Let's Make Robots!

Big Motor Driver Done

Run 2 Motors, Fwd and Rev with PWM Control


Well, it seems I might have smoked something. I am in the process of testing the motor driver, with a load, and using 2 motors. While my "B" channel seems to be working fine, the "A" channel seems to be stuck on. Not to mention, the FET driver sorta burned my finger when it was touched. All this and I have yet to blow my 20 AMP safety fuse! I'm letting it cool off a little and am going to try again. ---I am not looking forward to asking for "purchase authorization" from the wife for more parts!



What started as a simple question about running big motors has turned into 2 threads and over 110 posts.

First Post

Follow-up Post

What has come of this is wonderful submissions from BOA, GroG, Robologist, Krumlink with Rik and others filling in a bunch of gaps and disputing what everyone else has said. I have gleaned what I could from this firehose of posts and decided to go with the BaseOverApex/ Robologist plan. As an overview this is a high power motor driver with logic level inputs for PWM and "reverse", a FET driver chip and FET's for main power switching and 2 high power relays for reverse. Here is the original schematic from the second follow-up post "H Bridge Matrix":

BOA's Brilliant Hybrid H Bridge - Robologist Mod

BOA's Brilliant Hybrid H Bridge - Robologist Mod
Description BaseOverApex's design of a great Hybrid H-bridge. The hybrid is a combination of relays and MOSFETs. The relays are for forward and reverse switching. The MOSFETS can accept a high frequency PWM for speed control. This design has been built (not just theory) and is currently powering one of BOA's great bots. Hopefully he will post a version of the PCB art - although it might be good to rework it so that the PIC is not part of the design, as others might be using different methods of control.
Original Author BaseOverApex robologist
Built By  
Max Current 15 Amps - limit by relay
Max Voltage 12 Volts to 15 Volts
Build Time 3 Days?
Pros strong, low cost
Cons not quicklyswitchable forward to reverse, limit by relay
Max PWM Frequency  
Features reverse flyback diodes, MOSFET driver
Parts List
Desig  Qty  Part#           Each  Total    Dist      Description
*K1,K2 2 PB897-ND $4.44 $8.88 Digikey Tyco PCLH-202D1SP,000 12VDC/75 mA coil, 15A relay
*Q1,Q2 2 IRFZ44NPBF-ND $1.89 $3.78 Digikey IRF IRFZ44NPBF 49A 55v N-chan FET TO-220
*Q3,Q4 2 P2N2222AGOS-ND 0.36 0.72 Digikey On Semi P2N2222AG 600mA 40v NPN GP BJT TO-92
*D1-D8 8 497-2753-5-ND $1.09 $8.72 Digikey STMicro STPS1545D 15A 45v Schottky rectifier diode
*U1 1 TC4427CPA-ND $1.05 $1.05 Digikey Microchip TC4427CPA 1.5A MOSFET driver
*C1 1 4035PHCT-ND 0.81 0.81 Digikey Vishay 220 uF 35 v Electrolytic Cap
* Total $23.96
Schematic Image:M_relayhp.jpg
BreadBoard Image:M relayhp B.jpg
Gerber Image:M relayhp gerber.jpg


I bought all the parts on the list above, all from Digikey, with the addition of some blank 2oz copper PCB boards and were quite happy when my parts arrived within 3 days!


Next, using PCB123 layout software (which is not only free, but worked quite well) transposed the schematic into a PCB layout. In general, I was able to keep most + runs to the outside and - to the inside. I think I did pretty well, only needing 6 topside jumpers. I must admit, it was a nightmare to keep all the componant's polarity straight as I went as things are mirrored or double mirrored (depending how you think about it) in terms of the top side of the board, bottom side and the fact the transfer is reversed when applied to the board. The picture below is of the final transfer I used.


This layout, printed with a printer using ink, was taken to my local copy-shop and Zeroxed as dark as it would go using a standard copier using toner. Next, I ironed the ever-living-shit of the transfer onto my blank PCB board. I was careful to go over each line one-by-one in addition to placing the iron over the whole transfer and putting my whole weight on it.

That would be:

140 Lbs

63.5 Kg

10 Stone

Happy that indeed the shit had been ironed out of it, I soaked-the-crap-out-of-it and then rubbed off the paper.



With a little scotch-brite (green scratchy pad (for dishes)) I removed the toner from the circuit lines, drilled and soldered away. The tiny little drill bit came from a model airplane shop, chucked-up in a dremel tool and was done by hand. Also, when setting the parts I seemed to run into a big phatty problem with the layout of the diodes. If you run into the same problem, well, here you go.


