Suspend those pesky physics laws!

The flaw of your design is that you do not control acceleration and deceleration, which not only is brutal, but prohibits precision control such as you describe. Curing an excess deceleration problem by braking and adding more deceleration is not a sound approach. You are controlling position mechanically, and perhaps the second derivative of velocity via air pressure, but not the second derivative. You are adding excess energy that is mechanical noise showing up as precision error.

Figure out a way to decelerate evenly at the end of travel, so your mechanism comes to a controlled, smooth stop. You will get better precision, you will use less energy, and your mechanism will wear slower and last longer.

The obvious method would be to use a servo and program in the ramps. Too expensive at this point, maybe in the future.

I have considered a hydraulic damper near the end of stroke to bleed off the energy and not abruptly shock the system. But you see why if I unclamp the wire, because it has so little mass it's momentum is negligible. The trick I see is unclamping it at the right time so it coasts to a stop at the right place against the steel block at the far end.

So the system is over constrained, with the wire both being held by the roller and by the steel block? Ooh, bad you.

If the wire's momentum is negligible, can you adjust the tension on the clamping gears so that as soon as it hits the block it stops? Or do the gears fly back a bit after they stop at full extent?

I think a shock absorber on the cylinder (not 'damper', not for this, in the US at least -- believe me, I've been living the terminology for the last few months on That Damned Gate Project) would work quite well, particularly if you could arrange for it to have a bit of a progressive action. You can get little screw-in shock absorbers that are spring loaded, damp when something lands on them, then return to 'out' when you retract whatever it is that is pushing on the thing.

I'll bet you could select one of those that would work quite well, particularly if you put a spring on the end of it so that deceleration force is progressive rather than getting applied suddenly.

Alternately you could valve the cylinder so that as it gets close to full extent the feed gets choked off -- I couldn't tell without doing the math, but it seems like there'd be a 'right' combination of cut off position and residual valve opening that would start bleeding off the speed while still providing enough flow to complete the stroke at an appropriately slow rate. In theory this would act exactly like the shock absorber I'm suggesting, just using the cylinder as the damper and the air as the working fluid.

Can you just slow it down a little by restricting the airflow into the cylinders?

----- Regards, Carl Ijames

Tom

If I understand your application correctly, this just screams for a servo motor.

I used to work with Parker Compumotor products. They have unbelievably top notch customer service. I'd at least give them a call. Programming one of these compumotor units is very simple.

Karl

If these are 3" cuts, separate the operations. Cut, pieces fall into sorting/stacking bin, a different machine feeds the cuts to your existing machine.

technomaNge

Both cylinders have speed controls on both ends. On the bench, at slow speed, there is no over-run. But I have a 300ms window to feed.

I figure about $2500-$3,000. The guys at Carlson Tool (they make brush machines) do a lot of feeds with servos and mechanical methods and they have servo set-ups that could work. You'd be surprized how beefy it has to be. My original design uses a servo, but why stop there...the "X" and "Y" axis on the table might as well be servo driven too. That works great for brush machines that change set-ups often but this machine never will, so cam actuation works just fine and is oh so simple.

I don't mind spending the money but I'll explore the mechanical first.

That's how it's done with round wire. With this flat wire, you can never lose control of it.

Hmmm, the wire needs to be in place and in control to be cut in the 300ms window, that's all. This is the third wire feeder design in ten years. We've had to use different wire manufacturers over the years as some plants have closed and the qualities of the wire change...this or that feed doesn't work on this or that wire too well...etc. This feeder is trying to address the wire that has too much oil, too much scale, lead drag, ect. The wire manufacturer can't fine-tune their quality any better and they are the only domestic supplier. We also use a Taiwanese supplier but their QC and lead-time is worse. Those are the only suppliers in the entire world! And, there are only 4 customers in the world and we are by far the largest user.

Need something outside the box?

