Let's Make Robots!

IGBT motor driver/h-bridge

I'm hoping I can get some help in regards to whether or not it would be possible to make a simple H-bridge with some IGBT's and drive them directly with an arduino.  I'm looking for significantly high power IGBT's, here's one I was looking at:


Now i'm not an electronics expert, but I do understand H-bridge design at a simple level, and I realize how I could hookup 4 of these to create an h-bridge, and drive them directly from an arduino.  However, if I did do that, would it be proper?  Would the arduino supply enough voltage and/or current to open it up completly?  Also, if I have 16 of these, 4 in parallel in each section, would it be able to drive that?

Obviously i'm looking for something pretty hefty, and what I really want is something that is modular/scalable.  So that I could start with 4 total, 1 in each quadrant, and then if I realized that whatever I'm driving needs more power, I could scale it up as I need to. 

Thanks for any help


Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.

If you delve into the datasheet you'll see the spec for the 'Gate Threshold Voltage' which is basically the voltage needed to turn the IGBT on. In the case of the part you've linked, the threshold is typically 5.8V, which is too high compared to your Arduino's normal outputs.

The current needed to turn on a MOSFET or IGBT is usually very tiny, so you don't normally need to consider it. This also makes them good to use in parallel with each other.

Further down the datasheet is a nice graph showing the collector current that you can acheive based on your gate voltage, for a given temperature.

You've got 3 options here:
• Find another IGBT with the required specs, but a lower threshold voltage.
• Use the IGBTs you've already found, or ones like them, but use some interface circuitry between the Arduino and the IGBTs to drive the gate voltage high enough. You'll need a supply voltage at least as high as the required threshold voltage, either from your main supply or a voltage booster circuit.
• Switch to BJTs which require only a small voltage to turn on, but will eat up more current instead. If you really do need high power transistors this might not be a great option.

Can you provide any details on what you'll be using for a power supply, and what sort of load you're going to drive with the H-Bridge?

Oh yea, power supply is batteries, either deep cycle or I just got some new lithium ion batteries I may start using.  My battery banks are of course rated to do the high amperage i'm going for.

Thanks for your reply.  I see what you mean by the gate threshold voltage, here is probably a better IGBT:


I'm attempting to power any and/or all of my projects, over the years i've bought so many expensive speed controllers that just don't match up to their performance specifications.  I have some Roboteq controllers for example, they claim to be rated at 300A per channel but they overheat in under 30 seconds and max out at 50amps, and that's not even continous.  i'd probably call them a 30 amp continous controller, for how much they cost, that's just unaccecptable. They do good for a short amount of time, they have a microcontroller controlled thermal protection software mechanism that doesn't allow you to pull to much after the controller raises a few degrees above room temperature...

My projects generally run from 3HP to 10HP, electric motors.  The best controllers I use are golf cart controllers, alltrax, they do 300amps continous without question, but they aren't reversible (well digitally).  I do have some OSMC boards that I have yet to put to the performance test, I have high hopes for those, they claim 150amp continous, and for the price it would be a bargain, granted they don't have a case, or a driver.

Anyway, regardless of all that, I just wanted to try using some IGBT's, or perhaps power mosfets to make a simple h-bridge, and see how much power I can get.  If I can get a design that works, which allows for the IGBT's to be put in parallel, i could theoretically use the same design to scale up as much as I wanted.  IGBT's are complete overkill from a voltage perspective for me, but what difference does it make.

I really need a solid 24-48 volt 300 amp continous controller, that can be controlled digitally.  Now the IGBT's I linked above, I realize individually would come no where close to 300 amp, but what's a realistic expectation for them, they have a rating of Ic 148 amp contious.  With heatsink and fan, is 100 amp possible? less? more?  It says pulsed you can get 300 amp. 

