Assembly Sequences

advertisement
2.875 – Fall 2001
Annabel Flores, James Katzen
2.875 - Fall 2001
Mechanical Assembly and Its Role in Product Development
Term Project: Report #5
DESIGNING AN ASSEMBLY PROCESS TO PRODUCE COMPUTER MOUSE ASSEMBLY
November 28, 2001
Annabel Flores
James Katzen
Photos taken from:
http://www.petergof.com/x-ray/Mouse.htm
http://www.Mousemorf.com/images/Mouse2b.jpg
Page 1 of 15
2.875 – Fall 2001
Annabel Flores, James Katzen
DESIGNING AN ASSEMBLY PROCESS TO PRODUCE COMPUTER MOUSE
ASSEMBLY
Introduction
The Microsoft Mouse Version 2.21 is an ergonomic, dual-button mouse. The simple,
eleven-part design provides an opportunity to analyze the product’s assembly characteristics.
Previous reports analyzed individual parts’ design, the interface between a subset of parts of the
mouse, the key characteristics and the functional requirements of the product, the proper
assembly of the Mouse, and the design and layout of a particular Workstation for assembly of the
Mouse.
This report will present the chosen assembly sequence and its advantages and drawbacks in
comparison to other assembly sequences. In addition, the floor layout of the assembly process
will be addressed, including process equipment, material handling/storage, and operator
placement. Production rate, production volume, and cycle times for each operation will be
estimated. Finally, the flow of parts and operators involved in the assembly sequence will be
choreographed. Unfortunately, computer simulation tools were not successfully used to study
the chosen assembly sequence. Efforts will be made to find a suitable computer that supports the
recommended programs, and if found, a simulation analysis will be included in the next project
report.
Assembly Sequences
As in the previous report, we assume that specific product redesigns would have been made
to the Mouse to facilitate assembly. These product redesigns include:
 Relocation of connector block of Circuit Board to side closest to where strain relief of
Cord passes through Mouse Base.
 Shortening of Cord to accommodate relocated connector block and eliminate need to
route Cord around edge of Mouse Base.
 Redesign of strain relief of Cord to permit top-down assembly mating of strain relief into
Mouse Base.
 Redesign of snap fit features on Mouse Base and Mouse Cover to permit top-down
assembly mating of these two parts.
 Redesign of Ball Holder to a simpler disk design that eliminates a subassembly.
 Elimination of “mystery feature” on Mouse Base to permit top-down assembly mating of
Circuit Board into Mouse Base.
1
Refer to Appendix A and Appendix B, respectively, for Bill of Materials and Exploded View for part naming and
numbering conventions.
Page 2 of 15
2.875 – Fall 2001
Annabel Flores, James Katzen
Figure 1: The "Mystery Feature" on Mouse Base that complicates necessitates reorientation of Circuit Board
during assembly.
As has been described in previous reports, there are a number of different assembly
sequences that may be used to assemble the Mouse. At the extremes, the product could be
assembled using fully automated assembly, fully manual assembly or a hybrid combination of
the two. The following section outlines the assembly sequences with a comparison of these three
methods.
Selected Assembly Sequence – Hybrid Design
A previous report described a number of potential hybrid assembly sequences for this
product. The optimum choice resulted in better efficiencies as it minimized unnecessary
operations. It was decided that these operations would logically be grouped into five
Workstations: one manual station, followed by three robotic stations, and then followed with one
final manual station. The complete list of the stages required to assemble the mouse is shown
below:
 Station 1 (Manual):
o Step 1: Place Mouse Base onto Primary Fixture
o Step 2: Place Wheel onto end of Spring
o Step 3: Assemble Wheel and Spring Subassembly with Mouse
o Step 4: Transfer pallet between Station 1 and Station 2
 Station 2 (Automatic):
o Step 5: Locate Pallet on Locating Pins
o Step 6: Assemble Circuit Board with Mouse Base.
o Step 7: Attach Plug of Cord to Circuit Board and attach strain relief of Cord to
Mouse Base
o Step 8: Attach Horizontal Gear and Vertical Gear to Mouse Base
o Step 9: Attach Mouse Cover to Mouse Base
o Step 10: Transfer between Station 2 and Station 3
 Station 3 (Automatic):
o Step 11: Invert Assembly and place into Secondary Fixture.
