Deliverable 5: Closing Time

And so, after much work and strife, the semester comes to a close.







Cost Analysis



For prototyping 100 parts, our estimated total cost is $927. The cost breakdown is as follows:




set screws: $48

shaft sleeve bearings: $34

body nuts: $11

bowtie nuts: $11

polypropylene: $13

polystyrene: $20

aluminum: $70

shoulder screws: $1

bearings: $130

MATERIALS SUBTOTAL: $337

Labor for 71 man-hours at $8.25/hour : $590


COST PER YOYO: $9.27


For producing 100,000 parts, we would expect a number of variations. First of all, the cost of our outsourced parts (bearings and nuts to weigh down the bowtie) would decrease per part since we would be buying them in greater batches and could therefore negotiate lower prices in exchange for ordering a larger quantity. In addition, the costs in material of fabricating a mold would decrease per unit part since each mold would be used to manufacture a greater number of parts. Overall, our already low marginal operational and material costs would become more significant, at the expense of our high fixed cost.

The cost breakdown for full-scale production is:

set screws: $19,000

shafts: $11,220

nuts (body): $3,216

nuts (bowtie): $6,432

polypropylene: $12,600

polystyrene: $12,000

aluminum for molds: $700

shoulder screws :$30

bearings: $140,000

MATERIALS SUBTOTAL: $205,198

Labor: this is difficult to estimate, as the industrial process would likely be much different than the process we used in lab. I would expect a much greater level of automation, especially in the molding process. This, of course, would increase our mold prices considerably, but we would be able to make millions of units much more easily.

COST PER YOYO: $5.00>X>$2.05





Adaptation of Design



The 2.008 manufacturing equipment provided a series of constraints on our parts. All molds needed to be machinable on a CNC mill and lathe from Aluminum stock. The adaptation of our design to suit the manufacturing equipment is presented on a part-by-part basis.



Body: Other than changing the color, there are a couple other things we could have done to improve the body. We could have thickened the top rim to make it easier for material to flow throughout the mold better. We also could have lowered the window ledge to allow for a thicker ring and a stronger press fit.



Bow-tie: Other than optimizing for better bearings and a better body design, the only other change would possibly be putting ejector pins not on the actual bow-tie so as to lessen the chances of damaging the parts.



Window: The window was thermoformed out of a sheet of high impact polystyrene ‘HIPS.’ All sharp edges in the part were rounded to prevent discontinuities in thickness at the edges.




Ring: If made again, the ring would have been made thicker to produce a strong press fit. The thin ring deforms easily, thus weakening the yo-yo. All other aspects of the ring work well at all levels of production.





    For larger scale manufacturing, we would have done several things differently. Actual weights would be used instead of nuts in the bow ties. The ball bearings would be of higher quality. These two options would be more feasible at the larger scale since these outsourced components would be much cheaper if bought in bulk.

    More emphasis would have been placed on maximizing rate and minimizing cost at the expense quality. We would reduce the thickness of the yo-yo body in order to decrease cost and increase rate of production. This would not be worth the extra effort for small scale production, but would make a huge difference at the large scale. For the thermoformed window, we would have sacrificed quality to increase production rate by increasing the oven temperature and decreasing the heating time of our production run for each part. Although we would have a higher risk of small defects forming in the part, each part would take less time to manufacture. We could also make multiple molds so that multiple parts could be thermoformed simultaneously, which would further increase our production rate. These alterations to the design reflect our shift in priorities at the larger scale.

Constructive Recommendations for Improvement



   We appreciated the hands-on lab experience and being taught to operate heavy machinery. We also really liked having the freedom to design our own paperweights and yo-yo parts. Both Daves were really friendly and helpful towards students who were learning to use the machines and CAM software for the first time.



   The first couple of MasterCAM tutorials were difficult to follow for people who had no previous experience with this software. The pace was too fast and difficult to keep up with. Since all of the step-by-step instructions were oral, falling behind made it very difficult to catch up. It would have been extremely helpful to have a written handout detailing each step in the tutorial that could be referred to throughout the tutorial session. This would be helpful because students have various levels of experience with MasterCAM coming into the class. This way, slow students who fall behind could refer to the handout and catch up. At the same time, more experienced students could read ahead and complete tasks at their own pace, allowing Dave to progress through the steps of the tutorial at a slower pace that caters towards the new users of the software who require the most assistance.



