[TechGuy5002]

Hi,

I've been a long time lurker, but this is my first post. I am thinking about building a DIY CNC metal working machine (not a manual mill conversion). I was think about including linear encoders to address repeatability. I though about using something like iGaging DROs and just hacking them (I assume they probably use 2 or 3 bit encoding).

As an alternative option I was thinking of using optical rotary encoders that placed on the axises put use a wheel (perhaps hard plastic) that rolls against the frame or non-moving surface. The rotary encoders would not require much space (compared to a regular linear encoder) and would not be subject to dirt with optical film or unsealed linear encoders (like the iGaging models). I though about using a spring tension mount so that a constant pressure is applied between the wheel and the non-moving surface to avoid slip. I would also use wiper brushes to clear away chips.My thoughts is that this would have at least some better repeatability over shaft mounted rotary encoders, which don't really address backlash, thermal changes, etc. It would also be easier to set up, then using a linear encoder since I don't need to provide the space to store the linear rail and I don't have to worry about matching the machine to dimensions of the linear rail since they can't be lengthened or shortened to fit the machine.

Thoughts, comments? Has anyone tried this before?

One other quick question about limit switches. what would be a recommended distance between the position of the limit switch and a hard limit? I assume there probably needs to be about a centimeter or more for the controller to recognize the limit switch has been tripped and to stop the carriage (inertia). I realize some of this is dependent on motor size the mass of the carriage and the speed of the motor at the time the limit is breached. I am just looking for a recommend spacing since I need to factor in their position.

Thanks

[wdbox]

The wheel driven rotary encoder is probably not the best solution. You are depending upon the friction between the wheel and the driving element to rotate the code wheel. In order to minimize slipage you are going to have to apply substantial load to the wheel. The encoder mounting bracket assembly can get quite complex but the biggest problem will be the side loading on the encoder shaft. Most rotary encoders are not designed to take a lot of side load. The bearings will fail in short order. Another problem is compression of the drive wheel. Any deflection from the calculated radius, unless it is constant over the entire range of travel, will affect the accuracy. Rotary encoders are great, but they are meant to be used for rotary inputs only.

If you are not happy with the standard optical linear encoders, you can get ones that work with a magnetic sensor. The magnetic units are not that expensive and are almost completely unaffected by oil and dirt. I would think that they would be a lot easier to mount than the rotary encoder that you are thinking of,

[jsheerin]

I've thought about doing the same thing for an experiment (not for long term use on a machine). What you would probably want to do is mount the wheel on a shaft with bearings on both sides, and then couple the encoder shaft to the wheel shaft so there's no side load on the encoder bearings. But that is getting pretty complicated. In addition, you'll still need encoders mounted on your motors, and although I've heard people say you can setup a feedback loop using both a rotary and linear feedback device, I don't know that it's a straightforward process. That might be the bigger hurdle.

[Al_The_Man]

Here is something along those lines DRO SOFTWARE
There are products out there for doing this, it consists of a wire cord that has a retracting mechanism similar to a steel measuring tape, and the wire runs over a grooved wheel on the encoder shaft.
Al.

[ger21]

We used to have a large woodworking saw that used a rack and pinion drive, but had an encoder with a pinion in a separate fine pitch rack.

what would be a recommended distance between the position of the limit switch and a hard limit? I assume there probably needs to be about a centimeter

Whatever distance it takes to accelerate to the speed it'll be moving, plus a safety margin. 1cm doesn't seem enough, but there are far to many variables to even make a guess.

[wdbox]

Linear encoders for feedback to the servo motor controller require a fast conversion loop or you can have major stability problems. I would think that the easy solution to your setup would be rotary encoders on the motor with a good antibacklash setup on the screws.

[CalG]

The mechanism is already designed.

Proven and well liked by many.

[TechGuy5002]

Thanks for all of the input.

Based upon your comments I think I will not proceed with this method. I did some digging and I found this magnetic encoder:
LM10 - RLS d.o.o. rotary and linear motion sensors (encoders)

This uses a magnetic non-contact track/tape. Which I believe I can cut to my dimensions instead of a fixed length rail, and has a small foot print.

[cyclestart]

We used to have a large woodworking saw that used a rack and pinion drive, but had an encoder with a pinion in a separate fine pitch rack.

A length of xml timing belt and a pulley without flanges resembles a rack and pinion. I considered this arrangement for my manual mill but ended up building something that looks more like the works of a reel-to-reel tape head. The encoder signal is fed to the software mentioned by Al_the_Man. It works fine on a hand cranked mill but I'm hesitant to recommend it for the speeds cnc routers can reach.

edit/ btw: some of the parts can be found in any scrap scanner