Encoders vs. Hall Effect vs. Banner Sensors

[Jeff Waegelin]

After seeing some of the ideas people gave me in this thread, I decided to post a poll. Which do you feel works best for counting wheel revolutions: shaft encoders, hall effect sensors, or the Banner optical sensors (with some kind of pattern on the wheel)? If you could post your reasons why, too, I would appreciate it. Thanks!

[coastertux]

We used encoders on our robot because they were easier to use/install/program and seemed to give better results than the standard Hall Effect sensors in the kit.

[Sparks333]

We used our own quadrature hall-effects, but we wished we hadn't. The resolution wasn't good enough, and the counts were limited by the number of magnets we could squeeze into the wheel without cancelling the net charge.

Sparks

[MattB703]

Our solution may be a little bit of overkill, but it worked very well.

We used 64 pulse Bourn Optical Ecoders geared about 6:1 to the wheel speed. We wired these into a custome circuit box that had 2 COTS chips. I can get the technical details if anyone wants them. I'm a mechanical guy so I don't know myself. I do know that the first chip took the quadrature signals and turned them into one pulsetrain at 4 times the frequency and a digital signal for forward or reverse. The 4 times frequency pulsetrain was then sent into another chip that was an digital to analog converter. We read the analog output from that chip into the RC as wheel speed. No interupts required.

Matt B.

[Rickertsen2]

if you search there have been NUMEROUS threads on this issue

[mechanicalbrain]

i like optical sensors bcuase they are easy and less likely to generate errors like the hall effects

[John Wanninger]

Has anyone checked out the VEX pamphlets lying around their local Radio Shack? According to the brochures, among the inexpensive new sensors coming out for their VEX robots wil be a 100 pulse/rev optical encoder(#276-2156) good for up to 1020 rpm, or about 1700 pulses/sec. While 100 ppr isn't great, it should be good for 1/8" resolution even on the fastest robots if geared correctly. At $20 for a pack of 2, they're certainly worth a look. They have some other interesting sensors (ultrasonic, line follower kit, etc.) as well. (No, I don't work there, but as a kid, all my paper route money ended up there! )

[Hutch]

We used rotary encoders, and they worked... They were just a PAIN. I definitely would say that you should try to avoid having to count rotations in general (linear accelerometer anyone?) but rotary encoders are probably a good bet if you need to.