I was not expecting this - MOSFET H Bridge

[jocatch]

Building a new robot and decided to use 4 IRF 540 N channel MOSFETS for the H bridge instead of bipolar so I wouldn't be losing so much voltage in the transistors when switched on. I have never used MOSFETs before but I have seen lots of circuits for MOSFET H bridges and they are very easy to wire. The circuit I used is a standard 4 N channel MOSFETs driven from the output of a CMOS 4011 running at 12 volts, no resistors.

When I fired up the circuit the most voltage I can get across the motor is 8 volts from a 12 volt supply. The bottom MOSFETs appear to be switching completely on but the top 2, the ones where the drains go to Vcc, are not apparently. I just assumed the circuit would supply almost 12 v to the motor with 12 volts into the gates. This is the same circuit as in the Robot Bonnazas book. Yet they never highlighted the fact that that circuit would never supply the full 12v to the motor. So now my wheels are turning alot slower. Since I used a high RPM motor than needed, this may or may not be a problem. Still need to investigate. So I assume the IRF540s are OK, that I would have the same problem with any N channel MOSFET. Maybe if I had built the H bridge which uses P and N channel I wouldn't have this problem.

JC

[waltr]

The MOSFETs you used are ok.
Its just that the circuit is not fully turning the top MOSFETs on. Those N-ch MOSFETs start to turn on when the Gate is about 4V higher than the Source (Vgs) and fully turn on at about 7Vgs.

Look for another MOSFET H-bridge circuit that use P-ch MOSFETs on top.

[Soeren]

Hi,

[...] The circuit I used is a standard 4 N channel MOSFETs driven from the output of a CMOS 4011 running at 12 volts, no resistors.

Nothing standard about that circuit, as it will never work correctly.

When I fired up the circuit the most voltage I can get across the motor is 8 volts from a 12 volt supply.

Be happy, that's about half a Volt more than should be expected with this configuration 

The bottom MOSFETs appear to be switching completely on but the top 2, the ones where the drains go to Vcc, are not apparently.

Of course, N-ch. devices needs their gate AT LEAST 4.5V more positive than the source.
The proper device to use for the high side is a P-ch.

I just assumed the circuit would supply almost 12 v to the motor with 12 volts into the gates. This is the same circuit as in the Robot Bonnazas book. Yet they never highlighted the fact that that circuit would never supply the full 12v to the motor.

The circuit is rubbish. I am not familiar with the standards of Gordon McComb, but Myke Predko should know better!

• Using a weak "driver" like the 4011.
• No gate resistors, so the 4011 (which cannot deliver much current) will be brought to its knees (may even go into latch up) and will never be able to switch efficiently - i.e. inefficient and with high losses.
• No dead band, so same side low side and high side will be on concurrently for some time at every switch.
• Failure to realize the fact that N-ch. devices needs the gate abs. minimum 4.5V (with this device) more positive than the source.
• Use of way too slow protection diodes - but at least the MOSFETs will protect the diodes 
• 100nF buffer on the power line will do absolutely no buffering at all.
• Unleashing this on beginners that will typically take the words of a "specialist book" as the Gospel on this matter, creating disappointment and lack of faith in their abilities.

An amazing list of errors in a third edition - that alone makes the book not worth the price IMO, even if you get it as a present (your time reading it is an investment as well as paying real money for it).

So now my wheels are turning alot slower. Since I used a high RPM motor than needed, this may or may not be a problem. Still need to investigate. So I assume the IRF540s are OK, that I would have the same problem with any N channel MOSFET. Maybe if I had built the H bridge which uses P and N channel I wouldn't have this problem.

Exactly (apart from the weak driver and the missing logic (to create a dead-band keeping same side L/R devices from conducting concurrently, shorting the supply).

You have two ways to get the full 12V.
1 - Use P-ch. devices on the high side (and invert the logic to keep the switching OK.
2 - Use a driver that outputs 17V..25V with the high side N-ch. devices.
Remember to use ultrafast diodes, whatever way you go.

Better still, get a complete driver...
What are the metrics  of your motor(s)?
Voltage, current, RPM (or Kv), Locked Rotor current etc.

[jocatch]

Soeren, first let me say I have been following alot of post here of yours and I think it is great how you try to help everyone on various issues.

Regarding my motors, they are 80 RPM, 12 v (Pololu 1107) have a free running current of about 300 ma and a stall current of 5 amps. I couldn't find any ICs that could handle that cheaply.

And just in case everyone is not fimilar with the Robert Bonanz's ciruit, here it is:

http://www.joecool.org/hbridge.jpg

You are right about the book. I feel it is money wasted. I have the orignal revision and the third addition is almost the same, with the same errors such as this one. I question rather the author every actually built any of the stuff in it. By far the info in on this website is 1000 times better than the book. If someone could take everything on this site regarding building robots and turn it into a book, that would be awesome but probably too big to print! 8-)


My next challenge is to find an H bridge using N and P channel and then figuring out to retrofit my circuit to use them. Thanks again for your help.

JC

P.S. Looking some more I think to convert to P channel, I just have to replace the top two MOSFETs, switch the D and S around and connect the gates of the FETS on the same side together. I just need to get the complement of the IRF540s....

[waltr]

JC, read the details on Dead Band in Soeren's post again then google for more details. Do not tie the gates of the N & P ch MOSFETS together if you are trying to switch large currents.

Here is a link to a problem H-bridge and some discussion:
http://www.electronicspoint.com/h-bridge-popping-t219161.html
http://www.electro-tech-online.com/electronic-projects-design-ideas-reviews/40549-h-bridge-pwm-current-limit.html


Do a search on Soeren's posts as I believe he has posted MOSFET schematics in the past or at least a link to a circuit that works. There are properly designed H-bridge circuits out there but not many. Post a link and question here before building one.
It might be easier to use an L298 Dual H-bridge driver chip and add external current sense and limiting.

[jocatch]

Thanks waltr, I forgot to take into account the deadband. But now that you mention I do remember when I was bench testing the circuit that even after I removed the gate voltage, the motor still ran for a time. Didn't think the gate capacitance was so important. Like I said b4, I have always dealt with bipolar, not MOSFETS. One new circuit I was looking at using P and N channels had the gates tied together. I guess another bad H bridge design. I'll do a search on soeren and see if I can find his schematics. Otherwise maybe I should scrap the whole design and look into using an IC if I can figure out how to handle the stall current of 6 amps.

JC

[waltr]

I did see in the L298 data sheet that two L298's can be paralleled to double the current limit. That would be an easier solution.

Building and getting an H-bridge working properly will teach you a lot about electronics. So you may wish to pursue this just to learn.

Good luck and have fun

[jocatch]

I have been doing some searches on the net on how to avoid having both top and bottom MOSFETs on at the same time shorting out the power supply. I have seen ideas about increasing the gate current to reduce the time constant, etc but nothing practical.

So I started looking at motor drivers and I see Pololu has a Dual VNH3SP30 Motor Driver that can handle more power than what my motors need and it is about 1/5 the size of my H bridge. It would make life alot easier to just slap one of those puppies in then redoing my h bridge and having to buy more FETs and the like. I also did a search on this forum but didn't see any circuits on avoiding the problem with shorting out the supply.

JC

[Ro-Bot-X]

You may want to look into this schematic: http://letsmakerobots.com/node/18633
that was developed to make this product: http://letsmakerobots.com/node/21598