The Ceramic Samurai

1
The Ceramic Samurai

• Ceramic Lathe Concept by
• Roger Cortesi

Alex
Roger
rcortesi_at_alum.mit.edu http//pergatory.mit.ed
u/rcortesi/
Precision Engineering Research Group Massachusetts
Institute of Technology, Mechanical Engineering
Department
2
Basic Machine Requirements

• Turn Ceramic Parts of up to 15 diameter and 12
  long
• Target Part Accuracy of 1 micron (0.00004)
• Provide a Dry option for turning green parts
• Robust with respect to Ceramic Swarf
• Minimal Assembly

3
Early Machine Concepts

• Simplest to Manufacture

• More complicated to manufacture
• Swarf can fall away from bearing surfaces

• A possible design if it is not practical to move
  the spindle work piece

A few more variations on these concepts were
considered in the 1st pass sketch models
4
The Dry Machine and Wet Machine have sufficiently
different requirements, that they have developed
into separate machine concepts.

• Dry Machine
• Much lower accuracy since it will be rough
  turning parts while they are still green.
• No liquid coolants or lubricants to trap swarf
  and cause problems in firing the green part.

Roger in the desert and very dry (just like the
dry machine)

• Wet Machine
• A much higher stiffness and cooling requirement
  since it will turning finish parts to their 1
  micron target accuracy.
• Liquid coolants and lubricants are not a problem
  since the part has already been fired.

Roger underwater and wet (just like the wet
machine)
5
The Dry Machine and the Axtrusion
The Axtrusion is a linear motion concept
developed and implemented by Prof. Slocum and
Roger Cortesi.
It uses porous air bearings and linear motors to
make an easy to assemble, non-contact linear
motion system.
Because it is a non-contact air system it will
should be very robust with respect to the ceramic
swarf
6
2nd Generation Sketch of Dry Machine
Rough Error Abbe Errors were calculated for this
design with the following assumptions

• Average errors in Prototype Axtrusion used for
  the work piece and grinder carriages pitch, yaw,
  and roll errors.
• No MAPPING CASE Error magnitude is the actual
  error in prototype, 10 mradians (2 arc sec)
• PERFECT MAPPING CASE Error magnitude is the
  repeatability of prototype 2.5 mradians (0.5 arc
  sec)
• No errors in spindles
• Rough machine size based on a 15 dia. by 12
  long work piece
• All Magnitude Abbe errors are added for a worst
  case

• Maximum Total Abbe Error
• With NO Error Mapping
• Radial 13 mm (0.0005)
• Axial 10 mm (0.0004)

• Maximum Total Abbe Error
• With PEFECT Error Mapping
• Radial 3.2 mm (0.00013)
• Axial 2.5 mm (0.0001)

7
Notes on the Dry Machine Concept
The previous error estimates are for an Axtrusion
with a permanent magnet linear motor. A coreless
linear motor would dramatically reduces these
error motions further.

• Total RMS Abbe Error
• With NO Error Mapping
• Radial 6.6 mm
• Axial 5.0 mm

• Total RMS Abbe Error
• With PEFECT Error Mapping
• Radial 1.6 mm
• Axial 1.2 mm

• Estimated Total Abbe Error
• With NO Error Mapping
• Radial 9.8 mm
• Axial 7.4 mm

• Estimated Total Abbe Error
• With PEFECT Error Mapping
• Radial 2.4 mm
• Axial 1.8 mm

8
The Wet Machine
Hydrobushing Rail
Ballnut and Drive Motor Assembly
Ballscrew
Hydrobushing
9
The Wet Machine

• This machine was designed to maximize accuracy at
  a minimum of cost
• The motive force could come from either a
  coreless linear motor or a ballscrew.
• The linear guides could be either traditional
  ballrails or a Hydrobushing
• It should be very easy to assemble and error map

10
Wet Machine Concept
Better dynamic stability due to heavy (and
varying) workpiece mass mounted between rails It
is easier to integrate wheel dressing station,
since grinder is moving orthogonal to its axis.
Better Access to Work Piece
11
Minimizing the Abbe Errors
The machine was designed to bring the Center of
Motion (COM) of the two axes as close the contact
point as possible. If the COM for an axis can be
brought inline with the axis of the spindle
(either the grinder or work piece), then all the
errors associated with either roll, pitch, or yaw
of that carriage will vanish It is preferred to
use the above technique to eliminate error due to
carriage roll, since mapping the carriage roll
over the length of travel is much more
difficult. The Abbe errors in the X direction
for the wet machine are estimated to be about
half than the dry machine. The Abbe errors in the
Z direction for the wet machine are about one
third that those for the Dry Machine
12
Minimizing Motion Errors

• The motive force for the two axis can be applied
  through the COM for each axis. This minimizes
  errors as the carriages are accelerated.
• The Ballscrew and Ballnut/Drive Motor Assembly is
  designed to be removed straight of the end of the
  machine, making it very easy to replace when the
  swarf kills it.
• A similar implementation could also be used for
  a coreless linear motor in place of the Ballscrew
  (eliminating the wear problem)

Ballscrew acting on spindle carriage COM
13
The Wet Machine Structure
The wet machine is base on a very simple
structure. The two main pieces are shown below.
These could be aluminum extrusions, a casing
(polymer concrete, cast iron, etc) or pressed
aluminum oxide. A third piece not shown will be
needed to join the structure.
14
Hydrobushing Rails
Because the axis only require a short range of
motion the Hydrobushing rails can be completely
round. This means that fully round Hydrobushing
can be used. This eliminates the compliance due
to a partially round Hydrobushing.
15
Hydrobushing Stiffness Estimate
16
Questions Still Remaining

• Where does one get the workpiece spindle? (This
  component will heavily affect the design.)
• Are there suitable, infinite stiffness, and
  infinite life spindles available?
• Should we build our own with the Hydrobushing.
• Is there a standard grinding spindle that you
  like to use?
• What the tradeoffs between a linear motor and an
  easily replaceable ballscrew?
• What is the manufacturing cost goal?

