Simple question,
I need to run a motor drawing between 7 and 10 amps at about 18 volts. (I think a similar question has come up reciently)
I am looking a these which is a big brother to the L293D. It says it will handle 2 amps per channel.
Question: Is it as simple as ganging up in 5 of these units in paralell to get my 10 amps?
No problem
Here’s a schematic and a PCB layout. Uses DPDT relays to control direction. Will handle 10A at 40V. Speed control is by PWM. I’ve added an attachment with is the Eagle CAD schematic and layout.
Postive negative BJT FET temperature
Simple answer, No. It won’t work.
Long answer. The L298 consists of bipolar junction transistors BJTs as the load carrying devices that have a negative temperature coefficient of resistance when going through current overload. One transistor gets hotter as more current is passes, dropping it’s resistance which allows more current to pass, getting hotter, etc until failure. Essentially one L298 will attempt to carry as much current as possible, til it melts, then the next, so on, down the line.
Field Effect transistor FET devices like the LMD18200, have a positive temperature coefficient, in which they have a growing resistance as they get hotter. So one FET device will get hot, causing more resistance, which will push current to other parllel FET devices, allowing the first to cool, and allowing a sort of load balancing.
Or you could buy a couple BTS7960B half h-bridges that carry 20+ A up to 24 volts. (bottom of page, need 2 for full h-bridge) These are surface mount, but not a small package.
Hah!
I never thought about that. I just realised that’s what the OSMC is: a load of FETs in parallel. I didn’t realise that was an option. Nice one.
I swear we have the smartest users…
Great explaination of the resistance/ heat thing --good stuff.
I went over to the “fried/ square one” post and found some links to some ready-made controllers… If money were no object, this guy seems the most promising… However, money IS an object so I am liking the BOA approch more and more. I would hesatate, but I re-watched the video of BOA driving his plywood plank around via remote and his little homeade controller seemed to be working fine. -I like the PWM aspect as well, as the motors I am driving are wicked fast. I plan to start going through BOA’s controller tonight and getting a parts list together.
One question though: are there any FET’s available bigger than what you are using? 10amps will probably work for me as there will be alot of gearing down and the motors will not see much load but on the other hand, they are showing 7.2 amps just spinning in mid air… I might be crossing the 10 amp threashold from time to time…
Did I get it right, BOA?
So here’s what I got (based on your schematic)
Questions:
Do I have the Reverse on/off and PWM inputs right?
Is the 2N4401 (Radio Shack) NPN Transistor the right one to use?
What's up with the 330 and 470k resistors on the bottom -do I need them?
It's a 12v relay, right?
Is there a bigger FET to replace the IRF740 to allow for greater amps?
OK, that's should be it... Gimme what you got!
Wut? We’ve been through
Wut? We’ve been through this before, ya muppet 
Actually, I mean no disrespect… I think your design is brilliant! I can’t tell you the amount of pain you could suffer, designing voltage doublers circuits (or kludging additional batteries in series) or looking for rare, expensive, and inefficient P channels FETs … The relay FET combo is a GREAT design! It utilizes the best from both worlds, simple current switching with solid state PWM.
Kudos! And of course since emulation is the highest form of complement, I would like to compliment you and use your design 
However, as my reply at the end states, I would use some flyback diodes in the design. In my setup, it means the difference between smoke and no smoke. Another scary thing is when a FET burns from a voltage spike it will burn Open. This means your bot will go 100% full blast until it hits something which can stop it - the motors will stall - then more things will smoke.
Here is my bread board testing FETs and diodes since my last simple-H lit up. (see here) - the terminal wires were ganged up for more current capacity.
I'll still probably end up using my 555 PWM and DIO for speed control, which probably won't be relevant to Chris, because he'll just use the PWM on the PicAxe. Is that your plan Chris?
Good Luck,
GroG
@BOA - uh very cool design -
@BOA - uh very cool design - I have been dealing with P and N channel mosfet bridges for a while - pumping the voltage on the high side FETs is always a pain in the ■■■■.
@Chris - consider the IRF2804 i’m using them now after struggling with underated stuff previously…
One thing I noticed though is the lack of flyback diodes from gnd to the motors mains. I have been burned by the lack of these before and FETs seem expecially sensitive. I know the FET has its own internal diode, but this won’t recirculate the large spike of voltage when the motor stops.
Have you heard of this? Does your motor have diodes in it? (if I remember correctly - they are windsheild whipers - so I would gues s not)
No diodes
Omission of these diodes was an oversight. I don’t know if I’m just getting lucky but there are three measures I can think of which may be overcoming the requirement for flyback diodes:
I agree with your thoughts on custom H-bridge design. Solid state can be a complete PITA.
at a glance…
Nope. You really do need the transistor drive to pull the FET gate all the way up to your drive voltage. These are still something of a black art to me, so I won’t advise you, but I learned that driving the gate with only 5V from the PIC caused only 5V to appear at the source regarless of drain voltage.
Also, be aware that the transistor will invert the PWM, so 100% = off and 0%= full on. AND it does not fail safe. If the MCU fails or the FET lets smoke out, the motors will keep running. (Consider an additional SPST relay at the output so if you lose one of the above, the MCU still has full authority to switch the motor off.)
Pay attention to the recent posts on flyback diodes. My design SHOULD have them and there’s a distinct possibility that I’ve just been getting lucky.