The OSMC uses 4 IRF1405 n-channel hexfets per leg, they are 55volt, 131 amp continous drain.  The OSMC overall claims to be 150 amp continous (although i've never seen anyone really benchmark/validate this claim).  But that means they are taking the mosfets and claiming their contious rating is 29% of the contious rating on the datasheet.  Now this could be for many reason i don't understand, like power/heat disapation capabilities of the fans they use, other aspects of the mosfets.  Also, they are using mosfets where i'm talking primarily about using IGBT's, but I really don't care what I use, I thought IGBT's might be easier to drive directly off a microcontroller.

Also of course, doing it myself i'm hoping to be cheaper then the super expensive controllers. 

What does the Power-Max rating mean?  obviously people push power mosfets way beyond that rating, is that the max power if there is no active cooling? meaning no heatsink, no fan, nothing?  Basically it's the natural ability to disappate the heat, and therefore that's it's max power?


Hello again, sorry for the late reply =)

Those new IGBTs are much better suited for driving directly, but at the current you'll be loading them with it still might not be enough.
Take a look at the graph on page 3 which shows Collector Current vs Gate-to-Emitter Voltage. You'll see that a typical IGBT of that model requires ~7.5V between the Gate and Emitter to drive 100A.

The Ic MAX rating of 148A is only good at up to 25°C, which you'll fly past pretty quickly, if the air isn't already warmer that is. With a decent heatsink + fan combo 100A per IGBT would be reasonable, but you'd be better using 4 in parallel to drive a 300A load. Unfortunately you can't bet on all the IGBTs sharing the load perfectly equally, so using 4 in parallel gives you a decent safety factor.
Better to over-engineer the first circuit than having to build a second =D
This is probably also the reason for the OSMC controller having such an underrated capability - it might cost a bit more to build their controllers, but they can tolerate a wider variation in IGBT performance, and overall there will be fewer units that blow up.

The maximum power rating is how much power the IGBT can safely dissipate in terms of self heating. It doesn't directly have anything to do with the load you're driving, which is what can often confuse people.
The IGBT you linked to should have a Collector-to-Emitter voltage drop of less than 2.5V when driving 100A. The amount of power dissipated as heat by the IGBT will then be P = V*I = 2.5V * 100A = 250W, which is only half the rated max power, so you're in the clear.

Yea, i'd rather over engineer the circuit as well.  I've actually ordered quite a few more components to test with, and i've ordered mainly logic level gates, which the gate threshold is as low as 1.25volts on some of them, so 5 volts opens them up basically all the way.  

After blowing out the 2 large mosfets i've realized that heat disapation is the key, which of course I knew to an extent, but I honestly thought that if I put enough in parallel that Rds (resistance) would be low enough to have little or not heat.  I've learned that's just not going to happen, at least not without a large heatsink.  So I'm moving back to the smaller mosfets, (because of cost, the large ones are just not cost effective), and i've got some high performance fans and standard heatsinks.  One of the fans is a normal 120mm size in a puter, but it goes upto like 6000rpm or something, sounds like a jet engine spinning up, well a little bit ehhe.  I attached it to a water cooler thing and it started pushing it across my desk.

With a much smaller mosfet, and a heatsink I was able to drive a fairly large load, 80 amp for around 15 seconds.  Actually it was a cpu water cooler thing which I attached a TEC (peltier) cooler to, and then put the mosfet on, compared to every motor controller i've seen it would be considered extreme cooling.  The mosfet was actually only rated at 120 amp, so was kinda neat to see it work at all.  I actually had it do 20-30 amp continous for a few minutes.  I blew it out when the motor did 150 amps for a second, which is understandable.  For a TO-220 mosfet it's kinda interesting to see how capable it is if cooled properly, and really i probably didn't have it attached very well, so it could of done better if it was bolted.

What I think I might do is go back to the OSMC boards, and attach some of these higher performance fans, and then put some temperature probes all over it and control it with an arduino, vary the RPM as the temp increase, etc...  I've already made the box for it, and hooked up a bunch of wires, but I need to make the arduino connect up to them.

I have to many projects at the moment, so building the h-bridge from scratch may be put on hold for a bit.