Go Down

Topic: Another custom h-bridge problem... (Read 5080 times) previous topic - next topic

asrinivasan

Hey there,

I'm in the process of designing a mosfet-based h-bridge - I'm sure you've seen the grief others have had with the process, but this is the only way for me to go. I have 12 high-power motors to control (stall currents about ~8 amps each) and purchasing commercial controllers would be prohibitively expensive.

That out of the way, let me explain my problem - right now, I'm using a a pair of TC4427 mosfet drivers, and each is used to drive one side of the bridge. Here is the relevant schematic:

[img=http://img850.imageshack.us/img850/6099/bridge1.th.png]

(disregard the apparent pinout of the 4427's on the schematic - I couldn't find the Eagle part so I just used a random 8-SOIC part)

Here are the mosfet specs:
(n-ch) IRLZ44n , logic level gate, 47A cont. current - digikey link: http://search.digikey.com/us/en/products/IRLZ44NPBF/IRLZ44NPBF-ND/811808
(p-ch) NTD25P03 , logic level gate, 25A cont. current - digikey link: http://search.digikey.com/us/en/products/NTD25P03LT4G/NTD25P03LT4GOSCT-ND/1967261

As you can see from the schematic, the TC4427's are connected to 12v - and therefore are triggering the FET gates at much higher than "logic level." I quickly realized this when the bridge didn't work, so I then connected VDD of the TC4427 to +5v from the arduino instead. The new schematic:

[img=http://img214.imageshack.us/img214/3203/bridge2.th.png]

Yet this still didn't work - the motor simply stayed at a lower-power on state the entire time, and the h-bridge began getting hot. Could you help me figure out the problem? I've already ordered significant amounts of the logic-level FETs, so I'll have to use them. Would  not using the driver IC at all and simply driving straight from the Arduino pins make more sense (as in the case of the LawnBot [ftp://ftp.rediculouslygoodlooking.com/arduino/LawnBot400/Triple8_v3/])? Your help is appreciated!

RuggedCircuits

Triggering your FET's at higher than 5V -- the higher the better (lower on-resistance), unless you exceed the maximum gate-to-source voltage of your FET's. The ones you have chosen are good for 15V of gate-to-source voltage so there's no reason to lower 12V down to 5V.

In fact, with P-channel FET's in your H-bridge you HAVE to drive the gate to the same voltage as the motor voltage: 12V. Go back to 12V.

The transistors on the right hand side of your schematic are "upside down". They should look the same as the ones on the left, just mirrored. Take a look at the way the built-in body diodes of the MOSFET's are pointing. In your schematic, they would allow current to flow even if the FET is off!

Driving directly from the Arduino would not make sense at all. Stick with the TC4427.

--
The QuadRAM shield: add 512 kilobytes of external RAM to your Arduino Mega/Mega2560

asrinivasan

Thanks for your quick reply! Is this what you meant by fixing the orientation of the FETs?

[img=http://img440.imageshack.us/img440/1614/bridge3.th.png]

Thanks again!

RuggedCircuits

Yes, that looks better.

--
The Basic Motor Driver: simple, inexpensive motor driver for 1 stepper motor or 2 DC motors

asrinivasan

Hi,

I tried out your suggestions - but this time, the motor didn't move at all and the power terminals sparked when I connected them (which I'm assuming means a short somewhere).
This is the current schematic I'm using:
[img=http://img440.imageshack.us/img440/1614/bridge3.th.png]
I've also attached eagle .brd and .sch files in case they are of any help.

Is there something I'm doing wrong here? I've searched for details of what others have done to make these work - one thing I kept noticing is that many people said that the N-ch FETs needed to be driven at a higher voltage, and something about "bootstrap circuits" - like here: http://arduino.cc/forum/index.php/topic,75620.msg570711.html#msg570711

Just to be clear, these are the specs:
arduino - off of 5v usb
Motor supply - 12v
FETs- both are logic level(5v gate "on")
Using TC4427 FET drivers to do the switching

MarkT

Quote from:  link=topic=94490.msg709511#msg709511 date=1330576002

Triggering your FET's at higher than 5V -- the higher the better (lower on-resistance), unless you exceed the maximum gate-to-source voltage of your FET's. The ones you have chosen are good for 15V of gate-to-source voltage so there's no reason to lower 12V down to 5V.



