Newsgroups: comp.robotics
Path: brunix!news.Brown.EDU!noc.near.net!howland.reston.ans.net!europa.eng.gtefsd.com!uunet!world!jamcorp
From: jamcorp@world.std.com (Jonathan Priluck)
Subject: orders of magnitude causes immense problems
Message-ID: <CBzp1L.5DM@world.std.com>
Organization: The World Public Access UNIX, Brookline, MA
Date: Thu, 19 Aug 1993 04:41:44 GMT
Lines: 109


     I am sick of this quibbling about "toys" and "reals".  I have
a neat problem, and welcome input from one and all.  Here is the
deal, I am building a miniature milling machine.  The basis of the
project is to build small machines with which to build still
smaller robots.  Ideally (translation: ten years from now) we would
like to use some sort of virtual reality display to translate full
scale actions on a bridgeport type milling machine down to a milli
scale milling machine.  So a machinist would manipulate parts on
full size equipment making a full size part, these motions would be
duplicated, down on the miniature machine, to produce a
miniaturized duplicate of the full sized part.
     Ultimately we hope to develop a miniature machine shop,
capable of turning out miniaturized parts just like a full scale
shop.  The goal of all this is to be able to build very small
robots as easily as we can build large ones.  We have already
developed some miniature welding equipment, we can easily
manipulate and weld .0001 wires and build little tiny buildings and
structures.  Take it as given for the rest of this discussion that
we can do miniaturized welding with excellent weld quality and
precision of +or- .05 mills (a mill is one one-thousandth of an
inch, for those of you not familiar with machinist jargon) for both
the location and penetration of the weld.  But we can only do spot
welding, no fillets, and no butt joints.  If you want fillets and
such we can do +or- 1 mill.
     We have decided to approach the problem by first building the
miniature machines and worrying about the interface after we get
them working, so the rest of what I have to say relates only to the
hardware.  But dont let that discourage you if you wish to discuss
the controls, I am sure plenty of people here will be very
interested even if I have no idea what you are talking about.

Here are some of the major areas of difficulty we are already
encountering.  

1) Feeding in raw material, this cannot be done through an
electronic interface, raw material must be introduced into the
system somehow.  One option is to use larger machines to produce
scaled down "stock", for example .001 wires would replace 1" rod
and thin slices of sheet metal would become "bar stock".


2) HEAT.  Heat is a big problem, the trouble is that the laws of
physics are not invariant under scale translations.  Initially we
though the non-translation of force scaling would cause most of the
problems but generally the changes in the force applications due to
the change of scale work to our advantage.  But heat is another
matter.  Our little tiny parts don't have much total heat capacity,
and we are still cutting and generating heat right at the
interface.  The trouble is that the whole part heats up and deforms
unless serious precautions are taken.  To add insult to injury the
speeds of our cutting tools have to be very high, for those of you
who are familiar with machining the formula for the RPM of your
cutter is CS X 4 / diameter.   that means cutting speed multiplied
by four divided by the diameter of the cutter (or the diameter of
the part if you are on a lathe and the part is turning).  Needless
to say we need to turn a .01 endmill rather fast unless we are
machining wax (which has a very low cutting speed). So not only is
the trouble with friction increased but our little motors also need
to be kept cool. 
     One option we are considering seriously is to enclore the
whole "machine shop" in an artificial "atmosphere" of some
appropriate liquid.  There is some precedent for this in
watchmaking.  Divers watches (and expensive rolex watches) are
filled with oil or alcohol.  The reason for this is to help resist
water pressure by removing all the air from inside the  the watch
case.  Our reasons are different but an appropriate fluid with a
high specific heat could carry away a lot of heat.  And if a
current (convestion not electrical: like the current in a stream)
was introduced it has the potential bonus of providing both
lubrication and chip removal.
     Another option is to do everything in a very low temperature
environment.  This would increase the temperature gradient thus
increasing the rate of heat transfer out of the part.  Perhaps some
combination of cold and liquid would work.

3) Micromanipulators.  Since we have chosen to mimic the full scale
methods of machining we need some equipment to stand in for a
"person" who is the actual machinist in the full scale.  Some
options:  1) Move the miniature machines not the part. 2) Build
miniature waldo's (a daunting task which would be much easier if we
*already had* our miniature machine shop available for its
construction).  3)Build a simplified robot arm (same trouble as
with the waldos).  
     Currently we use commercially available hand operated
micromanipulators combined with the super secret HOT-M (which
stands for Hand Operated Tweezer Method :^) ).  I guess in general
this should hold us for a good long time, at least untill we have
built all the machines (miller, lathe, drill press, sander,
scraper, etc.)

4) Tolerances.  This is a MESSSSSSS I don't even want to get
started, its like a catch 22.  You need small machines to make
parts to high enough tolerances but you need the parts first so you
can build the machines to make those parts.   AAAAAAAAAARRRRG
<picture me sitting at my desk pulling my hair out :^)  >  


Well that should be enough to get a good discussion going, I look
forward to your responses.  And be creative, its just a discussion
:^) 

Regards, Jon Priluck

-- 
*   Jonathan Aerospace Materials Corp., 41 Naples Road, Brookline MA 02146  *
*            Tel (617) 731-3637, Internet:  jamcorp@world.std.com           *
*     Developers and future manufacturers of Lattice Block Materials ...    *
*                the world's strongest and lightest materials.              *
