Archive for April 2010

L293DNE H-Bridge

April 19, 2010

Today I’m going to talk about the L293DNE Quadruple H-Driver used to control 2 bidirectional motors when hooked up to a motor. I had a lot of trouble recently attempting to get this thing working. All of the tutorials written involved hooking it up to a microcontroller (I’m going to use the arduino in all examples from now on), and sending singles with code to change the direction. While that is the ultimate use I’ll be using the L293DNE for, I wanted to make sure it worked without a microcontroller first.

Example use of an L293DNE

The L293DNE has 16 pins. If you’re just driving a single motor you can ignore 5 of those pins (9-11, 14-15). Pin 16 powers the logic of the controller and should always be set to +5V from what I’ve been able to gather. Pin 8 provides power for both motors. Often you’ll want a stronger and independent power source for your motors than from the rest of the logic. Your motor power source should have its positive node connected to pin 8, and ground connected to the grounds of pins 4, 5, 12, and 13. Grounds for your logic control source should also be connected to the same pins.

The motor will be connected to pins 3 and 6. The basic idea is that when power is supplied to pin 2, pin 3 will become positive and pin 6 will become ground for the motor. When power is supplied to pin 7, pin 6 will become positive and pin 3 will become ground. Depending on which control pin (2 or 7) you have powered, the polarity of the circuit that’s going through the motor will change. If neither 2 nor 7 is powered, there will be no circuit going through pin 3 and 6 and the motor will not be powered.

Motor 1 and 2 are turned on and off by pin 1 and 9 respectively. If either of these pins does not have +5V connected to them, the motor associated with that pin will also not receive power.

One thing that I did not understand from looking at the datasheet for the L293DNE was that while you can add a voltage to pin 2 or 7 to turn the motor on in either direction, the other pin must be connected to ground to work. This means that if pin 7 has a +5V but pin 2 is not connected into the circuit, the motor will not work. The circuit for pin 7 is completed by pin 2 (and visa versa), not the grounds coming from pins 4, 5, 12 or 13. I assume those grounds are associated with pins 1, 8, 9, and 16, but I haven’t experimented with that and I can’t find that information in the datasheet.

So, in order to test if your H-Bridge works, do the following:

Quick test for L293DNE without a controller

Pin 1, 2, and 8 should connect to the +5V positive line of your power source. Pins 4, 5, and 7 should connect to ground. Hook pin 3 and 6 up to a multimeter and check the voltage. It should read around +5V (probably a little less). Change the circuit so pin 2 is hooked to ground and pin 7 is hooked to +5V and check again. The voltage reading should be reversed (-5V or so). Unhook either pin 2 or 7 and you should see no voltage going through the circuit. You’ll notice the same thing if you leave the original setup and unhook pin 1 or 8 as well.

So what caught me off guard from this was that all tutorials mentioned plugging the H-Driver into an arduino and setting pin 2 to HIGH and 7 to LOW. What didn’t click for me was that LOW doesn’t mean no current is going through it, it means ground. By not making the association that LOW = Ground, I was attempting to just add voltage to 2 or 7 without connecting the other, assuming that pins 4, 5, 12, or 13 would provide the ground for the circuit. This is not the case, the other needs to be connected to ground.

I don’t know if this will help anyone else out there working on making simple bots, but I found it a frustrating adventure in learning how to use the L293DNE. If I had just hooked it up to the arduino to start with it would have worked fine, but by going the longer route of figuring out how to use the driver without an arduino I learned more about the chip and the arduino itself.

Coaster Bot

April 1, 2010

There is currently a contest by Make Magazine to create a coaster bot. I’m not looking to really win or anything, but it is a good excuse to create a robot of some sort.

Wall-E is a good example of a Robot. Also, pictures make posts more interesting I've been told.

I’ve ordered a few parts such as a pair of motors, some wheels, and a bunch of hexes of steel and aluminum for structure support. Unfortunately I don’t have much in the way of sensors so I’m not sure how I’m going to have it interact with the outside world. The only sensors I’ve ordered were a few ambient light sensors, but that doesn’t really help with navigation other than perhaps telling the robot to stop when it becomes dark (because you’re likely to be eaten by a grue).

Fortunately, sensors are really just devices that send some sort of electric signal when they “sense” something. Digital sensors will be basically an on-off sort of thing such as a sound impact sensor that detects loud sound changes like clapping. I’ll need some sort of sensor to detect other objects so the robot can perform some sort of navigation.

Luckily, there’s a simple sensor that I’m pretty sure I can build myself without buying an expensive sensor module. That is a touch sensor. The idea behind a touch sensor is to detect when the robot has basically run into something so that it can react to that in the future. I can do this by placing some pieces of material on the front of the robot that will bend slightly when the robot hits something, causing a piece of metal to complete a circuit and send a signal to the microcontroller.

If it looks like a hacked up ugly version of a large button, well it is. The idea is taken from the Lego Mindstorm kit that I played with in College. I should be able to find some scrap metal (maybe a washer or something) for the touch contact and another washer and some wire for the internals. I’m not sure what to make the touch arms out of yet, maybe some sort of wood or see if I have any plastic pieces laying around. Ideally I suppose I would use a part of a CD, but I don’t have any good tools for cutting and working with them. I’ll look into it though

Very rough diagram. The gold lines are just wires or plates that will connect to the Microcontroller.

The touch sensor design can be augmented for other uses as well. If you place a rod with a wheel attached to it on the underside of the front of the robot you can add an additional contact that will activate when that front wheel drops too low (such as when stairs or a ledge are encountered). It might not look pretty, but it won’t look pretty if the robot falls and smashes either.

So the robot itself will need to be able to navigate a space, around objects and such, in order to qualify. The touch sensor is a good start on that path, but it might be unfortunate that the robot will only be able to avoid something after having rammed into it. My cat might not like this much.

Another aspect of the contest is that the robot needs to be able to accept additional sensors to help it navigate. That seems easy enough, though it’s mostly on the programming side (the side I’m least worried about). Ideally I’ll be able to acquire some range finders, either ultrasonic sonar or maybe some sort of laser range finder.

Of which, lasers eventually need to be involved. I’m not sure how or in what capacity, but lasers are cool.

So this is the start of my CoasterBot build. I’ve ordered the motors and a few wheels along with a bunch of other construction parts. I probably should have ordered more wheels but I’ll manage I’m sure. I don’t know what’s going to come of it, but I’m sure it will be neat.