   It would be nice to have some basic tools available in lab after hours. For assembly, our team required pliers and deburring tools for finishing operations, and an arbor press in order to assemble press-fit components. However, since these tools were all in the lab space that was locked at night, we had difficulty assembling our yo-yos outside of lab hours. A small arbor press and toolbox with pliers, files, and deburring tools in the 24-hour space right next to the lab would come in extremely handy.

    Team cabinets for storing our molds and parts were useful and convenient to have near the machining equipment. However, since they were inside the lab, they could not be accessed after lab hours. In the future, it would be nice to have team cabinets in the 24-hour space as opposed to the lab so that team equipment could be accessed at any time. There were some occasions when members of our team needed access to one of our molds in order to complete a deliverable assignment but could not do so until the following morning.

Deliverable 4

After numerous hours injection molding parts, we met as a team to measure critical dimensions and trim the sprue from each part.

 The windows were trimmed to size and stacked in order of production.  After optimizing cooling time and other thermoforming parameters, the windows were clear and free of bubbles.

The bow tie production run had some hiccups, mostly due to the parts not releasing from the core mold.  Sometimes, the ejector pins would dent or completely puncture the parts.  A spray of mold release was all it took to mitigate the problem for 30 or 40 shots. 
The press fit was slightly tighter than we had anticipated (~.003" on the radius).  Fortunately, it was not so severe as to deform or noticeably diminish the performance of the ball bearings.  However, it did require that we use an arbor press to fit the bearings.

 Body production went smoothly.  We recut the cavity mold for the bodies after noticing that our first run had significant dishing.  The dishing caused our string gap to be larger at the center of the yo-yo than at the radius.  This made the yo-yo difficult to wind.  The core mold modification fixed the problem noticeably.

A test shows that the bow tie does indeed spin:

Below is a list of dimensions on which we gathered data.  A range of data was gathered and averaged for the critical dimensions, while other dimensions are taken from a single part.  Critical dimensions appear highlighted in the table.

 Although most of our parts were not within the initially specified tolerances, most of the errors were mean shifts with relatively small standard deviation.  Ultimately, we modified the offending parts until our yo-yo became functional.  Noncritical dimensions were treated as noncritical.  As a result, they were allowed to vary as we modified more critical dimensions to fit the specifications. 



Table of Specifications
Specification	Value (with units)	Measuring Method	Measured Value	Explanation
Yo-Yo Diameter	2.500 ± 0.005 inches	Digital Caliper	2.514 inches	overestimating shrinkage during mold design
Yo-Yo Outer-Gap Diameter	2.368 +0.000	Digital Caliper
	- 0.005  inches
Yo-Yo Inner-Gap Draft Angle	2 ± 2 degrees	Digital Caliper, tangent of two sides
String Gap	0.075 ± 0.005 inches	Digital Caliper
Yo-Yo Width	1.700 ± 0.005 inches	Digital Caliper	1.604 inches	poor estimation
Yo-Yo Cavity Diameter	2.092 ± 0.005 inches	Digital Caliper	2.113 inches	overestimating shrinkage during mold design
Bow Tie Max Length	2.046 ± 0.005 inches	Digital Caliper	2.053 inches	close to spec
Bow Tie Inner Diameter	0.315 +0.000	Digital Caliper	0.309 inches	close to spec
	- 0.005  inches
Inner Shaft (for ball bearing) Width	0.197 +0.005	Digital Caliper	0.202 inches	within spec
	- 0.000  inches
ID of Ball Bearing	0.197 +0.000	Digital Caliper	0.195 inches	within spec
	- 0.005  inches
OD of Ball Bearing	0.315 +0.005	Digital Caliper	0.313 inches	close to spec
	- 0.000  inches
Yo-Yo Wall Thickness	0.1875 ± 0.005 inches	Digital Caliper	0.185 inches	close to spec
Retaining Ring Outer Diameter	2.378 +0.005	Digital Caliper	2.385 inches	close to spec - extra tight fit
	- 0.000  inches
Retaining Ring Inner Diameter	2.2185+0.000	Digital Caliper	2.135 inches	noncritical dimension varied as a result of adjusting linked critical dimension
	- 0.005  inches
Window Extrusion Outer Diameter	2.185+0.005	Digital Caliper	2.155 inches	thermoforming inaccuracy
	- 0.000  inches
Mass of Yo-Yo	0.166 pounds	Scale	0.171 pounds
Volume of Yo-Yo	4.10 inches cubed	Mass/Density
Max RPM of Yo-Yo	142.29 RPM	Tachometer sensor	2700	invalid assumption
Inertia in X and Y direction	0.0952 pounds*inch2	Calculations
Inertia in Z direction	0.1345 pounds*inch2	Calculations

