Re: finding sources of errors
Posted by
Jon Elson
on 2001-03-09 15:34:00 UTC
ballendo@... wrote:
reasonable accuracy, repeatability and orthogonality to start with.
Measuring these parameters is a very good start at building a facility
to make parts of some precision. If you have out-of-spec parts, it
is very hard to know the cause immediately. By isolating the potential
causes as much as possible, it becomes possible to determine the source
of the errors and start working on fixing them. I don't always know
what I'll be making, so I can't just say, "Oh, yes, this part is to spec,
therefore, everyting is fine." I am in much better shape knowing the
approximate errors in my machine, and then being able to determine
what parts in the future might run into trouble, and what parts will
be easily made to spec.
leadscrew dials as the only measuring system were NOT adequate!
You'd drill and tap holes in one part, and then drill holes in another
piece, and they would not line up well enough to put the screws through.
The problem was differential wear on the Acme screws causing about
.015" error from one point to another 6" or so away. (I was compensating
for backlash.) I then got a Bridgeport optical readout (mirrored engraved
glass scales and magnifier/interpolator boxes) and did MUCH better, but
things still weren't as good as I wanted. The Bridgeport opticals were in
horrible condition after being rescued from a scrap yard, and were just
hard to use. Then, I converted to CNC, and things have been much
better, since! When I drill holes in two parts that are supposed to match,
they DO!
Oh, yes. The professional machinists at work run into problems once
in a while, too. They made a big 'ring', part of a vacuum chamber, from
a piece of very thick wall tubing. They left flanges at various locations, but
cut away much of the remaining material, to make for a very thin wall.
When they took it off the fixture and checked the hole alignment with the
covers, they were horrified to find that you couldn't get more than 2 of the
8 holes to line up at a time. Not just a little mis-alignment, but you couldn't
even see any part of the other holes through the holes in the cover. A mis-
alignment of .1" or more! This was due to relieving the stress that was
put into the raw stock by the wrought aluminum finishing! Well, now, of course,
you needed an arbor press to spring the chamber enough to even get the
ring back onto it's own fixture! They are planning to attach the support
fixtures to it and put it in an oven and try to relieve the stresses while it is
held to the correct shape. We'll see if this works! Then, they'll have to
remachine all the flanges flat, again.
Jon
> Jon,What I was trying to point out, was that you need a machine that has
>
> This is exactly right! However, if parts are the end result, then
> the end result needs checked... These forces and issues will, as you
> say, "swamp" machine accuracy. That is precisely why I mentioned
> doing the test parts; if you can't get consistent parts, it may not
> have anything to do with the machine, per se. Could be the chosen
> feeds/ speeds/ tooling/ toolpath.
reasonable accuracy, repeatability and orthogonality to start with.
Measuring these parameters is a very good start at building a facility
to make parts of some precision. If you have out-of-spec parts, it
is very hard to know the cause immediately. By isolating the potential
causes as much as possible, it becomes possible to determine the source
of the errors and start working on fixing them. I don't always know
what I'll be making, so I can't just say, "Oh, yes, this part is to spec,
therefore, everyting is fine." I am in much better shape knowing the
approximate errors in my machine, and then being able to determine
what parts in the future might run into trouble, and what parts will
be easily made to spec.
> Point is, if you CAN'T make what you need to make, then you NEED toWell, I CAN say that the parts I made with my manual Bridgeport with
> find out why; but if you CAN make what you need, then WHY you can is
> an academic exercise... Better to spend your time making the parts,
> and optomising your techniques, which will probably eventually put
> you in the position of needing to know just what the machine is
> doing. Finding out at that point will teach volumes!
leadscrew dials as the only measuring system were NOT adequate!
You'd drill and tap holes in one part, and then drill holes in another
piece, and they would not line up well enough to put the screws through.
The problem was differential wear on the Acme screws causing about
.015" error from one point to another 6" or so away. (I was compensating
for backlash.) I then got a Bridgeport optical readout (mirrored engraved
glass scales and magnifier/interpolator boxes) and did MUCH better, but
things still weren't as good as I wanted. The Bridgeport opticals were in
horrible condition after being rescued from a scrap yard, and were just
hard to use. Then, I converted to CNC, and things have been much
better, since! When I drill holes in two parts that are supposed to match,
they DO!
Oh, yes. The professional machinists at work run into problems once
in a while, too. They made a big 'ring', part of a vacuum chamber, from
a piece of very thick wall tubing. They left flanges at various locations, but
cut away much of the remaining material, to make for a very thin wall.
When they took it off the fixture and checked the hole alignment with the
covers, they were horrified to find that you couldn't get more than 2 of the
8 holes to line up at a time. Not just a little mis-alignment, but you couldn't
even see any part of the other holes through the holes in the cover. A mis-
alignment of .1" or more! This was due to relieving the stress that was
put into the raw stock by the wrought aluminum finishing! Well, now, of course,
you needed an arbor press to spring the chamber enough to even get the
ring back onto it's own fixture! They are planning to attach the support
fixtures to it and put it in an oven and try to relieve the stresses while it is
held to the correct shape. We'll see if this works! Then, they'll have to
remachine all the flanges flat, again.
Jon