H Bridge Design Help

Discussion in 'The Projects Forum' started by Gumbercules, Feb 1, 2011.

  1. Gumbercules

    Thread Starter New Member

    Jan 11, 2011
    8
    0
    So I've been spending the past two weeks attempting to design and build an H Bridge for a Power Wheels Jeep. These motors generally draw 15 amps at 12V so the H bridge I need requires a bit more robustness than the general H Bridge designs I've been reading up on.

    Basically my issue is that I need some medium on which to layout all of the parts, without having to dish out the money for an expensive PCB board. I have a general perforated PCB, and while the 4 MOSFETs I'm using can be fit and soldered into the board fine, the 12 Gauge wire I'm using cannot. Does anyone have any suggestions as to how I can assemble this H Bridge at relatively low cost? Thanks!
     
  2. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    Do you have a schematic of your H-bridge? That's generally the first thing asked for around here, as otherwise things get confusing in a hurry, and we have to ask a lot more questions. A complete schematic with a full list of parts & part numbers where applicable will get you much more quickly to the information you seek.

    Besides, you really want someone else to have a look-see at what you're attempting; there are a LOT of mediocre-to-simply-awful schematics floating around the Internet. Posting it now could help prevent a bit of hard work and money spent for parts from going up in a puff of smoke.

    .png format image files are preferred; .jpg files are too "lossy", .bmp files are much too large and require external programs to view.
     
  3. Gumbercules

    Thread Starter New Member

    Jan 11, 2011
    8
    0
    Here's a link
    http://www.modularcircuits.com/pic/simple_bridge.gif

    I mean that's just the general H Bridge model I've been going off of. I'm using 4 IRL7833 MOSFETs in a T220 package, there are internal flyback diodes so I've left them out of the design. And of course I'm trying to connect it all together with 12 gauge wire. Is that what you meant by schematic?
     
  4. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    OK, but the schematic leaves off a very important section - the MOSFET gate drivers. You're going to need a high-side driver for the upper two MOSFETs, or they will very quickly burn up - or the thing just plain won't work. You will need to keep the gate at least 4.5v higher than the source terminal in order to get them to be fully turned on.

    It's good to include external diodes; otherwise the MOSFET package has to dissipate more power as heat.
    Yes, but that's just a basic schematic. Need one with more detail. What were you planning on using for the high-side gate driver?
     
  5. Gumbercules

    Thread Starter New Member

    Jan 11, 2011
    8
    0
    Ok, I'm driving the 4 N channel MOSFETs directly from the arduino I'm using, and here's why. I specifically bought logic level ones just so I could drive them at 5 volts. I'm also going to be using them at very low frequencies, so that way the lower two MOSFETs will be switched on relatively quickly, and the time that it takes the low to propagate through the motor to the source of the upper two MOSFETs should be relatively insignificant. I'll also be including a capacitor across the motor leads in order to help collapse the motor's magnetic field.

    They each can dissipate 140-72 watts, depending on temperature, and have an extremely low Rdson. So that can generally compensate for any extra current running through it, such as from the internal diodes. I will also be adding heatsinks in the future to further help dissipate heat.

    I'll even include the datasheet if that'll help
    http://pdf1.alldatasheet.com/datasheet-pdf/view/86647/IRF/IRL7833.html

    The motor I'm driving will be running at 12 volts, with 15 amps continuous and peaks of 150 amps, something I made sure the MOSFETs could handle

    I've already assembled this design before, using the following PCB
    http://www.radioshack.com/product/index.jsp?productId=2103798
    and it worked fine, the only problem was the wire I was using wasn't able to handle all of the current, and so there was an arc through the insulation to the grounded T220 heatsink.

    The problem now is that I can't come up with a better design for the H Bridge. The only two options I can think of are using a similar PCB board and drilling some of the holes bigger to accommodate the 12 Gauge wire, or just use no board at all, solder the wires directly onto the pin, and just heatshrink it all up. However, none of these seem to be very good solutions, so I was wondering if the forums had any suggestions as to how I can set up the project and assemble all of the components in a permanent manner.
     
