Anyone have any idea on how can i make this possible? For example a 2 wheeled robot bumped to a wall, then after few second it will start reversing.

Can anyone give me some idea how can i start to do this? Thanx!

i cant give you a circuit, but i would imagine that you would be monitoring the motor current. when you hit a wall, the motor will stall, at which point it will draw more current. when you get this current spike, you.... depending on configuration, but if using a relay to control direction, you would trip the relay to reverse direction. i imageing you would have to filter out the first spike after changing direction to avoid the starting current to the motor causing it to reverse again.

these are just whats bouncing around my head, and untested, so please go easy if ive made an obvious mistake.

wow thanx! i'll try all this idea this week.

Originally Posted by Nigel Goodwin

Bumper switches - that's all most use, as it bumps a wall it presses a switch under a bumper, and that tells the circuit it's hit a wall.

i thought from the thread title that you couldnt use a switch 'Reversing a DC motor without sensor/external signal '

obviously if you can use external input then nigel is spot on!

I also thought about the 'overload' current. When the robot hit something, the motor will draw more current. I thought i can use this to make as a signal. But is there any comparator for current? Pity me now i only know how to use voltage as signal, i can use inverter for example.

not sure this is right, as im an 'experamentalist' when it comes to electronics, but i would assume that youd run the current to the motor thru a shunt, then measure the voltage drop accross it. but as i know many people on here know the theory inside and out, im sure someone could elaborate.

I have decided to use a switch to trigger an external signal.



How to design, so that the motor will reverse when received HIGH INPUT from the oscillation? I have a reason why the T_LOW must be longer than T_HIGH.

The problem with my design now is, the current_a period is too long. I want the current_a period to be the same as current_b

Your oscillator circuit seems over-complicated and also seems to have a design flaw. It relies on Q1 and Q2 not having too much gain.

It would be simpler to get rid of Q1, Q2, R2 and U2A.

Just wire R1 from input to output of U1A, with C1 connected from the input of U1A to ground.

http://www.fairchildsemi.com/an/AN/AN-118.pdf, figure 6 shows how it is done.