2D parts tracer

If you made a fairly lightweight arm using two suitable resolution optical encoders, there would be no stiffness at all - the encoder bearings would act as the joints.
If it were very large, it may need triangulation with a support bearing directly above the shoulder joint.

Or you could use a bearing for a joint plus a parallel arm or timing belt link so the elbow encoder was not at the joint, to reduce the moving weight even more.

How accurate do you need it to be, positionally, and over what size or area range?

What's possible or practical very much depends on the accuracy required and how much you are willing to pay for the parts.

eg. With encoders such as these, programmed for the maximum line count:

You could get resolution around 25uM with a 1m (total length) arm.

AussieBoy said:

Hi, I would like to make a 2D parts tracer. A bit like Arcroid.

ARCDROID CNC

Compact portable cnc plasma robot

arcdroidcnc.com

Only just the tracer part.
What is the best way to go about this.
I think using a robot would be difficult as it would be stiff to move the arm to trace the part.
It would have to trace coordinates, may be GCode and then convert to dwg / dxf somehow.
Thanks,

Click to expand...

In order to design this you need to define all assumptions for size in XYZ, resolution , error tolerance , input method and trace method.

e.g. Perimeter limits, OD, ID, centres, straight edges? point to point line drawing merged to nearest grid size. From a video image, flat screen or 3D model? Don't worry about how yet, just what.

rjenkinsgb said:

If you made a fairly lightweight arm using two suitable resolution optical encoders, there would be no stiffness at all - the encoder bearings would act as the joints.
If it were very large, it may need triangulation with a support bearing directly above the shoulder joint.

Or you could use a bearing for a joint plus a parallel arm or timing belt link so the elbow encoder was not at the joint, to reduce the moving weight even more.

How accurate do you need it to be, positionally, and over what size or area range?

What's possible or practical very much depends on the accuracy required and how much you are willing to pay for the parts.

eg. With encoders such as these, programmed for the maximum line count:

EIL580P-SC10.5FF.01024.B | Baumer EIL580P Series Optical Incremental Encoder, 1 → 65536 ppr, TTL/HTL Signal, Solid Type, 10mm Shaft | RS

uk.rs-online.com

You could get resolution around 25uM with a 1m (total length) arm.

Click to expand...

Hi, I know I said I wasn't interested in the cutting part. But I am intrigued.
1.
I am guessing that for the cutting mode there are motors to provide the driving force. As Encoders only return feedback for the position to be determined?
2.
With this line of the quote below.
"If you made a fairly lightweight arm using two suitable resolution optical encoders, there would be no stiffness at all - the encoder bearings would act as the joints".
Are you suggesting a robot could be built for this purpose if using the optical encoders. Only a robot would be good, to use for other tasks such as welding may be.
3.
You could get resolution around 25uM with a 1m (total length) arm.
700 x 400 mm area would be more than adequate.
4.
I guess this could all be connected to an Arduino for the control side and the data pushed out via usb.
Thanks,

​

​

Back in the late 70's we used to design a 4 or 6 layer board with colored pencils on mylar grid in less than a week. We'd send it for digitizing to gerber code to Toronto which took less than a day. They used an xy pointer to capture the start/stop end-points on a 0.1" grid then send the Gerber code files to our board shop and we'd get the boards back.

I don't know what you are digitizing in 2D but G-code is very similar to Gerber code which has different syntax and added features for trace width and drill size.

AussieBoy said:

Hi, I know I said I wasn't interested in the cutting part. But I am intrigued.
1.
I am guessing that for the cutting mode there are motors to provide the driving force. As Encoders only return feedback for the position to be determined?

Click to expand...

??
That setup is only for a precision 2D measuring arm, not a robot...

rjenkinsgb said:

??
That setup is only for a precision 2D measuring arm, not a robot...

Click to expand...

Hi, you answered that question.
Just wanted to clarify.

I'm correct in saying no motors required.

Could I buy cheaper encoders for initial testing?

Thanks,

AussieBoy said:

Could I buy cheaper encoders for initial testing?

Click to expand...

Yes, something like these from ebay should work, but they have only about 1/100th the resolution.

These magnetic encoders have more steps per rotation: https://www.ebay.com.au/itm/194625197361

Another way to "trace" the outline would be to use a camera below a translucent table to capture the silhouette, and then run a simple threshold and contour function.

So what your wanting to make is a 2D Faro arm?

Google has many links to people making DIY Faro arms. You might get some ideas from them -

From the answers above I ended up looking at these pages
Really please and hope to give it a go.
Also found these