You're misinterpreting those values. The pin descriptions clearly state "The pin can be driven by signal levels of 0V to 15V (no greater than VDD)". (though that sort of thing should be in the table somewhere, not a text description). Where you are getting the "2 and change" value is probably the "High Level Input Voltage" in the Electrical Specifications. That is not a limit, but the threshold of what voltage is considered HIGH when applied to the input. In other words, once there is more than 2.5V on the input pin, the chip considers that a HIGH input.
You are not going to find a single chip solution for this, and having to work in P-FETs is just going to make things harder since P-channel transistors are worse than N-channels ones.
The HIP4081A really is the kind of thing you need, since it can bootstrap N-channels to drive the high side. Don't be fooled by the 2.5A rating. That's not the bridge current, but the gate drive current spec that determines how quickly it is able to turn the FETs on and off. Because the bridge is made with discrete FETs, the rating of the FETs will determine how much current the bridge can handle.
Are those regulators just for the bridge drive circuitry? You can't be thinking of powering the bridge with a pair of linear regulators, are you? You will toast those things without massive heatsinks if you're going to be running a few amps through them.
I don't have tremendous experience actually using some of these components so I take the specs more or less at face value and I'm used to having ppl tell me the product itself is less capable than the specs on the pdf. Bottom line is I need a nominal current of about 10A pretty much for any arbitrary amount of time without burning anything up. I can heatsink the chips if they can take it. I don't mind using a single chip H-Bridge either.. makes it easier for me not to have to re-invent the wheel but honestly I haven't seen anything that has a couple of arduino inputs on it and a single output capable of 10+ amps so I bought some N and P channel fets... the best I could find and planned to make the thing manually. If there's a better option, I'm still willing to go that way but I can't compromise on the amperage or the execution style (ie. I need to use arduino pins only to take a 24V single sided supply and alternate it back and forth). As for the regulators, is there something just as easy to use (like some 220 style chip with 3 pins on it) that doesn't present a heat dissipation issue? I'd rather not substitute it with some convoluted circuit requiring a dozen components just on it's own... essentially building a power supply. The only need for it at all is to prevent a variable voltage coming from a battery that will decay with use.
EDIT:
I imagine you were thinking of a switched mode regulator... something like a DC-DC converter. The ones I've purchased in the past, capable of 15A or so were kind of big'ish and I can't say much for the quality... either of the assembled product or the cleanliness of the output waveform, but I guess it could be made to work if it was well made and small enough and properly heat-sinked.
Single chip H-bridges can't use vertical current flow and are very limited in current. However there are some which have several MOSFETs and chips in one package, such as the VNHxxxx series (such as VNH3SP30 - polulo have a variety of pcbs using these).
Simple reliable MOSFET motor drivers require a driver chip to provide protection features like shoot-through-prevention, under-voltage shutdown, etc. Without protection circuitry MOSFET bridges just blow up too frequently to be usable!
MOSFETs with on-resistances in the sub 1-milliohm range are available, you won't need heatsinking if you pick a good enough MOSFET until you are in the 30A-or-more range. However its a case of balancing the cost of the devices with the cost of the heatsink - 10 milliohms is far more easy to source than 1 milliohm, especially for higher voltages.
MOSFET drivers are designed to reduce parts count, not increase it.
Thanks for the detailed response Mark. I don't think the particular part number you picked is a perfect match BUT what about this one?
It has higher voltage that I need and still covers me for 12A.
I believe you that homemade bridges would blow often. I can just see that happening all over the place. However in theory, if you staggered your arduino input pulses or simply heat-sinked the heck out of your fets and chose fets with relatively low RDS and high slew rates, wouldn't the shoot-through problem reduce to heat management where you just need Qin=Qout to be satisfied and your'e good?
I spent a good hour looking through digikey with a dozen filters applied and the best fets I could find still had much higher RDS values than 1 or even 10 mOhm. Perhaps that's because I had other filters eliminating those results (like low gate capacitance and low switching times). Because I've become aware of heat problems resulting from gate capacitance for example delaying the transition time through the conductive zone. These fets you're talking about with the low RDS... are they also good with capacitance and switching time?