  6. jpanhalt

    AAC Fanatic!

    Jan 18, 2008
    5,699
    907
    When the motor is running at 12V, the high-side mosfet that is on will have negligible voltage drop across it. Therefore, its source will be at almost 12V. You will need to drive the gate to 12V+gate turn-on voltage (say 5V). You cannot get 17V from an Arduino pin. Your top mosfets will not turn on.

    John
     
  7. Gumbercules

    Thread Starter New Member

    Jan 11, 2011
    8
    0
    Ah my mistake, it appears I didn't read the H Bridge secrets website carefully enough. So if I connect a transistor from a 17ish V source to the high side gates, and then trigger the transistor base via the arduino, would that then work to drive the upper half of the bridge?
     
    Last edited: Feb 2, 2011
  8. jpanhalt

    AAC Fanatic!

    Jan 18, 2008
    5,699
    907
    Please provide a schematic for what you are proposing, as requested by SgtWookie earlier. Do you have a link for the H-bridge secrets website?

    John
     
  9. John P

    AAC Fanatic!

    Oct 14, 2008
    1,634
    224
    I'll probably start some flak with this, but for hand-soldered construction I think you'll have to use much smaller wire than 12 gauge. You could probably go down to 18 gauge. It will lose a little power in the wiring, but you won't start any fires. I've looked it up--18 gauge wire has .0075 ohms per foot resistance. You don't have to make a fetish out of big wire. Just keep the wire runs short (not shorted, ha ha).

    But speaking of power loss and heat--those MOSFETs will need to be on a heat sink. Though I see the IRL7833 has .0038 ohm on-state resistance. That's about the same as 2 feet of 18 gauge wire! So not much of a problem, less than a watt per transistor.

    Surely there are circuits out there you can copy, for driving h-bridges for model robots and so forth, with low voltage and high current?
     
  10. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    That'll work OK for the lower MOSFETs (Q2, Q4), since their source terminals are connected to ground, and therefore the Vgs will be 0v or 5v. It's the upper MOSFETs (Q1, Q3) that you are going to have trouble with, as when you try to turn them on with a 5v signal that's tied to ground, the source terminal will rise to 5v-Vgs(th) (the threshold voltage) and no further. This means you'll only see about 2v across the motor, and a very hot MOSFET that is dissipating a lot of power. Your motor will not run well with only 2v supplied to it. See the attached circuit and simulation. Notice the power dissipation difference between M1 and M2; as M2 is being turned on fully, and M1 is not.

    Due to the motors' inductance, initial current through the MOSFETs will be relatively low, and will build as time goes on.

    A SMALL capacitor (<1nF) can help with slow diode turn-on times. Much larger than that, and you will wind up with large bursts of current through the MOSFETs and resulting high power dissipation when they are turning on

    If the die were mounted in a more robust package, and heat-sinked in liquid nitrogen, that would be true. However, the TO-220 package thermal resistance will limit you along with your heatsinks' ability to remove heat. The idea is to keep MOSFET turn-on and turn-off times low to minimize power dissipation, as that is when they are dissipating power as heat.

    Rdson=3.8m Ohms is certainly remarkable; at Vgs=5, you may have as much as 4.5m Ohms. Note that Rds(on) increases as the temperature of the die increases. However, the high side MOSFETs will not be turning on without a high-side driver.

    Just to keep it simple, with a high-side MOSFET on and opposite low-side MOSFET also on, with a theoretical 4m Ohm Rds(on) and a supply of 12v with a 15A load current, each MOSFET will drop E=IR = 15*.004 = 60mV across itself, resulting in P=EI = 15*.06 = 0.9 Watts power dissipation.

    When the MOSFETs are suddenly turned off, the body diodes begin to conduct; they have a Vf of 1.2V @ 30A. For simplicity's sake, let's just say 1v @ 15A, which is 15 Watts power dissipation. Therefore, you'll have ~16.7 times as much power dissipation in the MOSFETs when they turn off than when they are on.