o Step 12: Transfer between Station 3 and Station 4
 Station 4 (Automatic):
o Step 13: Place Ball into Mouse Base
o Step 14: Secure Ball and Mouse Base by inserting and twisting Ball Holder
Page 3 of 15
2.875 – Fall 2001
Annabel Flores, James Katzen

o Step 15: Secure Mouse Base and Mouse Cover by inserting and tightening Screw
o Step 16: Attach top Sticker Pad
o Step 17: Attach bottom Sticker Pad
o Step 18: Attach Hologram Sticker
o Step 19: Transfer between Station 4 and Station 5
Station 5 (Manual):
o Step 20: Plug in Cord’s connector into Test Fixture Connector Block
o Step 21: Functional Test (computer controlled with automatic data collection)
o Step 22: Remove Cord’s connector from Test Fixture Connector Block
o Step 23: Bundle Mouse Assembly with Product Documentation and Software
o Step 24a: Pack into OEM packaging OR
o Step 24b: Pack into aftermarket packaging
Appendix C shows the proposed floor layout for the hybrid assembly sequence.
Fully Automated Assembly Sequence
Successful incorporation of a full robotic assembly process would require a significant
amount of product redesign. These redesigns would be required to ensure adequate delivery of
all key characteristics, especially the KC between the Mouse Ball and the Gears (as this has
largest effect on proper mouse functionality).
It was found in prior analysis that the KC between the Mouse Ball and the Gears is met
through the use of a spring loaded Wheel that pushes the Ball against the Vertical and Horizontal
Gears. The wheel must contact the Ball in a specific location to minimize friction. The current
design uses a long thin steel Spring and a small plastic Wheel to achieve this function. The
current design must be manually assembled first into a subassembly then onto the Mouse Base,
because of the asymmetrical properties of the Spring. To allow for automated assembly, which
includes robot feeding, handling and assembling, the Spring and Wheel would need to be
modified or eliminated.
In addition, design changes would have to be made to the Cord. Since we believe that
there is a need to test at least a representative sample of the assemblies prior to shipment, it is
necessary to plug the Cord into a test fixture at the functional test station. Since the current Cord
is quite flexible, it would prove very difficult to program a robot to deal with the Cord’s
flexibility while the Cord’s connector is manipulated into and out of the test fixture.
As can be seen, there are drastic changes that need to be made to the current design of the
mouse to allow for full robotic assembly. As such, this assembly sequence was judged to be
being infeasible at this time and was discarded from further consideration.
Fully Manual Assembly Sequence
A possible assembly alternative is to assemble the product using only manual assembly.
This assembly sequence would allow a cellular design to be used as detailed below:
Page 4 of 15
2.875 – Fall 2001
Annabel Flores, James Katzen
Racks
From
Kitting
Workstation 2
Workstation 1
To Shipping
Workstation 3
To Kitting
Test Area
Workstation 4
Figure 2: Full Manual Assembly, Cellular Design
This layout requires that the individual components to be combined in a kit outside of the work
cell and then fed into the cell next to Workstation 1. At each station, the operator would take
parts from the kit and perform the required assembly operation(s). The subassembly would then
be placed onto a tray. The racks and trays would then be moved to the next Workstation, with
kits traveling on the racks above the work surface and trays being slid along simple roller type
conveyors. At the completion of Workstation 3, where all components have been assembled and
the kit is thus empty, the kit container would then exit the cell and be returned to the kitting area
for a new kit to be made. This full manual assembly would still use the same assembly sequence
as other assembly designs, however in this case, there would be fewer Workstations and the
work would be divided slightly differently.

