Hi all,

I am posting this for a friend, who's been consulting me about his project. Ok, I ain't the best guy to ask, but I am the best help he's been getting (so now you know how desperate he is!)

Its a pretty interesting project: Vector control for Induction Motor. The basic requirements are: His module has to control a motor position with a certain accuracy under possibly varying load. The drive is controlled through a PWM. The immediate query is :

He needs to know if there are any PWM ICs that allow you to configure both the frequency as well as the output waveform amplitude. The one IC he knew about was the HEF4752 from Philips, but that is not manufactured anymore. Could anyone suggest a similar IC for him? It would also help if anyone could suggest one widely available in India.

My solution in absence of any such IC was to modulate the output of a normal PWM using any other configuration, or simulate a PWM with a timer and DAC (And now you know exactly how much help I really am! :-)) )

If you are interested in knowing more about the project, read on:

Right now I've only had a brief discussion so my understanding is limited, but the control algorithm looks like a heuristic thing. If anything doesnt make sense, blame it on incomplete requirements gathering. Basically the controller is given a position to reach. It then gives a PWM output, and before the next encoder input comes in, it must decide what the best next PWM output is to reach the required position in the most optimum manner. There are hard real-time constraints: encoder is sampled every 6 ms, hence 2 ms are allocated for the control algorithm and 4 ms for sending the PWM output.

The control algorithm is quite complex. There are two formulae to use while controlling the motor:
T = (S*V^2)/f
and
T = Phi*Ir
Where
T = torque
S = Slip
V = Stator voltage
f = freq
Phi = Magnetic flux density (??)
Ir = rotor current =directly proportional to V

Now, he has to come up with an algorithm which decides which of these formulae are to be used to calculate the best next PWM output . Also, he has to determine the values of all the parameters (V, f etc) that will lead to the optimum torque.

Unless somebody can give a thorough explanation of how this is normally done, I would say its a pretty heuristic thing. The guy actually asked me about the applicability of Artificial Intelligence algorithms to solve this! My gut feel was that it could be done with a lot less than AI, maybe some simple fuzzy logic. Hopefully somebody on the forum can point to something with more concrete information than me! :-))


Texas Instruments has a DSP which does exactly what is required in hardware, making things really easy if my friend could afford one, but alas $2000 per unit is a bit out of range for his budget. Since he asked ME for an alternative, I of course could only suggest an 8051 derivative <grin!>, but he is also considering PIC and Motorola microcontrollers, if there are any variants that would make his life really easy. Any help in that department also appreciated.

I did a search on this site and this link seems to be the closest to this post. http://www.8052.com/forum/read.phtml?id=8268&top=
What I read from the posts only matched with what he told me.

The part I like most is that its got several elements of interest to me... embedded systems, motor control, and quite possibly Artificial Intelligence!

If any clarifications are needed, he would be the best person to give them. He would be visiting this site himself the next time.

kundi