    You'll be adding them right after you burn up your first set of MOSFETs, which will be about 10 seconds after you power it up.

    Thanks. I already downloaded the datasheet and SPICE model from IR's site.

    OK, you're going to have problems there, too.
    Read: http://www.mcmanis.com/chuck/robotics/projects/esc2/FET-power.html
    If you try running more than about 70A through a TO220 package, the legs will melt off. Read Chuck's page through several times; there's a lot to digest there - but it will give you cause to re-think your H-bridge and heat sink requirements.

    I am sorry, but I cannot classify that as "working fine"; more like "Whoops!" :eek: ;) That's OK though, even burning something up can have very strong educational value, as you usually won't want to repeat it.

    Well, you need to start with re-visiting your load requirements vs MOSFETs and thermal requirements. Chuck's page will help a great deal there.
    You also need to learn about high and low side MOSFET drivers, as without them, you won't be able to turn on those high-side MOSFETs. In order to turn those upper MOSFETs on, you must maintain Vgs at 5v. Since when they are ON, the source will be at ~12v, so the gate will need to be at 12v+5v or 17v. High-side gate drivers reference the gate voltage to the MOSFETs' drain.
     
  11. Gumbercules

    Thread Starter New Member

    Jan 11, 2011
    8
    0
    Well thank you! Your explanations were very helpful and I've read through the links which helped a lot. It looks like the wire wasn't the problem in my last design then. So it looks like I'll be dissipating a lot of power, in addition to having to calculate switch on times and other forms of math (yuck!).

    However, I'm currently operating under some time constraints, and am trying to get a working robotics base so I can test some software and microcontroller ideas. Would you recommend using relays instead for now, while I attempt to learn more about and properly design an actual working MOSFET Motor Controller?

    I feel like with relays' electromechanical nature, they'll be easier to understand, involve less math, and be less likely to release magic smoke if done incorrectly, and assuming I get properly rated ones and can wire some proper drivers for the electromagnetic coils. At the moment I can afford the extra power consumption.
     
  12. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    You can use relays, but they are going to be an "all or nothing" proposition - no PWM.
    They simply won't switch fast enough, particularly power relays. It might take 100mS or more for the contacts to "settle down"; they bounce for awhile.

    You might use a pair of heavy-duty automotive relays that are SPDT.
     
  13. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    Here is a simple H-bridge constructed from two SPDT relays and a couple of pushbuttons.
    If one button is pushed, the motor runs one direction, the other button makes it run the other direction. If neither or both buttons are pushed, the motor is braked, because both terminals are connected together.

    [​IMG]

    You will need to change the pushbutton switches to driver circuits. Such a circuit could be as simple as a 2N2222 transistor, collector connected to the low side of the relay coil, emitter grounded, with a 330 Ohm resistor from the base to the uC output pin. However, you will need to tell us what the relay coil resistance measurement is, as you'll want to make certain that such a driver will be able to handle the coil current without burning up. The above described transistorized coil driver will be good for up to around 130mA.
     
  14. Gumbercules

    Thread Starter New Member

    Jan 11, 2011
    8
    0
    OK, so I've been looking into this recently and doing the math so I can determine exactly what I need to design this H Bridge, and here's what I've determined.

    For starters, I've misread the motor data sheet. It turns out 15A is the most efficient current draw at 12V, however it is also likely to fry the motor at stall currents. However, at peak power it can draw roughly 50A. Here's one of the data sheets I've found, although there are conflicting ones, but this this is the most legitimate one that I've seen.

    [​IMG]

    Now due to this fact I've determined that relays would be far too bulky and expensive for a temporary solution, so I'm just going back to using MOSFET's again. I've read through Chuck's site several times, done the math, and done the research into parts. I've even managed to find some cheap IRF1405's like he's been using. I'm also reducing the voltage to 6V so I don't fry the motors.

    So based off of his design, since he's designing for a 200A load, I can divide that by 4, since my peak current will be 50A and most likely less, I should be able to reduce the # of MOSFET's per leg to 2 each, and throw on some machined scrap aluminum heat sinks, in addition to a power supply fan, and be able to dissipate enough heat to make that work, correct?