Station 1:
o Step 1: Transfer kit from delivery area onto rack
o Step 2: Transfer tray from empty tray storage area onto conveyor
o Step 3: Place Mouse Base onto Primary Fixture
o Step 4: Place Wheel onto end of Spring
o Step 5: Assemble Wheel and Spring Subassembly with Mouse Base
o Step 6: Assemble Circuit Board with Mouse Base
Page 5 of 15
2.875 – Fall 2001
Annabel Flores, James Katzen



o Step 7: Transfer tray out of Station 1
o Step 8: Transfer kit out of Station 1
Station 2:
o Step 9: Transfer tray into Station 2
o Step 10: Transfer kit into of Station 2
o Step 11: Attach Plug of Cord to Circuit Board and attach strain relief of Cord to
Mouse Base
o Step 12: Attach Horizontal Gear to Mouse Base
o Step 13: Attach Vertical Gear to Mouse Base
o Step 14: Attach Mouse Cover to Mouse Base
o Step 15: Transfer tray out of Station 2
o Step 16: Transfer kit out of Station 2
Station 3:
o Step 17: Transfer tray into Station 3
o Step 18: Transfer kit into of Station 3
o Step 19: Place Ball into Mouse Base
o Step 20: Place Ball Holder into Mouse Base
o Step 21: Transfer kit from rack to takeaway area
o Step 22: Secure Mouse Base and Mouse Cover by inserting and tightening Screw
o Step 23a: Transfer tray out of Station 3 (go to Step 28) OR
o Step 23b: Transfer tray to test area
o Step 24: Plug in Cord’s connector into Test Fixture Connector Block
o Step 25: Functional Test (no operator required)
o Step 26: Transfer tray out of Station 3
Station 4:
o Step 27: Remove Cord’s connector from Test Fixture Connector Block
o Step 28: Transfer tray into Station 4
o Step 29: Attach top Sticker Pad
o Step 30: Attach bottom Sticker Pad
o Step 31: Attach Hologram Sticker
o Step 32: Bundle Mouse Assembly with Product Documentation and Software
o Step 33a: Pack into OEM packaging OR
o Step 33b: Pack into aftermarket packaging
o Step 34: Transfer finished product out of Cell
o Step 35: Transfer empty tray to empty tray storage area
A cellular design allows a single-piece flow operation that is very flexible to product changes.
Much more skilled and active operators are required since it is the operators that not only
physically assemble the product but also control the material flow throughout. Because the cell
requires that all the components be present before assembly, additional employees are needed to
create each kit. However, a missing operator does not necessarily shut down production, as
operators will likely be cross-trained and shift between stations as batches are completed. No
sophisticated machinery is needed with this sequence other than the Test Fixture.
Page 6 of 15
2.875 – Fall 2001
Annabel Flores, James Katzen
Analysis of the Chosen Assembly Sequence
An analysis of the complexity of the Mouse assembly process and the estimated volume
led us to decide upon a hybrid assembly process that uses both manual stations and automatic
stations. A fully automated assembly sequence is clearly not a viable option even with the
proposed redesigns mentioned earlier. The section below describes the general advantages and
disadvantages with pursuing the proposed hybrid design instead of the cellular design.
Advantages of Selected Assembly Sequence
The selected hybrid design provides a number of advantages over a manual assembly.
The automated Workstations provide a uniformity and repeatability of task completion. We
assume that the cycle time for a product is much longer in the manual assembly sequence and
therefore a larger product volume could be achieved with an automated assembly. Because most
of the costs of a manual assembly are variable, they increase proportionately to an increase in
product volume. An automated assembly, however, is comprised largely of fixed costs causing
unit costs to drop as the volume is increased. The heavy machinery is typically customized to
respond to the product’s design. However, since computer mice are relatively similar, the cost of
machinery can be distributed over product families. Minimal manual labor is required in the
hybrid design, thereby reducing the cost of fringe benefits and salary associated with manual
labor. Similarly, much of the employee training can be eliminated.
In general, the capacity of automated machines can be modified to respond quickly to
changes in product volume. While it is a viable option to use a full manual assembly process, we
feel it was important to complete our analysis of the hybrid assembly selected from previous
reports to explore the capabilities of robotic assembly. A subsequent report will compare the two
assembly processes on cost of capital, cost of labor, and product cycle time among other things.
Drawbacks of Selected Assembly Sequence
There are a number of disadvantages to a hybrid assembly sequence that make the full
manual assembly more attractive.
To put an automated assembly in place, invariably higher up-front costs of assembly
equipment and material handling equipment are needed. However, not just the cost of the
equipment must be considered. The cost of the integration software, which is substantial for
each new product, must be included. This raises the initial costs of building an assembly line
that has some automated equipment.
In addition, once the assembly line is operational, there are still additional drawbacks to
the selected assembly sequence. First, to accommodate the inherent variability in manual
operation output, it is necessary to keep relatively large buffers of work in process between
stations. Also, because product tests are not performed until the end of the assembly process and
visual inspection is difficult to integrate into the sequence, it is more likely that defective
products are created that must be either scrapped or reworked. Finally, the high cost of training
operators to troubleshoot and maintain the automation equipment could be considerable.