Our report can be found here.

The bowtie showed a very controlled run.  The critical dimension is linear in time with the noise making up one or two thousandths of variation.  Our shift in cooling time from 30 seconds to 20 seconds is not visible in this data.

A histogram of critical measurements shows a well behaved distribution.  The specification for this measurement was 0.315" +0.000" -0.005".  Here, the mean landed slightly below the minimum accepted dimension of 0.310" that we had decided on in the beginning.  The bearing interference averaged at 0.013" rather than the specified 0.010".  This proved to be benign as the bearing did not experience too great of a compressive force and still spun freely.

The C_pk calculated for this process was .4978. 

 Although that qualifies as "wildly out of control," it is mostly a reflection of the mean shift of our data. 

Calculating C_p without considering the mean shift yields a value of .9436 - not great by industry standards but reasonable by 2.008 standards.

Deliverable 3

Thankfully, shooting two parts in one mold was successful.  We had doubts as the overall surface area of our parts was quite large.

Fro and I spent a couple of hours with the injection molding machine, adjusting parameters like extrusion temperature, rate, cooling time, and the pressure profile.  The end result was a pile of data (pictured above).

We optimized to reduce shrinkage around the critical dimensions, those being the press fits for the bearing and ballast nuts.  The nut press fit was very tolerant of changes in shot temperature and cooling time.  The most extreme values deviated by no more than three thousandths of an inch from one another.
The bearing press fit dimensions were more variable. The OD of the bearings we're using is .315", we aimed for an interference fit of ten thousandths.  Regardless of variations in parameters, the dimensions stayed very consistent.  However, we found that the cavity shrank into the shape of an ellipse.  The major diameter averaged to about 0.311" while the minor averaged to 0.307". 
This made for 0.009" and 0.013" of interference respectively.  The fit was slightly snug, but went easily with the help of an arbor press.

Later, we increased cooling time to 30 seconds in an attempt to further refine dimensional accuracy.  During our trial run with clear plastic, this did not pose any issues.  However, when we added black dye to the plastic, the results were catastrophic.  Bowties wedged themselves in the core and punctured or cracked when the ejector pins were fired.  Although the critical dimensions were preserved, the parts were not.  We observed that this was provoked by long cooling times. We solved the problem by reducing the cooling time from 30 to 20 seconds, and by using copious amounts of mold release spray on the core. 

We also had to tweak the feed stroke to maximize packing while preventing flash.  We did this by trial and error, increasing the value by 0.05 cubic inches from our initial value of 1.6 to our final value of 1.85.  We noticed flash began to appear at around 1.95 cubic inches.

The injection hold pressure profile did not require changing.  We played with the values to see their effect on the part, and found that it had none in the range to which we changed them.  They seemed uncorrelated.

One of the damaged bow ties.

The remainder of the parameters were unchanged.

A full production run, mostly successfully

Process Parameters Sheet:

Injection Hold
Injection Hold Pressure Profile P7-P16
653	600	600	600	600
600	400	376	350	350
Injection Hold Time			Z2 =	8s
Cooling time			Z4 =	20s
Set Screw Feed Stroke			C1 =	1.85 cubic in.
Injection Boost
Injection speed profile: V12-V21
6.3	6.3	6.3	6.3	6.3
6.3	6.3	6.3	6.3	6.3
Injection Boost Pressure			P6 =	1100 psf
Screw Feeding
Screw Feed Delay Time:			Z3 =	20s
Ejector
Ejector Counter			AZ =	2
Ejector Pin Lengths:	4x 5.690 in.	6x 5.570 in.	1x Sprue
Total Shim Thickness:	0.062 in.