The stops and starts are killing ya. So, don't stop. Put your cutoff mechanism and clamp on a turntable. Spindle diameter determines length. You'll probably need multiple stations to get the layout to work, say do four stations and use a twelve inch circumference spindle. You can use cams, air or electric solenoids to move devices.

I used to work for a company that made a million rolls of adhesive tape a day. We had a few of this idea machine for cut to length tape. Of course, the first machine we made this way was a total POS. We rebuilt it and it was only crappy. Then we built one from scratch and it would run 24 hours straight without a stop. Quite a few more operations than you have. The turntable was eight feet in diameter.

Karl

On Thu, 4 Mar 2010 14:31:33 -0500, the infamous "Buerste" scrawled the following:

What kind? Mangrove? Orange grove?

That doesn't sound quite like "perfectly" to me.

Are you sure about that? I've seen some pretty clever mechanical means for feeding and orienting small parts. I'm guessing there's something proprietary about the process, but if you could reveal a bit more detail, that might be helpful.

One thought I had would be to feed through rollers with one way clutches to prevent the wire from slipping backward. Then feed through a set of jaws that work like those on strapping tape tension tools. These jaws would be on a linear slide, to be reciprocated by a crank. The mechanics of the crank will provide deceleration at both ends of travel. Stroke length easily adjustable in the same fashion used by shapers. Either an electromagnetic clutch or a servo motor to drive the crank, if there is need for dwell between each feed cycle.

Jon

Let the Record show that "Buerste" on or about Thu, 4 Mar 2010 14:31:33 -0500 did write/type or cause to appear in rec.crafts.metalworking the following:

Sounds like a "delay" between when you say "stop" and it does actually stop. Reminds me of a story about teaching a bunch of farmers to use small boats. They were used to tractors, which only turn when the steering wheel turned. Turn the wheel hard over, make a

90 degree turn, then straighten the wheel, "plow on". Small boats, otoh, keep turning. One needs to stop 'turning' before completing the turn, so that the boat is going "straight" when it gets to the new heading.

So "re-calibrate". That is, if X amount of time produces Y amount of wire, shorten X. Seems to me the air limit switch is an idea for doing just that. Cut off the air so that the "overrun" produces the last amount of travel needed to get the right amount of wire. That "over travel" is not a bug, it is a feature.

pyotr

- pyotr filipivich We will drink no whiskey before its nine. It's eight fifty eight. Close enough!

This is kinda sorta the absolutely worst sort of application for steppers. High acceleration = high torque; steppers don't produce much torque for their size or power input compared to DC motors, and when you ask for more torque than they can deliver they lose their place completely. So pushing a stepper for more torque leads to reliability issues -- you can sacrifice great chunks of an otherwise enjoyable engineering career to trying to make an undersized stepper reliably accelerate and decelerate.

Besides, by the time you get a stepper and a stepper controller designed in, you could well have designed in a brushless motor and a servo controller -- and with careful design, the only thing that happens if your brushless motor lacks oomph is that things slow down a bit.

There are applications for which you'd be crazy to use anything other than a stepper, by the way -- if you've got space and power to burn, you only need moderate precision for the speed, and tuning a servo motor may be a questionable exercise, steppers are superlative little gems. But "go from point A to point B damn fast and stop" is not a strength of a stepper, in my experience.

Let the Record show that Larry Jaques on or about Fri, 05 Mar 2010 06:17:53 -0800 did write/type or cause to appear in rec.crafts.metalworking the following:

Oh it works perfectly, just as designed and built! But the results aren't what he wanted. I hate it when machines do what I tell them, not what I want. B-)

- pyotr filipivich We will drink no whiskey before its nine. It's eight fifty eight. Close enough!

Ball screws work very good at this if combined with a PLC and sensor

Gunner

Whenever a Liberal utters the term "Common Sense approach"....grab your wallet, your ass, and your guns because the sombitch is about to do something damned nasty to all three of them.