H-Bridge circuit, please critique/review for me

Discussion in 'General Electronics Chat' started by El3ctroded, Feb 14, 2010.

  1. El3ctroded

    Thread Starter Active Member

    Feb 4, 2008
    63
    0
    I have this H-bridge circuit. I would appreciate critiques and/or recommendations.

    I realize the Optoisolators limit my switching frequency, but is 100kHz really necessary? If I switched at, say 10kHz, the optoisolators should be fast enough as they have a typical 4us turn on/off time and max time of 18us.

    Thanks!
     
  2. peranders

    Well-Known Member

    May 21, 2007
    87
    0
    Those optocouplers are rather slow.

    Why don't you check HIP4081 or something newer. One IC does all the job.
     
  3. ifixit

    Distinguished Member

    Nov 20, 2008
    638
    108
    1. When Q1 or Q3 are switched on the source will follow the gate voltage, however when the gate is at 24V the source will only be at approximately 20V depending on current. This will result in inefficiency due to power loss in the driver.
    2. Q2 and Q4 have a similar limitation because of the Vgs(th) being 2 to 4 Volts.
    3. When re-designing, don't forget the max. gate voltage is only ±20V.
    4. The turn off time will be quite slow with a 10K pull down as the gate capacitance is typically 5700pF.
    Good Luck
     
  4. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    Check out the HCNW-3120, HCPL-3120 gate drivers.

    The way you're trying to drive your gates now, you'll wind up with a melt-down very quickly, even at low frequency.
     
  5. hgmjr

    Moderator

    Jan 28, 2005
    9,030
    214
    I agree with ifixit. Q1 and Q3 cannot be fully turned on since the gate can not be driven more positive than the source by the Vgs needed to insure that the MOSFET is turned on.

    hgmjr
     
  6. Wendy

    Moderator

    Mar 24, 2008
    20,765
    2,536
    Have you seen this thread?

    http://forum.allaboutcircuits.com/showthread.php?t=33615

    You aren't biasing your MOSFETs correctly. They are simple devices on the surface, but...

    • You can not have less that 10V between Gate and Source when they are on.
    • You can not exceed the breakdown voltage between Gate and Source (typically 20V).
    • You should not exceed the breakdown voltage in reverse also.
    • The Gate voltage is referenced to the Source, not ground, to make them switch.

    The thread I pointed you to covers this, zener diodes are often used to prevent over voltage conditions. Upper side MOSFET drivers actually generate voltage greater than Vcc to drive the MOSFETs (turn them on).
     
  7. lmartinez

    Active Member

    Mar 8, 2009
    224
    6
  8. El3ctroded

    Thread Starter Active Member

    Feb 4, 2008
    63
    0
    Thanks everyone for that information. What has been said followed with what I was thinking.

    However, the circuit I posted I built a while back and it worked, though it was only tested for about 30min. I'm wondering why it didn't blow out and the only thing I can think of is that I used sufficiently powerful FETs on a sufficiently low-current motor at a slow-enough switching speed. Would you agree?

    I had 2 model-car type motors under moderate load, drawing around 7 amps total, and the FETs I used were IRF1404 (overkill I know).

    Thanks!
     
  9. El3ctroded

    Thread Starter Active Member

    Feb 4, 2008
    63
    0
    SgtWookie: I've looked over the HCPL-3120 drivers and don't understand something:
    The datasheet says it has a wide operating Vcc range of 15-30V, which I assume I have to provide. However the example schematic I found had the Ve pin connected to the source of the FET, so do I understand correctly that I have to somehow generate 15-30V with respect to the source pin of each FET???

    If so, in an H-bridge I'd have to do this 4 times, so what would be the easiest way to accomplish that?

    Thanks!
     
  10. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    No, you don't have to do that.

    Use the HCPL-3120 for the upper MOSFETs. You could drive the lower MOSFETs directly, if they are logic level.
     
  11. El3ctroded

    Thread Starter Active Member

    Feb 4, 2008
    63
    0
    Don't have to do what? Supply the Vcc, or supply it with respect to the source pin?
     
  12. El3ctroded

    Thread Starter Active Member

    Feb 4, 2008
    63
    0
    Unless I've just missed it, I don't see a good explanation of exactly what the HCPL-3120 drivers are doing or how they work in the datasheet or in the app note I found, so I don't really understand how to use them.

    Edit: I guess what's confusing me is that if Vee is connected to the source, but the source is currently = the voltage at Vcc (which it is when the gate turns on if I use +24V_FUSE for the H-Bridge's power), how does it generate drive voltage/current to sustain Vo? Or does this work like a bootstrap circuit that that has to be switched to work properly?

    Here's what I've got so far, let me know if it's anywhere near being right... Thanks!
     
    Last edited: Feb 22, 2010
  13. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    It's a bootstrap circuit that has to be switched to work properly.

    Select a low-side logic-level MOSFET to turn ON, and keep on.
    Then do your PWM business with the high-side MOSFET to keep the boost cap charged.
     
  14. El3ctroded

    Thread Starter Active Member

    Feb 4, 2008
    63
    0
    Thanks, I've got it, and it makes sense why not to use a driver on the low-side; seems obvious now ;)

    Now, how do I go about determining how much gate charge I can drive off of one of these drivers? If I for the moment ignore the thermal limitations of the HCPL-3120, use the 1.5A typical drive current rating from the datasheet, and use a formula I found elsewhere (Igate=Qg/t_transisition), the IRF1404 Qg_max=196nC, and assume a t_transisition of 1us, I get:

    Igate=Qg/5us => 196nC/1us = 0.196A to drive one IRF1404. So this driver could drive 1.5A/0.196A = 7 IRL1404s?

    Or if I use t_transisition of 10us I get Igate=0.0196 and at this speed I could (not that I would) drive 70 IRL1404s?

    Is this right or have I done something wrong? (Again :)

    Thanks!
     
  15. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    You're not likely to get that kind of performance out of one of these drivers and seven MOSFETs. The parasitics in the circuit board would kill the performance.

    Don't tempt fate with this kind of thing. I don't know what your load is, but the more you load up the gate driver, the hotter your MOSFETs will become. Besides, no capacitor is perfect - where do you think the high-side driver gets it's power from? A cheap cap will tend to blow the lid off a high-side driver.
     
  16. El3ctroded

    Thread Starter Active Member

    Feb 4, 2008
    63
    0
    :) I knew there had to be a gotcha in there somewhere, there always is. What would a safe margin be? Or is there not a good rule of thumb here? If not how do I understand when I get out of the realm of a reliable design?

    Some of this is practical info I need for a project, which will apply to a larger project. Basically I'm experimenting with doing my own small but powerful DC motor controller for something small like a go-cart. Not too hard, the schematic I've got now should work for that with single IRF1404s per leg.

    However I would like to eventually scale it up, so doing a discrete driver stage could be in the cards unless you know a driver that can handle far more current.

    Another Question: Could I use multiple gate drivers to drive one leg of an H-bridge? For example, if I used 6 FETs on each leg of the Hbridge, and drove 3 of the fets with one driver and 3 more with an identical driver, would it work? I expect the answer to be NO, so if No, why? Is it due to timing (doesn't seem like it would be that) or minute differences in drive voltage that would cause the 2 groups of FETs to conduct slightly differently, or something I haven't considered?

    Thanks!
     
    Last edited: Feb 23, 2010
  17. El3ctroded

    Thread Starter Active Member

    Feb 4, 2008
    63
    0
    Guess I can always do some destructive testing and see how it holds up.
     
Loading...