17
Hydrobushing Supplemental Information
18
Hydrobushing Modular Bearings Concepts

• Aesop Inc.s self-compensating water hydrostatic
  bearing technology can be used to enable machines
  where
• Bellows can be eliminated
• Bearing maintenance and wear can be virtually
  eliminated
• Bushings can run on standard hard ground 2
  diameter Stainless Steel shafting
• Self-cleaning action can enable a friction drive
  to be used, thereby also eliminating the
  ballscrew
• Modular all features injection moldable or
  lost-wax castable (using rapid prototyping)

Compensator-to-pocket connection grooves on OD
Upper compensator for lower pocket
Picture frame upper pockets for greater damping
Supply groove
Lower compensator for upper pocket
19
RodWay Self-Aligning Hydrostatic Ways

• The RodWay system minimizes the need for
  precision alignment of bearing ways
• Accommodates change in way parallelism if machine
  foundation changes
• Machine accuracy is not affected by self-aligning
  function
• Enables low-cost use of water hydrostatic
  bearings in rugged industrial environment
• Old US Patent (4,637,738) OMAX can improve on
  with damping materials and rod-end bushing to
  create RodWay design and trademark
• Operating principle
• One hard ground stainless steel round shaft (the
  fixed shaft) is mounted in a normal manner using
  supports that can be replicated to structural
  steel so the shaft remains straight
• The second round shaft (the pivot shaft) is
  mounted as parallel as is reasonable to the first
• One hydrostatic bearing carriage rides on the
  fixed shaft, and it is bolted to the machine
  structure (e.g., bridge structure risers)
• One hydrostatic bearing carriage rides on the
  pivot shaft, and it is connected to the machine
  structure (e.g., bridge structure risers) by a
  spherical bearing
• A steel hemisphere stud mounted in the machine
  structure and resting in a replicated spherical
  seat in the bearing carriage, where the center of
  the hemisphere is a distance H above the round
  shaft center
• Alignment errors (pitch and yaw) between the
  round shafts are accommodated by the spherical
  bearing
• Center distance errors (d) between the round
  shafts are accommodated by roll (q) of the
  bearing carriage.
• Vertical error motion (D) of the hemisphere is a
  second order effect
• Example d 0.1, H 4, q 1.4 degrees, and
  D 0.0012

20
Axtrusion Supplemental Information

• The slides that follow contain
• Background information on the Axtrusion
• Performance Data on the Prototype Axtrusion

21
Axtrusion Components
Not Shown Position Encoder and Position Encoder
Scale
22
How the Axtrusion Works
The attractive force between the motor coil and
magnets preload the air bearings.
Changing the values of q, ym, and zm the values
for Fside, Ftop1, and Ftop2 can all be set
independently
23
The Bench Level Prototype
Magnet Adjustment Screw
Magnet
2 Rollers (side surface)
Way side surface
Way top surfaces
3 Rollers (top surface)
The first Axtrusion ever built! It uses 5 CAM
rollers as bearings
24
The Pitch, Yaw, Position Accuracy,Vertical
Straightness, and Vertical Stiffness Setup.
The basic setup for the pitch, yaw, and accuracy
measurements. A laser and a variety of optics was
used to make each measurement.
The basic setup for the vertical straightness
measurement. The probe is suspended above the
straight edge on the carriage.
The basic setup for the vertical stiffness
measurements The load is applied in the center of
the carriage and measurements made above each of
the top bearing pads.
25
The Pitch Data
Period of Variation 29.9 mm
This period is the distance between alternating
poles on the magnet track!
Measurements every 10 mm Carriage Speed of 10
mm/sec 6 passes (3 in each direction)
The way also appears to be slightly curved (from
0.5 arc seconds to 0.5 arc seconds).
Pitch data was also taken at a carriage speed of
40 mm/s, after incremental movements, and as a
function of time. There were no major
differences. Please see the appendix for the
complete data.
26
The Yaw Data
Raw Yaw Data
Yaw Data with Linear Slope Removed
Measurements every 10 mm Carriage Speed of 10
mm/sec 6 passes (3 in each direction)
Period of Variation 30.1 mm
Pitch data was also taken at a carriage speed of
40 mm/s, after incremental movements, and as a
function of time. There were no major
differences. Please see the appendix for the
complete data.
27
The Linear Accuracy Data
Raw Linear Accuracy Data
Linear Accuracy with Linear Slope Removed
This is the difference between the actual
carriage position and the position that the
controller thinks it is in.
28
The Vertical Straightness Data
An Abbe error of 1.2 microns vertical
displacement at both edges of the carriage.
Data for displacement of carriage center
Period of Variation is 28.6 mm
Notice that the hourglass shape of the data is
due to pitch errors measured away from the center
of pitch rotation. To get better data, the test
must be rerun with the probe mounted in the
middle of the carriage and the mirror suspended
above it. The data from the center of the
carriage shows the pure translation of the
carriage to be about 0.3 microns.
Data for displacement of carriage edges.
Two passes are shown for a carriage speed of 10
mm/s
Because the plane mirror could not be leveled
perfectly, any linear slope in the data was
removed.
29
The Vertical Stiffness Data
Masses were place on the top center of the
carriage and the displacement above the four (4)
top air bearings was measured.