Selected Assembly Sequence
The choice to design a hybrid assembly process instead of a full manual assembly process
was driven largely by our own intellectual curiosity about robotic assembly. Both fully manual
and hybrid assembly are viable options for this product, and they should be thoroughly analyzed
and evaluated before either one is implemented. A curious discover is that, as the product was
Page 7 of 15
2.875 – Fall 2001
Annabel Flores, James Katzen
redesigned to make robotic assembly possible, it became much easier to assemble the product
manually. Thus, many of the justifications for the need for robotic assembly, have been rendered
moot, after design simplification.
Production Rate, Production Volume, and Cycle Times Calculations
The production volume was based on Microsoft’s claim that 2 million products were in
the market. We have made the assumption that a production run would last roughly 2 ½ years.
Using these estimates, this requires that roughly 800,000 assemblies would be required each
year. We know from the product label that the assembly was made in China. Assuming that
there are roughly 250 workdays in a Chinese production year, a production rate of 3200 products
per workday is required. Assuming there are 8 hours per Chinese shift, and 2 shifts per Chinese
workday, a cycle time of 18 seconds per Mouse Assembly is required.
In the previous report, we stated that it would be efficient for multiple assemblies to be
placed on the same pallet to minimize the total number of tool and gripper changes required.
Based on recommendations, we selected a pallet design that would hold 4 assemblies at once.
Thus, using a batch size of 4, with the cycle time (per Mouse Assembly) of 18 seconds, a pallet
of four assemblies must be produced every 72 seconds.
Note that this cycle time calculation assumes a production yield of 100% fault-free
assemblies. Since this is an unrealistic assumption, we must increase the production requirement
for the process. Since we can expect that the circuit board (the major source of product failure)
to be tested at the supplier’s facility, it is likely that there will be few unusable assemblies. Thus,
if we assume a yield percentage of 95%, the required production rate becomes roughly 3400
Mouse assemblies per day, or a cycle time of 17 seconds per Mouse assembly, and a resulting
batch cycle time of 68 seconds.
Note also that this batch cycle time of 68 seconds is the end-of-line rate. Upstream
stations in the assembly process will have to produce at a marginally faster rate, to account for
minor production losses due to repositioning/rework at the manual stations, which could result in
unscheduled downtime losses due to blocked or starved conditions. This source of yield loss
could compound over time to be a major source of lost production volume, so it is vital that it be
accounted for when planning the cycle time of the process. Assuming an in-line yield loss of
90%, the first stations in the assembly process should be capable of producing 3800 Mouse
assemblies per day (a cycle time of 15 seconds per Mouse assembly, or a batch cycle time of 60
seconds). However, these stations should just be capable of this production rate, and must not
produce at this rate unless it is required to account for in-line losses of yield. Producing at this
higher rate would cause unneeded in-process inventory, which would negatively affect the
balancing of product between stations in the process.
Finally, planned downtime must be accounted for in any cycle time determination.
However, since we assume that the assemblies are produced with a two- shift operation, it is
expected that any required preventative maintenance could be completed during the third shift.
In the previous report, a batch cycle time of 55 seconds was calculated. This calculation
was based on empirical numbers suggested by the Boothroyd & Dewhurst design analysis
method. Revisiting these calculations, we believe that this is an overly optimistic estimation.
Four assembly steps, involving four parts, are required in the stage that was studied. In addition,
the pallet must be located and later released from this Workstation. Assuming it would take 1.5
seconds to raise the pallet off the conveyor and then locate the pallet properly, and another 1.5
seconds to lower the pallet back onto the conveyor, 52 seconds remain in which to accomplish
Page 8 of 15
2.875 – Fall 2001
Annabel Flores, James Katzen
the required assembly stages. This would require rapid accelerations and steady-state speeds,
which could result in less accurate positioning of parts in the assembly. Recalculating the
assembly time using an estimated time of 5 seconds per assembly stage (rather than 3.3 seconds,
suggested by Boothroyd & Dewhurst) yields a total time of 20 seconds per pallet per assembly
stage. In addition, a gripper change (estimated to take 5 seconds per change) is required for each
assembly stage. This adds another 20 seconds onto the cycle time. Therefore, the total cycle
time to produce a pallet of four Mouse assemblies would be 103 seconds.
