Paul and Sabrina's Cheap 3 Phase Inverter (AC Controller) with Field Oriented Control

[MPaulHolmes]

Ya, it's an optical sensor. A super bright LED sort of thing and a ton of black dots or something weird like that.

[Christ]

The ABS sensor setup is a laser with a sprocket-like rotor. I works just like a laser mouse does. 90% of the time when ABS lights come on, you remove the sensor and clean it, and it goes back out.

As the rotor spins, the laser "Sees" the high points of the rotor and the low points, and there is a signal change depending on how far the laser has to travel before it hits the sensor inside the laser housing.

I just figured that they can do the job well, since they report wheel speeds to the car's ECU in real time.

[roflwaffle]

If it's an optical sensor it could be useful, depends on how many signal tracks are there.

[Christ]

Seems like there are something like 32 around the circumference (not the same for all cars).

[dcb]

a mouse style optical sensor can detect speed and direction at a decent resolution, but if you turn off the computer and move the mouse, then turn the computer back on, the computer will not be aware that the mouse moved. Likewise an abs doesn't really care what position the tire is in, just its speed.

though I expect that two optical sensors on an output shaft where one has one less tooth could create a position signal, based on the time between pulses of the first wheel and the timing difference of individual pulse events of the first and second wheel.

Paul, you got a part number for that encoder?
Edit, sorry you already posted the link
http://usdigital.com/products/encode...rotary/kit/e6/

[roflwaffle]

As long as we could save the position of the rotor count according to mod "# of signal tracks" when off it shouldn't be an issue, unless we decide we need to turn the rotor while the sensor is off for some reason. Still, iono if 512 is suitable or overkill, so 32 may not be enough.

[dcb]

re: saving the position, you would probably have to completely disengage the motor from the wheels while the controller is off for that to be remotely feasible. Otherwise I'm sure it would drift. And even then an occasional missed interrupt or something would cause drift as well.

I think you need a sensor that says "you are here!"
Edit: or at least a sensor that can tell you closely enough where you are so you can get things moving and then compute position based on rpm and the timing of gross sensor events (a-la 3 hall effect BLDC, crankshaft "position" sensor, etc).

[Janis]

In this closed loop application there is no need for position control of shaft (these would be so called absolute encoders) , because here we need to control speed only. we ussually have used encoders with 1024 lines, less than 1000 only for high speed applications over 10000 rpm, if we need to control high precision position at low speeds we would go for 2000-5000 lines. This application should not go under 500 lines 1024 would be optimal

Edited: Actually controler will know position of shaft, because there are A and B signals for speed and direction, and index signal which is once per rev so controller will know position

[NiHaoMike]

Use an encoder wheel with a wider gap or cog in one spot. That's how most crankshaft sensors are designed. I don't think exact position is that critical for induction motors, only switched reluctance or synchronous AC. (In the case of switched reluctance, the motor itself can be used as a position sensor by measuring inductance.)

The ABS sensors (as well as camshaft and crankshaft sensors) I have seen are actually inductive. The sensor circuit applies a high frequency AC signal to the sensor and uses it as part of a voltage divider. The inductance varies depending on whether the sensing coil is over a cog or gap, which changes the impedance and the voltage division ratio.

[Janis]

For this application it is, when you use closed loop vector control, for Park and inverse transforms you need angle input. and there is said in Microchip docs not to use less than 250 lines encoder