Robotics for Module and Stave

ugliestmysticAI and Robotics

Nov 14, 2013 (3 years and 8 months ago)

61 views

Paul Tipton

Yale University

May 3, 2007

Robotics for Module and Stave
Production

P.Tipton

UCSC May 3, 2007

2

New and Growing Effort

Collaboration of BNL (D. Lynn,
D.Lissauer, Y.Semertzidis), LBNL
(C.Haber, G.Gilchriese), and Yale (W.
Emmet, A.Martin, P.Tipton)

P.Tipton

UCSC May 3, 2007

3

Motivation

7500
-
14000 barrel modules needed

Ideally production should last of order
one year (
e.g.,

not three)

A few barrel module production sites
(<5?)

Leads to ~10 modules/day/site

Additional production capacity = schedule
contingency

P.Tipton

UCSC May 3, 2007

4

Other Motivating Factors

Likely lowers cost


Smaller standing army, shorter production time


not enough information yet to estimate true savings


But due diligence demands that we explore automation

Leads to uniformity of production techniques and final
product over many production sites

Quality control inspection also automated

Robotic technology is mature

Having a plan for automated production makes
U.S. stave design more compelling & attractive


Robotics likely to help in almost all ID upgrade
scenarios, not just current U.S./RAL stave design




P.Tipton

UCSC May 3, 2007

5

What exactly are we proposing?

ATLAS SCT used automation in


Automatic glue dispensing


Module placement on the CF cylinders


But module fabrication used fixtures + techs

Want to add robotic ‘pick
-
and
-
place’
technology, integrated with optics and
glue dispensing, for use in both module
fabrication and module placement on
stave frame (
e.g.,

stave production)





P.Tipton

UCSC May 3, 2007

6

Why Start Now?

Because we cannot start any sooner

CMS gantry development effort took ~4
years from purchase of prototype until
they were ready for module production

We have two processes to perfect
(module and stave production) but
arguably have more expertise at t=0.


P.Tipton

UCSC May 3, 2007

7

Three Robotics Options
(+fixturing) being explored:

1)
Buy pieces


servo
-
motors, stages, control
system
-

and build system ourselves

2)
Buy large sub
-
systems and
integrate/customize, adding optics, vacuum,
and glue dispensing to an off
-
the shelf motion
-
control system

3)
Buy a multipurpose work
-
cell with pick
-
and
-
place, optics, and gluing capabilities that
requires little in
-
house engineering &
retrofitting

4)
Bag robotics and use fixturing

P.Tipton

UCSC May 3, 2007

8

Rotating

pick
-
up


tool

cameras

Glue dispense

Cartesian gantry with work head

Sketch of an Option 2 System

Y

X

P.Tipton

UCSC May 3, 2007

9

1 meter

Pitch adapter and hybrid staging area

25 10 x 10

detectors

Module Work Surface

P.Tipton

UCSC May 3, 2007

10

Robotic Assembly Process

1.
Load 25 detectors in work space. Nominal positions are set by pins or edges. Apply vacuum.

2.
Load chip packs containing 100 pitch adapters

3.
Load 100 hybrids into staging area. Only rough placement is required as determined by
footprint marked on work surface.

4.
Survey detectors with cameras on head to determine actual positions. Use focus to determine
detector height.

5.
Survey pitch adapters

6.
Survey hybrids

7.
Dispense adhesives onto detector surface

8.
Pickup first hybrid

1.
Calculate rotation and translation to arrive at correct position on detector

2.
Move to position

3.
Check local fiducials and recalculate correct position for placement

4.
Place hybrid down on detector surface. Vertical drive set by detector thickness.

9.
Pickup first pitch adapter. Follow sequence similar to 8.1
-
8.4

10.
Repeat Steps 8
-
9 for the rest of the components (99 operations).

11.
Inspect, report, end.

12.
Technicians remove workplate and set aside for overnight adhesive cure.

13.
Next
-
day inspection on OGP as cross
-
check, as needed

14.
Load each module into a holder for wirebonding and test.

P.Tipton

UCSC May 3, 2007

11

Stave Production

Aerotech 10000 with 2mx1m work
space is ~$90K, well suited for even the
longest proposed staves.