EDIT: Bit of a snag... See what I was planning on doing is using a complimentary inverted pair of outputs coming from one of the arduino timers... say pins 9 and 10, inverted from each other to drive the 2 halves of the H-Bridge. This way, when there's a trough on the left side of the bridge and the magnetic field is collapsing, I can get a peak building on the other side in perfect synchronization and vice versa. How would I achieve this sort of result using the pinout on this single chip solution? I see it has a single PWM input but is that enough to get the alternating effect I'm after? How would that work?
You don't need 1 or 10 mohm, 25 mohm will work ok, with current 10A the power on it is only 1-2 W, transistor is getting hot because of wrong switching timing = post #13 drawing, that circuit is ok for slow switching , 1Hz.
Connecting 2 gates together is bad for fast switching , there is time where both transistors are half open and that is main reason why transistors are hot.
For fast switching mosfet driver produces pulses with a dead time .....
I believe you.. but there must be a solution for transformers. The one mark suggested would work wonderfully for motors but I realized that if I can't flip both halfs of the bridge at high frequency, I can't use a device like that. So if you think an indictive coil circuit would work, do you have an example of a finished product that is plug-n-play that'll work if you connect a transformer to it and drive it with an arduino?
Shoot-through for high power bridges wreaks complete havoc as 100A + pulses of a
fraction of a microsecond wide generate massive interference and dump loads of heat
(can be many kW) into FETs and internal resistance of the batteries. Typical symptom is
exploding FETs, vaporizing PCB traces, overheating decoupling caps and wiring. Timing
is temperature dependent, so mild shoot-through can degenerate rapidly into bad
shoot-through.
Its all bad. Do not have any chance of shoot-through!
That's why you need to use proper MOSFET driver chips instead of kludging your own half-baked solution. There's always performance tradeoffs, and I wouldn't be surprised if lower resistance FETs had higher gate capacitance. That's why gate drive chips like the HIP4081A have their output rated in amps, it's to shove all that charge onto the gate as quickly as possible. That's what they're designed for.
Hold the PWM input high, and apply the complimentary outputs to INA and INB.
Jiggy: You realize I am more than happy to appease you and Mark and others in the interest of safety but unless I know how to implement the "correct" solution, I have no alternative but to kludge things together. Fortunately you've pointed out something extremely important about the inputs of that HIP chip (or the VNH5019 which seems to work the same way) that should allow me to use it. If I can do as you suggested, then I believe there's no problem. I'll double-check that what you said will work and buy one. Thank you.
In the meantime, I see no harm in prototyping my kludge unit since it'll be fully supervised and I have a fire extinguisher. Sometimes there's value in doing things the wrong way..and I have time to burn and it'll probably still work so I can gain valuable data on the rest of my circuit while I wait for the mail
also seems to have suitable voltage, current and input pin specs but it doesn't explicitly say whether I can put a 20kHz square wave on the direction pin and get the same shape on the output. It implies that a 50% duty (which is what I'd be using) would result in a net zero output. This is for a motor of course, not a transformer, but given the similarities electrically, I'm concerned that trying to use this board this way will result in nothing more than a null output.
In my case 1Hz is a tad low. Need more like 20k. But the last circuit I linked to says it can handle 100kHz. But that's on the PWM. I am hoping it will work on the direction pin and that the waveform won't get filtered to 0. That's the bigger question.
There doesn't seem to be evidence either way, as the inputs are going into a generic "Logic" block. Who knows what capabilities that chip has in terms of polling rate or whatever dictates the maximum frequency. The fact that they made a deliberate point of declaring a limit on the frequency makes me wary of calling any bluffs. The G2 on the other hand doesn't explicitly mention any problems with feeding it high frequencies so if I'm going to experiment, it might as well be with that 1 I suppose.