Since this is quite a lot more than the required batch cycle time of between 60 and 68
seconds, additional capacity is likely needed. Constructing another production line or adding
additional production hours could achieve additional capacity. However, before the capital
investment is made, it is crucial that firm values for production volume and production rate
requirements are identified.
Part Flow and Operator Movements
Workstation 1
Since the Mouse Base is durable, easy to pick up and proper orientation of the part is not
a concern, parts can be fed into Workstation 1 in crates or boxes directly from the supplier, either
in-house or out-sourced. The Spring, on the other hand, is very difficult to handle and feed into
the Workstation therefore it is ideal to create the part in the Workstation. Thin metal rods can be
easily fed into the Workstation, can be quickly made and the part, by nature, is forgiving to
minor manufacturing errors. The Wheel, though small, is easy to handle and can be fed into the
Workstation directly from the supplier. As mentioned in an earlier report, the Spring and Wheel
are the most complicated parts to assemble and cannot be easily redesigned to facilitate
assembly. A manual operator is needed to create the Spring and Wheel Subassembly as well as
to install it into the Mouse Base. Once the operator creates the subassembly, there is no need to
create a feeding mechanism to install it into the Mouse Base as it is done directly. The size of
the Spring may require a magnifying glass be placed into the Workstation to facilitate placing the
subassembly.
In order to minimize the effect of tool changeover times, four products will be assembled
at one time. As the operator finishes the assembly steps required in the first Workstation, they
can place the Mouse Base on a part tray to present to the subsequent Workstations. As the layout
demonstrated, the operator must continually rotate within the Workstation to assemble their
components and feed the parts into the second Workstation. This rotation could prove to be a
source of repetitive stress, which is an ergonomic concern. Efforts must be made to limit the
potential for operator injury.
Workstation 2
The four-product tray can be transferred from the first Workstation to the second via a
mechanical or gravitational conveyor belt. As Workstation 2 is robotic, a trigger mechanism is
needed to allow product trays to enter the Workstation only after the robot is ready to begin a
new assembly sequence. The additional components needed at the Workstation are the Circuit
Board, the Cord, the Gears and the Mouse Cover. Pallets of parts will need to be made available
to the robot and requires a significant amount of space within the robot arm’s reach. Much of the
detail of this Workstation design was presented in the previous report, and will not be repeated
here. The same conveyor system used to send pallets into the station will be used transfer the
Page 9 of 15
2.875 – Fall 2001
Annabel Flores, James Katzen
tray onto the next Workstation after a second trigger releases the tray when the assembly process
is complete.
Workstation 3
As the trays leave the second Workstation, there is a necessary, non-value added process
each product must undergo. Each Mouse must be inverted before entering the next Workstation.
This intermediary station will receive the trays through a similar conveyor system. A robotic
arm will extend over a Mouse and gripper tool will pick and lift the Mouse. The gripper tool will
then rotate 180 degrees to place the inverted Mouse in a secondary tray. Each tray will need to
constrain the product thereby locating it accurately for assembly.
Conversely, a robotic arm can take the second tray and place it over the first, effectively
encapsulating all four products within. The entire sandwich can be inverted and lifting the first
tray will present the mice in the proper orientation for the subsequent Workstation. This method
reduces the time and number of steps required to present the inverted mice to Workstation 4.
However, this increased speed may be an unnecessary investment since this would require
additional gripping tools and a higher lifting capacity. In addition, this area is not a bottleneck
and there is no reason to accelerate the process if the products will just sit and wait for the next
Workstation to become free.
In either method, the new tray will be transferred from Workstation 3 to Workstation 4
via a conveyor system.
Workstation 4
The fourth Workstation is the final value-adding Workstation in the assembly sequence.
The Ball, Ball Holder, Screws and Stickers must be fed into the Workstation. The Balls can be
presented to the Workstation via a chute that is filled by a parts feeding hopper. Since the part
orientation is immaterial, the chute will need little specialization. The Ball Holders are simple
disks that can also be fed into the Workstation through a magazine. Because the orientation of
the part is critical, the magazine will have to be modified to prevent incorrect loading of the
components. Similar to other components, the suppliers will be responsible for creating the
magazines for assembly. Feeding the Screws into a robotic assembly has become a standard
process. A bowl feeder can store the Screws for a tube feeder to present the parts into the
Workstation. Finally, reels can be used to hold the three Stickers that can be simultaneously
installed onto each Mouse as the tray exits the Workstation. The design of the reels will have to
either store enough stickers between line shut downs or facilitate reel changeovers without
interrupting the assembly.