As you can see, I did have a few issues with some lines bumping each other but a cut-off wheel in a dremel and a very steady hand, I was able to tidy up everything.


And here she is!

Here you can see all the parts, the 2 green LED's are tied to the PWM input while the red LED's light with reverse.

Once again, much love and thanks to all who helped with this project. At the time of this post (and the video included) is of my first -low current- test. I hope to have some new videos up soon including this unit running the 2 DeWalt drill motors it was built for!


Comment viewing options

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

Hey Grog,

The resistor diode combo that you see on the gates of the OSMC FETs were for a couple reasons. The resistors were to limit current from the driver, to keep the heat down, and the diodes were there to hasn't shut down of the FET, to prevent shoot through. There is not upper FET here to have shoot through, so no diode on the gate is needed.  A resistor of some sort might be helpfull to limit the current of the driver a little.

The TVS devices are really more useful in the OSMC, as the IRFZ44 has a 55 volt reverse breakdown Vds compared to the voltage supply of 12 volts, where the OSMC was made for a supply from 12 to 50 volts operation, much closer to the 55 volt Vbrdss of those FETs used there.

Hi robologist,

Interesting points.. I was surfing around trying to find what might make the bridge full reverse shock proof .  I have mostly seen protection diodes wired to the drain of the FET as you mentioned previously.   If the FET is clamped by an adjacent diodes, are all the other diodes necessary?  I think we are all interested in the Most Robust / the Fewest Components sweet spot.

I see the additional reverse protection diodes as useful if the FET reverse breakdown voltage is less than twice the supply voltage, otherwise I'd just use the FETs intrinsic diodes. As above, the OSMC if running a 36 volt pack or even higher up to 50v, is dangerously near that 55 volt reverse breakdown of the IRF1405s. If it was never to run more than 18 or so volt pack, the external TVS diodes probably wouldn't be needed.

This is getting more towards opinion, as I can't exactly spell out why that point is. I really prefer to provide a factual back-up as to why I picked the "double" figure, but can't lay my hands on it right now.  

Ya might be thinking of the wrong candidtate, don't you know it's the Obama-nation of desolation that brings on the apocalypse?
"George Dubya." That's all I have to say. Whetever happens, the wrong guy will win. Let's debate this elsewhere.
Guess smiley faces actually could be helpful in indicating "jokes".

I'd missed the Update above initially, but saw the shoutbox and messages that something went wrong. Kinda sounds like a FET quit, and did it while being locked on. Grog mentioned here that FETs can fail to being on, when hit with a voltage spike. I'd mentioned here the connection for the flyback diodes to use the FET drain as ground, but failed to implement that in the circuit by following standard flyback connections. Don't think that is really a problem, but might be something to think over.

FET input protections to prevent the driver from over volting the gate might be to add a zener diode limiting the voltage to 10 or 12 volts. Additionally, a resistor can be placed between the driver and the gate to prevent "ringing" between the 2, though that probably isn't happening here. The over-volting might be happening if the motors are spiking the voltage enough for it to feed through the drivers and be transmitted to the gate, though it might be expected for the driver to fail before sending the voltage. Something like the OnSemi 1N5349BRLG would be a good zener, connected anode to ground, cathode (bar end) between the FET gate and driver connection. Digikey part 1N5349BRLGOSCT-ND 12v 5w zener, (sold in q10) The resistor between the driver and the FET gate might be a 10 ohm 1 watt type, Vishay NFR0100001009JR500, Digikey PPC10CCT-ND . Either or both of these might help the driver from getting too hot, though it is probably only getting hot being forced to drive a failed FET.

It is also very possible the FET died from getting static zapped on installation, which might allow it to operate for a bit, then fail. FETs can be very static sensitive, as other devices that use them can be but typically have some sort of protection. Sadly it's difficult to say what caused failure, without hooking up an oscilloscope to watch how the device is switching.  Might be as simple as dropping to a 15 A fuse too, to give some headroom and adding a small heatsink to each FET.

hey ctc,

 While doing a search for "and" gate usage, I ran into this site that had a particularly interesting hbridge setup.


 the backlink where I found it was from here:




That h-bridge has 10k resistors that would slowly bleed off the charge from the FET gates, causing a bit of a slow turn off. That wouldn't allow a very high PWM frequency.

Assuming there's no fault in the board it's possible the fet was damaged by static or too much heat when soldering. I've had this happen once before with BoozeBot and found that a circuit breaker is a worthwile investment since you don't have to buy another fuse every time an experiment goes haywire.

PS. I don't see any heatsinks on your fets. Even if you don't need them under normal conditions they help a fet to survive surge currents.