I’ll tell you something: motors that are drawing 7A at idle seem likely to pull a LOT more under load. Consider measuring the coil resistance and calculating the stall current. It seems you don’t need bigger FETs. Neighbouring posts seem to indicate that you can just bung a load of them in parallel. It looks like that’s what’s going on here.
I can’t comment on the suitability of the transistors, but I’m pretty sure basically any old NPN transistor will do so long as it can sink the current for the coil i nthe relay. Do the V=IR thing for the relay coil to find your current.
You’re nearly there, though!
future reference
Please, pls, plz, please, alsjeblieft, PLEASE.
Could someone write a half decent, tagged, summary of these discussions on powerful motor controllers? Once there is something noteworthy to write up, that is. The taxonomy (tagging) system on LMR for motor controller only provides nodes on the little stuff. Which is nice for all the starters (including moi). But I could easiliy see myself asking this kind of questions a year from now. And you guys, friendly neighbours that you are, would scramble to recover these exact threads in the forums and blogs and robot pages and the Internets.
Please let me make good use of that tag, a few robots from today. Dankjewel.
8ik
"Yes"
…is the short answer.
Well, I don’t know about someone writing a summary, but there is certainly about to be a tidy up of the “tags” system.
Once Chris has learned what he needs to learn, I’m going to encourage him to write a walkthrough. It’s not that I’m too lazy, but if I did it, it would be written in BOA-speak. There would be too much detail on the bits which interest me and not enough detail on the bits that interest everyone else. I’ve done this sort of thing before and I tend to leave out things that are obvious to me and less obvious to others. It needs to be written in the style of someone who makes no assumptions: someone who’s just figured it all out.
Chris - that’s YOU, by the way.
go chris!
YOU THE MAN!
Hi BOA,I found a book once
Hi BOA,
I found a book once on the “black art” and it explained it this way:
FET:
1. not happy unless fully on or fully off. (unhappy = heat)
2. you need a .7 voltage boost on the Gate relative to the Drain to be fully open (Happy!)
3. this is why N channel FETs are easy to deal with on the bottom end of a motor - because the voltage drop after driving the motor is much more than .7 volts - so if the gate is being driven by a rail it turns on ALL the way!
A picture’s worth a thousand words, so here
GroG
Possible added parts for a relay h-bridge
Here’s a few things that might be helpful making things work well. That IRF2804 is a pretty strong FET, that could push 75 A it looks like. Another is the old standard IRFZ44 at 49 A. There are “logic level” FETs that will mostly turn on with low 5 or even 3.3 volt signals, but what FETs really work with is about 10 to 12 volts slamming them on or off. There are devices called FET drivers, that have a mini half-h-bridge inside to quickly switch the big FET outside. Microchip makes a TC4426, TC4427, and TC4428 that drive a pair of FETs each, and can actually be used as h-bridges themselves for small motors. I would add a TC4427 to drive the 2 FETs in BOAs relay circuit. With this driver, you won’t have to worry about the signal being inverted. And add some diodes in the manner as those added to L298 devices to the outputs of the relay as had already been suggested. And also one across each relay coil as well, probably a 1N5818 or so. An MBR20150CT dual diode might be able to work for flyback on the motor with a 20 A forward current capability, 150 A peak. Maybe overkill, maybe not.
Oooooh, baby!
See, that’s BOA-speak. I coulda used that explaination a few months back…
Relay coil diodes
Actually that’s a very good point. Being an inductive load, they do “induce” at switch off. Funny how we tend to only think of motors continuing to spin after powering down as being inductive. Probably the effect is small, though?
I agree to do it!
I think you guys are right… I am probably in a unique posistion to explain to people what I have learned here -here’s what I mean: All you guru guys out there might assume one allready knows a basic concept or term as you probably do and know like the back of your hand. I, on the other hand, am an idiot and know only what I can hold in my hand and know only after it actually works. This whole “electrical theory” crap does nothing for me, just tell me how to hook it up! Therefore, I can bypass the"BOA" speak and well, speak in plain English (Well actually, now that I think about it, plain English still might not help here)
So here’s my offer. Gimme a little time to tinker with the above knowledge, and build a controller. If I get to a point that I feel comfortable letting you all know what I know, I will do a walk-through. In addition, I will see if I can piece together a bullet-point list of all the high points we have hit here as well as consolidate all the links (i.e. parts, discussions and links to ready-made controllers (there were quite a few good ones from the “Fried/ Square One” post)) I might even make a video letting you see my pretty face again.
For now, PLEASE (to the gurus) keep this discussion up and MORE SCHEMATICS!!! MORE SCHEMATICS!!!
that’s what i’m talking about
I am a patient robot builder. I can wait. 
Hi Chris,I think you have a
Hi Chris,
I think you have a good point of view. I would like to suggest a couple things.
The first is inter-related to issues I have with Drupal and this site at large. I come from a wiki environment, where everything is editable by everyone. (BOA Admin Queston) Would it be possible on this site to have a "Tip & walkthrough" which is editable by all of us (at least who have responded to this forum post)?
If that were possible - the resulting document has a better chance of being helpful to a wider audience. The quality tends to be better too, when someone tells me "GroG, your grammer stinks here", and they fix it.
The comments can contain a wealth of info - but, that leaves the reader to sift through the "irrelevant" parts.
I think we all could come up with a very good step by step doc, which contains all the parts, skematics, revisions, etc. Being a collaberative effort would make the best outcome.
If Drupal is not capable I would be willing to provide a wiki.
Regards,
GroG