    I haven't had time yet to familiarize myself with spice or other sim software, and the ebay deal on the FET's ends tomorrow, so while I'm going to try and learn enough to give it a computer generated try, I just want to check and see if my math and educated guessing is sound enough to go ahead and order them.
     
  15. thatoneguy

    AAC Fanatic!

    Feb 19, 2009
    6,357
    718
    The IRL7833 is a low gate threshold MOSFET, but not a Logic Level. The difference is very small, but on the RDSon curves, Logic Level MOSFETS will show RDSon with a gate voltage of 5V, the datasheet for the IRL7833 uses a 10Vgs.

    The IRLZ44 is a Logic Gate MOSFET that also fits your needs if you end up ordering more.

    With either one, you will need a high side driver to get a 10V Vgs on the high side. There are ICs, such as the IRF6224 designed for driving N MOSFETs on the "high side" of an H-Bridge.
     
  16. Gumbercules

    Thread Starter New Member

    Jan 11, 2011
    8
    0
    Thanks thatoneguy, I was looking at digikey and all, but I can get 10 IRL1405's for $12 which is the cheapest I can get most FETs for, so it seems like the most bang for my buck. Thanks though!!!
     
  17. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    The trouble is that Ebay sellers are not authorized IRF nor Vishay distributors. There are lots of counterfeit parts being sold that simply will not perform up to the real components' specifications. So, if your H-bridge blows up, you won't know if the design was faulty, or if you simply had counterfeit parts.

    The IRL7833's have a far lower gate charge than the 1405's, and are much more modern than the IRLZ44's. The IRLZ44's are rated for Id=50a, where the IRL7833's are rated for 150a; even though you would not want to run that much current through them. The main advantage is the lower gate charge; about 10nC's lower than the IRL44's and FAR lower than the 1405's. This lets you switch them on and off much more quickly than the other MOSFETs, which means less power dissipated as heat.

    If you can't get rid of the heat fast enough, your MOSFETs will die a smoky death, and very quickly.
     
  18. John P

    AAC Fanatic!

    Oct 14, 2008
    1,634
    224
    You probably can't get this information, but I wish you could find out how much current your battery can actually supply. I would guess that it's nowhere near the maximum that you've talked about, but there's no way to be certain of that. Anyway, if you aren't doing PWM control, just using the transistors for turn-on and reverse, timing issues are unimportant and power loss won't be very large. Start switching the FETs in PWM mode, and things change a lot.
     
  19. Gumbercules

    Thread Starter New Member

    Jan 11, 2011
    8
    0
    @John P. I'm not sure at the moment how much my batter can actually supply. I'm using 4 UB670 6V 7Ah batteries in parallel at 6V. I will, however, be moving the PW jeep into my dorm room tonight, and so I should be able to test out the weight vs. current motor curves and hopefully be able to see just how much current I need to deal with.

    However, after looking at the stock PW batteries, I noticed they have 30A fuses on them so that can at least give me a ballpark answer. I also don't think I'll be doing much PWM to start off with and so that'll make the switching easier to deal with. As for buying parts from eBay, I understand that it's risky to buy questionable parts from there, but the seller has great feedback, and it's $8 cheaper per 10 MOSFETs and since I'm a poor college student, I'm going to have to take risks if I have to buy a large amount of them. However, if it turns out that I only need 10-20, I might dish out the extra $16.

    As for switching time, if I'm doing my math right, and not using PWM, the switching time should only max out at around 1 microsecond. (Gate charge of 170nC/ transistor current of 200mA = 850nS). Now I guess that is a lot, but I'm hopefully going to be using scrap Aluminum as well as CPU Heatsinks and a power supply fan in order to dissipate heat. And even though I don't have the specs, I know cpu's can give off a lot of heat and so in theory, shouldn't that be enough to cool off a constantly running FET and be able to survive the occasional switch on or off?
     
Loading...