Given the voltage spikes that fly around a high power MOSFET H-bridge, running 12V into a gate whose abs-max rating is 15V is asking for blown FETs.  Drive logic level FETS at 5V if you can (and all gates really need protection zeners if you are handling significant POWER...)

In fact expect to spend more components on protection than the actual circuit.
[ I will NOT respond to personal messages, I WILL delete them, use the forum please ]

asrinivasan

Thanks for your reply - but since I'm not able to get this work with the TC4427's and I've seen just about a hundred different opinions about triggering FET gates, I've decided to stick one that I've seen to work  - the "Lawnbot" controller - http://www.rediculouslygoodlooking.com/site/lawnmower.shtml

I thought this would work for me since I too am using the logic-level type FETs.

Could anyone knowledgeable about this driver tell me whether the schematic would work for just 1 FET for every quadrant? He uses 3 FETs per quadrant, for higher-current capabilities, but my current needs are quite a bit lower than the motors he uses.

Thanks

asrinivasan

This is the current schematic I'm using - it is based on the LawnBot's "Triple-8" controller, but uses only one FET per quadrant. Could anybody take a look at this and tell me if something's not connected properly? Because this design doesn't seem to work either... of course it could be that I burned out my FETs long ago but I always disconnected them before they overheated, and I want to make sure it isn't something to do with the design.

zoomkat

Quote
of course it could be that I burned out my FETs long ago but I always disconnected them before they overheated, and I want to make sure it isn't something to do with the design.


Seems like time to individually test the MOSFETs for damage.
Google forum search: Use Google Search box in upper right side of this page.
Why I like my 2005 Rio Yellow Honda S2000  https://www.youtube.com/watch?v=pWjMvrkUqX0

asrinivasan

#9
Mar 04, 2012, 05:58 pm Last Edit: Mar 04, 2012, 08:27 pm by asrinivasan Reason: 1
FINALLY I found a design that works - I was just about to hold up the white flag and go the Sparkfun route (Did I mention earlier I'm controlling 12 windshield-wiper motors for a nearly life-size biped? ) since I'm only one month from my deadline. As you can imagine, I would have spent in excess of $400 on motor controllers alone - the motors themselves were cheaper. Anyway, I found this design that actually works with the transistors I'm using (look at earlier posts for specs) . One thing I noticed that's different is that this schematic tells you to trigger each side separately, rather than do the cross-configuration I see in so many other designs. This is the same for BJT-based bridges, so it does make sense... anyway, I'll post the link here for those who are as frustrated as I was - http://www.cadvision.com/blanchas/hexfet/h-bridge2.htm

This isn't 'smokeless' , and one thing I noticed is that one direction drove the motor at a lower speed than the other - maybe lower the resistor values of the high-side P-FETs?

Thanks for all who responded; solved.

[edit]
Here's the eagle sch for quick and dirty prototyping - As you can see, I'm trying to keep parts count and complexity low (since I have about a day to etch and solder all of them). Following up my previous remark about the motor not moving as fast in the other direction, I think there's some shoot-through happening; I'm willing to accept it as-is since it only needs to work properly for a day or so (for my demonstration) but if anyone else has suggestions, they're always welcome.

zoomkat

I find the h-bridge design in the below openservo project to be interesting. Also bottom is an h-bridge design that appears to use boot strapped NPN mosfets on the high side. You would have to purchase one of the bottom h-bridges to get the schematics, but if one needs 10 h-bridges, then DIY cloning might be cost effective.

http://www.openservo.com/Schematic

http://secure.oatleyelectronics.com//product_info.php?products_id=206&osCsid=30093449b694ae247d94c2bcce5d0560
Google forum search: Use Google Search box in upper right side of this page.
Why I like my 2005 Rio Yellow Honda S2000  https://www.youtube.com/watch?v=pWjMvrkUqX0

dc42

When building high power H-bridges, it's easier to use an H-bridge driver IC. These are generally designed for driving 4 N-channel mosfets, avoiding the need to use P-channel mosfets (with their higher Rds(on)). Examples include http://www.farnell.com/datasheets/32590.pdf and http://www.irf.ru/pdf/irs2453d.pdf.
Formal verification of safety-critical software, software development, and electronic design and prototyping. See http://www.eschertech.com. Please do not ask for unpaid help via PM, use the forum.

Go Up