Step-motor accurate but physical movement not

[CTruax]

I have a fabricating machine which uses a stepper-motor-chain-feed combination to move 2x4 wood up to 24' along its axis. The stepper is pulsed for the calibrated/required movement but the actual movement accuracy is lacking.
I am looking for a measurement method/device which will give the woods location within 1/32" tolerance.
A rotating/pulse signal wheel that rolls along the top with movement doesnt work because of wood characterisics (sap/nonhomogeneous/etc).

Any suggestions?

 

[btrueblood]

CT,

Are you saying that the wood slips on the driving rollers? I.e. the motor and drive roller turns the required number of turns, but the wood is sliding on whatever is supposed to move it? Essentially, my suggestion would be to then somehow clamp the wood to the conveyor/chain, or use a pinch roller to force the wood to make good friction contact to the drive mechanism.

 

[CTruax]

The servo motor (very accurate) connects to cogged chains which grip the lumber. The slop in the chains, gears, & rollers yields errors in excess of 1/2" in 20' length. My thought is to more accurately measure the movement somehow and send the information to the servo's instructions. Remember I'm trying to get 1/32" accuracy on the movement.

 

[btrueblood]

Is the slop repeatable enough, when motion is in one direction only, to cancel it out in software? The next alternative would be to attach a displacement sensor directly to the chain, but if that's the biggest source of slop...

 

[CTruax]

The slop is not repeatable, only consistently varying. Currently the operator halts production and manually adjusts out the tolerence with a tape measure: thereby establishing the need for better automation.

 

[notnats]

Can you attach a spring loaded line to the end of the 2x4 which wraps around a drum to actuate a rotary encoder to give a more accurate position signal?

Or do you have enough power in the stepper to let you load up the chain drive by friction or counterspring to keep the chain under tension and eliminate the slop?

Or can you operate the stepper at a slower speed over the last 1" or so to sneak up on the right position?

Or can the clamping mechanism pull backwards as it operates before cutting, so thatthe slop is pulled out of the chain?

Or can the chain drive be replaced with a ball screw? A toothed belt?

Is 2x4 the only size of timber being processed? 24 ft of 2x4 is not very heavy, is the chain drive unnecessarily heavy for the duty?

Maybe the same technology that is used in optical mouses might be adapted to measuring the timber going past.

Jeff

 

[CTruax]

The machine & chain drive is heavy duty ($200,000), its just not accurate enough.

The reel idea would work, but require manual labor to place the cable after each cut, and thereby slow the process still.

The optical approach sounds interesting:
Can you get 1/32" tolerance in 24' though?

 

[notnats]

I don't know how optical mouses work, in fact I can't believe they do work.

That chain is obviously too expensive to replace. It can't have a tensioning idler fitted to it?

Jeff

 

[jraef]

I don't think the optical mouse-like sensor idea would work for the same reasons your current system isn't working, non-homogenous surface reflectivity (that is how they work by the way).

The rotary encoder on a cable reel is probably your best bet, just automate a simple process by which an "L" shaped bracket sticks up and catches the leading edge of the 2x4 and stays on it, then attach the cable sensor to that bracket. I've seen that done on what are called 'Flying Cut Off Saws" in the lumber industry. If I can remember the mane of the cable pull encoders I'll post it.

"Venditori de oleum-vipera non vigere excordis populi"

 

[Mike2468]

Have you thought of having a precision laser beam prox switch at the 24 foot mark. It would be very repeatable.

 

[CTruax]

I don't know what a "laser beam prox. switch" is.
Can a beam be shot up to 24' at the lumber end and reflected back to a measuring device?
It seems like the end cut of wood would be too variable a surface and, the 1/32" tolerance is still desireable.

 

[GStuartS]

Would a rotating/pulsed wheel made out of something like a circular saw blade/star wheel work? Might need to be cleaned once or twice a shift, but thats better that operator involvement every operation.

 

[CTruax]

A saw pulsed-blade with teeth forced into the member and hope that the non-homogeneous material / sap don't vary the accuracy.
I like the sound of that laser idea (no physical contact) better though. Anybody know if it's possible?

 

[Mike2468]

A laser prox is a prox switch that you would locate at the 24 foot mark. When you break the beam, you know that the lumber is there. Many people make them, here is one:

http://www.rad-online.com/files/f20 flyer.pdf

(Think of the cheap light-beam circuit such as on electric garage door interlocks, department store "entry" buzzers, etc.) Same idea. You just need a higher precision version.

Naturally, the end of the lumber may not be perfectly flat. So, where is the reference point? You might end up using 3 or 4 laser proxs and have a circuit such as: "when 3 of the 4 lasers sense lumber = trigger the cut-off" to avoid triggering too soon due to a wood fiber protrusion.

 

[CTruax]

Won't work.
I am trying to monitor the position constantly as it moves. It may stop & start 30 times in 24'. The whole idea is to know where it is all the time so that the requested movement can be made.

 

[BrianR]

I think that a laser distance measuring device placed so the wood is moving directly towards it on the conveyor will do what you want. eg

http://www.senix.com/

may offer a solution.

 

[CTruax]

This looks great. I'll let you know what develops from my inquiry.

 

[SERVOCAM]

Does the chain drive alway run in the same direction or does it feed the ?? 24 feet ?? and then retract?

If it does not continously feed in the same direction, then I would suggest you add an external linear scale for position feedback. I have seen some people on continous converyor systems put an encoder on the distal end to compensante for chain/belt slop.

You start out saying stepper, then swtich to servo, which is it and what brand/mfg?

Is the gripper a positive gripper - I mean does it actually clamp and hole the wood? Or does it just push the wood and not actually clamp on it? If it doesn't have a positive grip, that can cause problems.

 

[CTruax]

The chain has cogs on it that grip the wood.
The 6' chain travels through about 2.5' of flat slotted track then loops around an 8" diameter toothed gear and always runs the same direction.

 

[SERVOCAM]

>The slop in the chains, gears, & rollers yields errors in excess of 1/2" in 20' length.

When you first start out, does it make accurate moves? And drifts over time - tolerance gets worse and worse? Or is it always sloppy?

For example, say you move 2ft of the wood and process it (your cutting, drilling?). That, lets just use hole, that hole is in tolerance (+/-0.0156" or +/- 1/64). Then you make another move and you accumulate a little more error, but still in your tolerance. Over time you keep accumulating error (repeatable error) and when you hit 20 feet your off by 0.5 inch?

If this is the case, there could be a problem with your scaling in the motion controller. Maybe you do not have your conversion factor right from rotary to linear. If it is correct and you notice a repeatable accumulation in the error, adjust your scaling to compensate for the repeatable error.

>The chain has cogs on it that grip the wood.
The 6' chain travels through about 2.5' of flat slotted track then loops around an 8" diameter toothed gear and always runs the same direction.

After you advance, the gripper comes back around to grip the wood again and advance it. Is there anything holding the wood when the gripper releases? When you grab or release the 2x4, it could advance a little on you..thus maybe that is why your looking for someway to sense it.

Are you feeding multiple piecies at a time? $200K chain sound pretty expensive. Could of purchase a 30 foot actuator less than that and controller to get your desired accuracy. Must be all the wood/sap and other stuff that collects on the system that led you to a chain drive? If only feeding (1) piece at a time and your accuracy problems are in the current griper set up, maybe you could redesign the gripper assembly. Have a parallel jaw gripper that clamps to the side of the 2x4. You feed your distance, then another clamp holds the 2x4 in place, you release the actuator gripper, move back you 2.5 feet, reclamp on the 2x4 then release your other holding gripper.

Maybe more detail on your process and your actuator setup if you can share?