Eight meter linear position sensor.

I've been talked into helping with a project to make carbon fibre masts. Part of the work involves winding carbon thread onto an 8 meter long mandrel. To do this requires that a bobbin holding the thread moves up and down the length of thea mandrel.

How can I accurately sense the position of something moving over a distance of 8 meters? So far my best guess is to use an ultrasonic range finder, these claim an accuracy of a couple of centimeters, but I really need a precision of a couple of mm.

Any suggestions?

I would think an encoder or a laser distance sensor would work, but I need to see what the bobbin and mandrel look like.

Could you post some pictures?

Ohmycod, are you in the uk please ?
Reason is iv been looking for someone to wind a rocket casing.
Only co i could find was in NZ and shipping was prohibitive.

To be a bit more helpful.

These are similar to the way electronic vernier calipers work.

I have seen similar on machine tools several metres long, not sure who makes them though.

They are expensive but not overly so, just a case of finding someone who makes an 8 m version.

Could try making yout own but your etch tank will need to be 8 m deep :fearful:

Here is a different type

Works like an extending rule

Could you post some pictures?

No - so far this project has gone no further than discussions in the pub!

similar to the way electronic vernier calipers

Yes, I've already googled this, but like you I'm not sure an 8 meter version would be practical.

a laser distance sensor would work

Are you able to point me in the direction of one that can be easily interfaced to an arduino?

Here is another type.

Works like an extending rule.

I wondered if you can find the right rule if you could count the divisions with an optocoupler.

8ft of toothed belt and an optical rotary encoder. Connected the same way as the Shapeoko2 motors are attached.

Or 8ft of rack and pinion with an optical rotary encoder.

The only laser range finder I know of that is with a "reasonable" price is this one HERE.
but its max range is about 2.4m and not the 8m you require.

However there is also this site here that has better distance sensors but for more money.

I would say to keep costs down, you go with an encoder.

OhMyCod:
a bobbin holding the thread moves up and down the length of the mandrel.

What makes the bobbin move? Can that also be used for position control.

Why do you need to "measure" the position of the bobbin?
I imagine the mandrel rotates and the bobbin just needs to move Xmm for every rotation of the mandrel.

...R

Of course.
Slaps forhead.

Weather you use a leadscrew or belt just use a rotary encoder on the drive motor.
If you are worried about slippage use an absolute encoder.
Many cnc machines just use steppers and dont bother with feedback.
Missed steps can cause a problem but the encoder will take care of that

Isn't this, in principle, the same as screw cutting on a pre-CNC lathe?

...R

you could do it like a coil winder have a ball screw or an R an P that runs beside the bobbin.
you would have the bobbin move at a constant speed and the carbon thread would be moving a long at a speed that allows it to wind on how you wont it to do.
use encoders so you know what speed everything is running at

Robin2:
Isn't this, in principle, the same as screw cutting on a pre-CNC lathe?

...R

Yes but in screwcutting the ratio is fixed.

Winding carbon fibre involves variable ratios. For spiralling and overlapping.

Boardburner2:
Yes but in screwcutting the ratio is fixed.
Winding carbon fibre involves variable ratios. For spiralling and overlapping.

Ahh.... so it's like a CNC lathe :slight_smile:

More seriously - when do the ratios have to change?

If each traverse of the length is at a constant spiral angle there doesn't seem to be any need to know (as in how many mm) the exact position of the bobbin. A limit switch at each end and a constant traverse speed in between should be sufficient.

...R

Continuously.

Best explained here

Runout which is sometimes seen at the ends is needed to retension the fibre when the wind direction changes by more than a few degrees.

Robin2:

Boardburner2:
Yes but in screwcutting the ratio is fixed.
Winding carbon fibre involves variable ratios. For spiralling and overlapping.

Ahh.... so it's like a CNC lathe :slight_smile:

...R

Yes but an 8 m long one which threw be a bit.

A bit more research indicates that 8 m leadscrews can be had, being the max length that the cutting machines are designed for.
Not managed to find a 16 m timing belt though.

You can buy timing belts on a roll. You don't need a complete unbroken loop. Fasten an 8m timing belt to each end, then a rotary encoder with a toothed pulley and two idler wheels to wrap the timing belt half around the toothed pulley. The encoder moves with the moving part.

Like this:

If you really can't put the rotary encoder on the moving head, then a 16m piece of timing belt can be wrapped around and the ends clamped together. The belt will never move more than a little less than 1/2 turn, so that part won't have to go around any curves. The clamped end is opposite where it attaches to the moving head, or the join can be where it attaches to the moving head. That is how many 3D printers use timing belts for the drive.

$2 per foot, sold as one continuous piece:

Many other sources.

Another way to do will be to use an I beam and two lengths of 10 x 50 x 50 angle iron you would have squire rails on the top of the I beam and a belt or R an P running along the side of the angle iron then you just need to make a carage that will contain the coil of fibre.

to change the way the fibre is layered you would just change the speed of the spindle and how fast the carage moves.

just think how threads are cut on a lathe yes they are fixed but its the same principle.

to make it easier to run it all use servos to move the spindle and carage then you just need a program them so when they go in one direction they go at a set speed stop them come back at a different speed if you need to change the angle of the laying of the fibre just have a small arm that controls the angle of the fibre.

it is the same as using a lathe to cut threads as its is set to, turns per inch or what ever you just need to work out turns per inch.

think of a bolt the threads go around at a set amount turns so it just a calculation fibre X radius the pitch would just be the fibre to fully coverage one inch side by side