Re: [CAD_CAM_EDM_DRO] Rotary to linear translation - an interesting idea.

>
> "D.F.S." wrote:
> >
>
> > Because the Length of that compressed surface will not be the same
> > as the Length of the centerline on the ribbon or cable.
> >
> > To restate your statement:
> > The wheel interfaces with the surface of the ribbon, which is
> > compressed. Not it's centerline which is the length we are
> > actually trying to measure.
>
> Ok, I think I see your logic here. Well, I'm math challenged, but CAD
> capable. I drew a 1" dia circle and a .998 dia circle. Trimed to a 60
> degree arc for both circles. I assumed a .005 ribbon. When metal bends,
> it stretches more than it compresses and the theoretical centerline does
> not follow the exact center of the material through the bend, that's why
> I went for .002 difference instead of .0025. Across this arc, the 1"
> segment was .523599 and the .998" segment was .522552 long, for a
> difference of .001. We'll forget about the 47 millionths as that's
> beyond the scope of hobby level machining.
> So we do see a change to the distance of the ribbon where it wraps
> around the wheel. However, I think we need to take into account the fact
> that there are two bends in the opposite direction which will be
> -stretching- the same surface we are trying to measure against. Since
> the material stretches more than it shrinks in a bend, one would have to
> really pay attention to these effects to make sure they don't cancel out
> the very compensation they are trying to effect.
> I can't quite explain why, but I still don't think it's real-world
> workable. Maybe I'm too hung up on the Travdial model of operation.
> However, I'll admit it's possible in theory, and would really like to
> see someone make a working example and prove it in practice before I
> fire up the BBQ...<G>

Ok, some clarification and more speculation.

The spindle that keeps being mentioned is 5/16? or was it
metric? I'll assume .3125.

You can't get too thin on the cable, or it will be too fragile.

Figure ~60 thousanths for kicks.

With all the trouble all the traffic would suggest people have finding
thin flexible cable, I have to wonder about where they are going to
find thin stainless metal strips of the quaility and finish this would
require.

So anyway, figure 60 thousandths cable and .3125 shaft.

The distance at the surface of the shaft for one revolution,
rounded to 6 places is: .981748.

Presuming the resulting centerline of the cable is 1/3 diam out from the
inside edge, that gives a diameter of .3125 + 2 X 10 Thou. or .3325

Distance for one revolution at the new virtual surface is:
1.044578

The difference is substantial at almost .063" / inch of travel.

This is not an issue IMHO, as it is linear.

Forget all the calculations, build it then calibrate it.

The only real issue is the software needs to be able to accept actual
figures as the distance one "Click" measured on the wheel represents.


The other issue I mentioned is the fact this compression and expansion
of the cable or metal band is taking place on the surface of the measuring
wheel shaft.

Presume 180 Degrees of contact. with the cable going over the top
of the staft. At points 9 & 3 O'clock the band is either squashing
together or pulling apart relative to the surface of the band and the
surface of the shaft.

How much creep is this going to introduce to the relative position
of the relative surfaces as a whole?

Will it be linear?
Will it be repeatable?
What will a small bead of oil on the band or cable do?
Will it tend to create an oil wedge in the uptake side and
make the shaft creep faster than the speed of the surface of the
band as the pay-out side of the band expands with less than an
equal conteraction from the uptake side?

Will it creep depending on traversal speed? will something like
running a power feed with a slow feed and quick reverse cause the
position to creep over time?
Note, I'm not talking about a quick jerk throwing it out of whack,
I'm just talking about a 10 inch/minute feed, and 30 inch/minute
return to make a new pass.

The only way to really know would be to build it and try it.

One real advantage is in many cases we are talking precision eqpt
and computer controls.

It would seem reasonably easy to set a system up to run a test machine
for a few days and hammer on the DRO system.

Move all over the place in random, cyclic, fast, slow, symetrical,
asymetrical movements in loaded and unloaded conditions.

Re-zero to known locations on a regular and automated schedule and
save the results in a data file.

We could disect it to death, the only real way to know will be to
make an educated guess, and test the results.


I don't know how plausible a system it is either.
A good test will answer that question.

Also as I stated before, I already have too many projects, and
This system is not something I need myself.
I'd do it if it was.

Reality needs to hit at some point, we could guess forever.

Like they say:
If you answer all the questions,
then question all the answers,
only the doubt remains.


I think the only definitive answer is an actual test.

Marc