The robot will need to pick and place the components. The Ball’s symmetry makes
placement relatively easy. The Ball Holder, however will have to be carefully gripped to allow
the disk to fall into the Mouse Base and avoid the Ball. In addition, the robot will need to twist
the Ball Holder to lock into place. Inserting and driving the screw into the Mouse Base and
Mouse Cover is a standard process that presents no difficulties for automated assembly.
Workstation 5
Workstation 5 consists of testing and packaging the products at the end of the assembly
line. Testing each product would validate the performance of each before shipping to the
consumer. However, holding the suppliers, whether internal or external, responsible for the
quality of the components would simplify the tests needed at the end of the assembly. Because
Page 10 of 15
2.875 – Fall 2001
Annabel Flores, James Katzen
the yields of the components should be relatively high, the test on the assembly line would only
need to verify the assembly process and the interaction of the different components. Using
statistical process control would reduce the number of products that would undergo testing and
could be limited to one test per tray of products. A computer generated random list could be
developed that would statistically verify the functionality of a single batch. However, there
would be little opportunity for a visual inspection during assembly unless the robots were also
capable of verifying the physical location of the components’ interface.
The operator would need to remove the tray from the assembly line and choose a preselected product to plug it into the Test Fixture Connector Block. As described in the previous
report, this block would be able to test the functionality of the product by moving the Mouse Ball
and seeking the desired electrical performance. A computer station will be needed to record and
assess the information. While one product is testing, the operator would bundle the remaining
mice with the required documentation and software. The documentation and software are
located around the manual operator so that they would need to rotate to obtain the necessary
material. In addition, the operator would need to package the bundles for either the OEM market
or the aftermarket. The cycle time of the product may require additional operators to test, bundle
and package the mice. Because this is a manual workstation, the area can be modified relatively
easily to accommodate additional employees or to facilitate material handling for the operators.
If defects are discovered, an intermediary rework station may be added after Workstation
5 that could disassemble the mice to discover the source of the problem or send it through the
assembly line once again.
Conclusion
This report required a detailed analysis of the overall assembly sequence and plant floor layout in
order to assembly the computer Mouse. This analysis showed that there are two feasible options
for assembly, a fully manual process, and a hybrid manual and automated process. A fully
automated process was shown not to be feasible, due to product design complexities. The
manual process would use lower cost equipment, but would require more operators and yield
lower production volumes. The hybrid process developed would require substantial upfront
costs (relative to the manual process), but would yield consistent assembly results and higher
production rates. Additional work and analysis, including computer simulation of each process,
would be required in order before a final recommendation for the best assembly process can be
made. However, this analysis would likely show that the decision would ultimately depend upon
the expected production volume and required production rate, as well as economic factors, such
as labor rates, and maximum initial capital investment.
Page 11 of 15
2.875 – Fall 2001
Annabel Flores, James Katzen
TABLE OF FIGURES
Figure 1: The "Mystery Feature" on Mouse Base that complicates necessitates reorientation of
Circuit Board during assembly. .............................................................................................. 3
Figure 2: Full Manual Assembly, Cellular Design ........................................................................ 5
TABLE OF APPENDICES
Appendix A
Appendix B
Appendix C
Page 12 of 15
2.875 – Fall 2001
Annabel Flores, James Katzen
APPENDIX A: Bill of Materials
Bill of Materials
Mouse Assembly
Mouse Base Subassembly
Mouse Base
Spring/Wheel Subassembly
Spring
Wheel
Circuit board
Gears (2)
Cord
Ball Holder Subassembly
Ball Holder
Ball
Mouse Cover
Screw
Sticker Pads (2)
Part No.
8
9
10
11
7
6
1
2
5
3
4
Page 13 of 15
2.875 – Fall 2001
Annabel Flores, James Katzen
APPENDIX B: Exploded Part View
Page 14 of 15
2.875 – Fall 2001
Annabel Flores, James Katzen
APPENDIX C: HYBRID ASSEMBLY SEQUENCE
Page 15 of 15
Download