PDA

View Full Version : Encoder drawings



svenakela
03-13-2004, 10:27 AM
Hi,

I found the posting below from Tom and went truly curious. Has anyone done such a thing, and could maybe help me with the drawings?

Best regards,
Sven



If you know a little electronics and are willing to invest some time you can build an encoder with arbitrary resolution for under $5.
Use an optical slotted switch (an LED/photodiode pair), one digital gate to condition the output, and a disk printed on clear acetate.
The switch and gate are available from digikey.com, mouser.com, allelectronics.com, and many others in a range of prices.
The disk can be drawn in a cad program and printed on any laser or ink-jet printer.

robotic regards,

Tom

Al_The_Man
03-13-2004, 01:24 PM
Sven, The problem is you will not be able to get very high resolution for CNC use with this method, commercial encoders use a photo etched glass disk and use an optical phenomenon called the Moire effect to be able to read graduations higher than 100 per rev.
Al

ToyMaker
03-13-2004, 05:12 PM
Sven:
Here is a simple encoder circuit that uses discrete LEDs and phototransistors at the sensor end. Two LEDs, 2 transistors, 2 ICs, and 4 resistors. Add a 5V supply (from the PC?) and you have direction and pulse count. Not too shabby.

ToyMaker
03-13-2004, 05:15 PM
And here are some LOW resolution encoder wheels.

ToyMaker
03-13-2004, 05:23 PM
Let’s crunch some numbers.
My machine has 1/4-20 all-thread lead screws. Remember that each segment on the encoder disk gives 2 pulses, 1 on the black-to-white transition, and another on the white-to-black transition, so a 50-segment encoder will produce 100 pulses per revolution.
On my 1/4-20 rod, 1 revolution is 1/20 inch = 0.050 inches. So, 0.050 / 100 = 0.0005 inches. That is a half-thou per count. Let me assure you, the mechanics of my machine will NOT hold this kind of tolerance. Even with a more practical 5 pitch screw that is only 2 thousandths per count (about 0.5mm): not unreasonable for a 4x8 foot (about 1.25x2.5m) wood carver. And, one hundred segment encoders, while not as easy as those shown, are not impossible.

robotic regards,

Tom
= = = = =
"Unlike my predecessors, I have devoted more of my life to shunting and hooting than to hunting and shooting."
- - English labor leader and government official, Sir Fred Burrows

svenakela
03-13-2004, 05:59 PM
Thanks ToyMaker, marvelous!

I agree with with your tolerance arguments, and with a little bit of a wider disc a couple of more counts should fit in. I'm also in need of a belt drive to gear stuff down, which gives even higher resolution. :)

While I'm running hot on this, what is CPR? I thought only the pulses where important...

Regards,
Sven

ToyMaker
03-13-2004, 07:15 PM
Sven:
Sorry about the CPR calculations. I'm thinking in terms of mobile robot odometry where CPR translates into distance traveled.
But, even in CNC you need CPR to calculate resolution.
In one kind of servo control system the computer sends step/direction signals to the servo-amp controller which drives the motors and then compares computer steps to encoder pulses to position the load.

robotic regards,

Tom
= = = = =
A magician pulls rabbits out of hats.
An experimental psychologist pulls habits out of rats.
- - unknown

svenakela
03-13-2004, 07:58 PM
No no, don't be sorry. ;)

I think you misunderstood my question: What is the definition of CPR? Cycles per revolution, yes. But what does it really mean?
Isn't it enough with the PPR value?

Regards,
Sven:confused

svenakela
03-13-2004, 08:54 PM
Duh, Im replying myself... ;)

If I got it right, in a quadrature encoder one cycle is one "wave" and the pulses are on the edges - four of them. So if the CPR value is 64 the PPR value must be 256.
Right, or wrong?

Sven

Konstantin
03-14-2004, 11:19 AM
Some kind of encoding happens inside every standard ball mice. X Y coordinates are encoded with 2 slotted wheels, would those encoders be useful for cnc?

How about optical mice?