View Full Version : H-Bridge Question
01-19-2008, 06:00 PM
I want to drive a two winding bipolar stepper motor with the propeller with a pair of H-bridges. I have a bunch of spare IRF740 MOSFETS (see figure attached). I want to use the MOSFETS in-place of the NPN transistors in the attached H-Bridge schematic.
Question: Since the IRF740 MOSFETs have an internal zener across the drain and source, do you think I can eliminate the 1N4004 protection diodes in the H-bridge schematic? This is my first time dealing with these parts.
Edit: In my first post I forgot to add credit for the H-Bridge.jpg circuit attached to Jason Johnson. I scraped the diagram from his excellent stepper tutorial page (circa 1998) at:
Post Edited (Drone) : 1/20/2008 5:16:52 PM GMT
01-19-2008, 08:02 PM
Those IRF470 parts aren't all that rugged for low-voltage high current stuff as they are designed for high-voltage. The biggest worry though is the Vgs(th) gate threshold which although they might quote a figure of around 3V it is only for a 250ua load!! Beware, you will need around 8V to drive the gate on the bottom transistors and +8 above the supply voltage to drive the top transistors. So that means if you are driving +12V motors then you need at least +20V for the high-side drive as the high-side needs to see +8V across it's gate and source while it's source is sitting at +12V (you do want it to switch right up to the supply don't you).
I typically do this driving with common LM358 opamps to translate the logic levels while I power the opamp from a voltage doubler made up of a 555 timer.
BTW, yes you can discard the external diodes as the internal diodes are much faster anyway than a common rectifier diode. Also as an alternative the IRF3205 are quite cheap and feature a very low RdsON of 0.008R vs 0.55R plus a lower effective Vgsth (although most mosfets are never low enough to run directly from logic level drive and handle high currents).
01-19-2008, 09:56 PM
Thanks for the great reply Peter. You are right the IRF740 isn't the ideal part for this application. I see that now. I'll try to find a spice model for the IRF3205 you recommend and see what can be done with a LTSpice sim. Thanks for clearing up the question about the internal diode.
01-19-2008, 11:50 PM
I think IGBTs are better than MOSFETS for high·current apps like this, where speed isn't so important.· I'm using some very large ones right now in a high-power H-Bridge circuit.· But, there are a wide variety of smaller ones with reasonable drive requirements.
01-20-2008, 03:37 AM
i would prefer to use N-FETs to ground and P-FETs on the +12 side!
This way you get a simple and reliable H-bridge without any bulky driver.
01-20-2008, 03:50 AM
Even though the MOSFETs have internal diodes, I would still recommend you use external diodes too. I've spent countless hours with a scope on motor controllers that I built myself, and I have observed that removing the external protection diode from the circuit causes the flyback volts to more than double even though the MOSFET has an internal diode. I believe this is due to the inductance in the wires to the MOSFET. Conventional advice tells you to place the flyback diodes near the transistor; that's bs - why wouldn't you rather have the diodes as close as possible to the source of the spike, which is the motor? So I put my flyback diodes around the motor, not the MOSFETs, and that really reduces the spike before it even gets to the MOSFET. Also, a diode costs about 7 cents, while a good MOSFET is $3 or more; why would you want to skimp on a 7 cent part to increase the heat and electrical stress on a part that costs 40 times as much?
Edit: Also, if you use an ordinary op-amp to drive the MOSFET, be aware that a MOSFET gate is not quite like a CMOS digital logic input! It has hundreds of times more gate capacitance than a digital input and can require a lot of drive current to properly drive it at higher frequencies. Op-amps won't provide enough current (at high frequencies) to drive a MOSFET, and so although the circuit will appear to work, you will burn out your MOSFETs under load. HOWEVER, if you limit yourself to low frequencies (a few hundreds to 1 or 2 thousand Hz) then you should be ok with an op-amp drive.
Post Edited (Dennis Ferron) : 1/19/2008 8:56:03 PM GMT
01-20-2008, 08:06 AM
Joerg: I would prefer to use P-FETs as well except that they are not as rugged as N-FETs. I have used P-FETs in smaller circuits where it is much easier to find devices that are matched. N-FETs are cheaper and at low switching frequencies a cheap opamp is all that is required to drive them. Besides, Drone had N-FETs on hand.
Dennis: I agree and there is merit in adding diodes at the source of interference although we are talking about running stepper motors which mostly operate at much lower frequencies. I have designed high-speed micro-stepping controllers before and I prefer that the pcb is mounted as close as possible to the motor as long leads generate EMI anyway. It's all much neater and more serviceable when everything is on the pcb sealed inside the box.
I would disagree however as regards to opamps not being able to drive mosfets at high frequencies. The humble lm358 may not be up to the task but there are a 101+ pin compatible opamps (aka adsl drivers, buffers, etc) ready to put their hands up.
01-20-2008, 06:41 PM
Only a comment to the integrated diode in FET's:
After a short view in some of the lie collections (data sheets) here some facts.
The reverse recovery time of a real fast diode is about 30ns where as the reverse diode of FET's have around 100ns!
The stored charge is about 10nC and the reverse diode of FET's have more than 100nC!
That is why i recommend extra diodes for H bridges.
01-20-2008, 10:06 PM
Thanks everyone for the worthwhile feedback. I've got lots to chew on.
01-21-2008, 06:20 PM
FYI, I've got the IRF3205 simulated in LTSpice/SwitcherCADIII (free from www.linear.com/designtools/software/ (http://www.linear.com/designtools/software/)). Attached is a .zip file containing (1) the subcircuit definition file that models the IRF3205, (2) an example circuit for LTSpice that tests Vgsth, and a text file explaining how to use 3rd party subcircuit definitions with LTSpice.