DIY Incremental Encoder attempt

[epineh]

Having decided to retrofit my mill with servo's I decided to be a cheapskate and try making my own encoders. This is a work in progress so may take a while.

Software I used to make the pattern is here, and not a bad read either:

http://users.adelphia.net/~skbosko/Robot_top.htm

I have servo's that have a pulse generating type of encoder, I will be removing these and adding my encoder (or a US digital one if this fails )

Here are my servo's, there is a permanent magnet in the end cover to provide a field for the windings to generate a pulse and it goes out via the second set of brushes. Second photo shows the end cap removed and the commutator assembly/brushes removed :

[epineh]

Next I have printed out the encoder wheels, I went with 2mm slots, 50mm diameter, and 1mm spacing between the two "phases".

For some reason they printed smaller than what I had intended but they will fit in the end of the servo this way, I was going to make some kind of enclosure to house them before, if I can squeeze them in the existing spot then all the better.

Russell.

[epineh]

I have used a resolution in encoder designer of 50, so this being quadrature should give me 200 CPR for my encoder... I think

This will equal my 200step/revolution steppers currently on my machine, giving me a working resolution of 0.01mm accuracy.

Tomorrow I will try to make some rings to hold the disc's in place and look for a suitable pickup to read the raw pulses.

The wheels are printed on standard transparency film, the result looks pretty good close up, I just hope it is good enough for the pickup's

Russell.

[epineh]

If all goes well with the encoders, I will be feeding these directly into EMC, and using simple H-Bridges for the servo drives, I did say I am being a cheapskate!!!

Russell.

[picbits]

I've been considering doing something like this for my cheapo mdf router build. I'm planning on using the encoder wheels and sensors out of cheap (<£0.50) mice though with PIC microcontrollers for the decoding.

It won't give me a perfectly accurate pulse per step but for general checking that I haven't missed a whole bunch of steps it should be fine

[epineh]

I was considering using a mouse encoder, I have heard that the trackball mice have high counts per revolution, though I haven't pulled any apart to find out (yet)

Another advantage of a mouse is possibly using all the hardware, mechanical and electronic to simplify things. That is plan B for this little experiment... plan C is as mentioned earlier, to purchase encoders from US digital, lol.

I would like to include microntroller control for a stepper system later on, mainly to add any missed steps as soon as they happen, or close as possible, though servo's seem just so much simpler overall, especially since EMC does the PID side of things.

Russell.

[ahren]

Cool project. I think you want your black and white sections to be the same width, however -- otherwise you will get weird effects with quadrature turned on, as transitions from light to dark will represent one distance travelled, and dark to light will represent another.

[epineh]

You may have a point ahren, I will have to play around with the settings in encoder designer to tweak things a little.

I have two options at the moment, to purchase an encoder chip, something like the Sharp GP1A71R should do the trick, and uses only one section of slits, or pull apart an old ball mouse and use the pickup's from that. I have an old mouse that looks quite easy to hack for my needs, just have to make a different, possibly larger wheel to get a decent resolution. I haven't tried one of those trackball mice to see what res their encoders have, will look into that as well.

The advantages of buying a pickup would be simplicity and not having to scrounge around second hand PC shops for the parts, downside is getting a distributor in Australia.

Advantage of using a mouse is that it is practically free !!!!! All I have to do is make the wheel, and a small PCB to hold the pickups and I'm in business...

Russell.