Or use smaller workspace and ‘index’
stave through

P.Tipton

UCSC May 3, 2007

12

Work Accomplished in FY07
*

Coalesced as a collaboration around the
need to, and how to, explore robotics

Made progress in understanding
production steps, robotic requirements

Survey products (Areotech, Newport)

Study CMS system in detail (4 of us to
visit FNAL next week)

We have a ball
-
park cost estimate for
the hardware for options 2 & 3

*

On an Upgrade R&D budget of $0

P.Tipton

UCSC May 3, 2007

13

Proposed Work for FY08



Our plan for the development work is to factorize
problem into:


motion control
-

Yale


vacuum distribution and control system
-

Yale


optics/pattern recognition
-

BNL


glue dispensing


LBNL


pickup heads/parts carriers


LBNL+Yale

Specify prototype motion and optics systems

Prepare for prototype gantry/optics or work cell
purchase in FY08

Design and construct the vacuum distribution
platform

Glue dispensing

Pickup tool development

P.Tipton

UCSC May 3, 2007

14

Budget Considerations

Materials cost of prototype gantry and vacuum
system covered by Yale


$101K of gantry purchase and materials for vacuum system


Will also subsidize engineering (cost to project is $62/hour with
no additional overhead)

Asking for:


Balance of engineering costs to specify gantry and design
vacuum distribution platform at Yale


Machining time for fabrication of vacuum distribution
system, also at Yale


Optics hardware for BNL work to begin


Glue dispensing


Pickup tool design


P.Tipton

UCSC May 3, 2007

15

Budget Details

FY08
WBS 4.1.4.4 Prod. Automation
Upgrade R&D
Base
Other
Funds
Funds
Funds
Materials, Procurements
OPTICS
Cognex or Matrox, camera, optics
framegrabber card, image analysis
DAQ and interface to motion cont.
32,000
MOTION
AGS 10,000 2mx1m gantry w/
5cm aluminum baseplate,
Z and Phi axis, DAQ, Control SW
91,000
PC for AGS control
3,000
Vacuum System
vacuum distribution platform w/
valves, valve actuators,
Additional DAQ channels
7,000
GLUE DISPENSING
LBNL
Pickup Head Development
LBNL+Yale
subtotal
32,000
0
101,000
Personnel/Travel
Yale Engineer for motion system spec
and vacuum system design
43,000
18,000
21000
Yale shop for vacuum system fabrication
23,000
subtotal
66,000
18,000
21000
GRAND TOTAL
98,000
18,000
122,000
P.Tipton

UCSC May 3, 2007

16

Conclusions

In FY07 our progress was good, but very soon
further progress will require us to start spending
money (engineering, then materials)

In FY08 we plan for success, want to be ready for a
timely purchase of an appropriate prototype
automated production system, soon after it becomes
clear what will be the comprising pieces and
construction steps for module and stave

This timeframe looks to be the second half of FY08.

FY08 R&D funds will be highly leveraged in this
activity

P.Tipton

UCSC May 3, 2007

17

Backup Slides

P.Tipton

UCSC May 3, 2007

18


The CMS Gantry

~1998 technology


Aerotech AGS 10000 Gantry


Added optics and ‘frame
-
grabbing’


Added custom vacuum plates & chucks


Added pneumatics for glue dispensing

Production recipe:


Establish coordinate system


Find position of objects


Glue dispensing


Pick
-
and
-
place


Inspection


Load next plate

Uses fiducials to locate and place each piece

3 Modules per ‘tray’, up to 8 trays per day

Second
-
day inspection on a separate machine as a cross
-
check