source of the upper FET, Pin 2 connected to the lower FET and their respective output lines to the GATEs of each FET with a 10 to 22 OHms 1/4 resistor.

    Make sure that the connections are short, specially the grounds of the DRIVERS.

    The charging Diode for the storage capacitors need to be HIGH VOLTAGE and HIGH SPEED, the 10 OHMS helps to reduce the peak current when the lower driver is ON to charge the 0.1 to 0.2 microfarad ( this is not your problem ).

    You Basic problems is time between the MosFets which should have the driers with a delay close to 200 nanoseconds -- THE DRIVERS will NOT guarantee such time isolation and the FET will cross short circuit and that the drivvers as well.

    You have as well another problem and it is the intrinsic diode that the FETS have and sincce you have an inductive load, when you turn one FET OFF a spike with opposite polarity which clamps the FET diode ON which is slow compared to the FET and then the other FET turns ON which needs to discharge this diode fast causing a high current spike.

    SOLUTION: In power supplies of this type two diodes are needed with each FET, a Schottky diode in series with the Drain Cathode toward the Drain and an antiparallel high voltage high speed connected between the Anode of the Schokky and the Source of the FET -- this done on each FET.

    You will need a heat sink set up since this type of design may be around 90 to 93 % efficient.

    I am a power supply designer, retired and I built this type of product from low watts to about 500 KW capacity.

    Good luck
     
Loading...