Direct Tooling for Injection Molding
Winter and Spring 1996

The Need

Manufacturers have a strong desire to develop methods which will allow them to produce small quantities (1-25) of functional prototype parts in their end use material. Typical prototyping methods, while capable of providing parts incorporating many of the functional aspects desired in working prototypes, cannot provide the product characteristics associated with the end use material. The fabrication of conventional hard tooling requires a significant investment in fixed cost and considerable fabrication time, which can only be economically justified through long production runs of thousands of parts. Direct tooling, a soft tooling method in which an SLA resin based mold is used directly as a tool for subsequent molding operations, holds the most promise of all of the current soft tooling methods. Direct tooling will significantly shorten the lead time associated with developing true functional prototypes. However, the challenges associated with developing this technology to a functional level are also the most daunting.

Primary Objectives

• Demonstrate the feasibility of producing a resin based mold which will endure at least 10 injection molding cycles and produce at least 10 low density polyethylene working prototype parts. A possible secondary objective may be to produce functional prototypes using other plastic resins.
  
• Document the time, cost, procedures, successful and unsuccessful techniques and approaches used to generate the resin tool. Investigate and document the size, fit, and finish of the final prototype parts through analysis of accuracy, surface finish, and functionality.
  
• Operate the tool until failure, determine the cause of the failure, and propose methods which could be used to either improve the functionality/finished quality of the prototype parts or extend the life of the tool.
  

Completed Tasks and Results

1. Prepared VanDorn injection molding machine for use.

The mold bases from the VanDorn injection molding machine were removed from the machine to learn how the our molds would have to be designed. It was determined that the current mold bases could be used if ejection pin locations could be drilled into the ejection pin plate. After determining that our pin locations would not interfere with the locations being used by other projects, the machining on the mold base could be completed. A drawing used for the ejection pin locations is included in the Appendix.

Ejection pins (Catalog # CT9-M6, 1/4" and 1/8" diameter) and a blank sprue bushing (Catalog # A-6601, O=5/32", R=3/4") were procured from DME Company. The sprue bushing was machined for one-sided molding only. The sprue bushing currently with the VanDorn machine can be used for two-sided molding.

2. Built mold insert for VanDorn machine using SLA machine.

The first step was to define the orientation of the keychain in the mold halves and the runner system. For ease of molding, it was determined that the keychain should be oriented parallel to the mold face with the parting line being the middle of the keychain. The runner system would be a straight runner from the sprue to the keychain cavity.

The mold halves were then designed using Pro-E. The ejection pin holes and the passages for the sprue were included in the design. The files containing the Pro-E model of the molds are called solidcore.prt and solidcavity.prt and are located in the usr/people/bvanhiel/molds directory on the SGI.

This model was then prepared on Maestro and built on the SLA machine. The resin used in the build was SL-5170. The following parameters were used in building the molds:

• Aces build style
  
• 3-mil layer thickness
  
• 15 second z-waits
  
• 3 sweeps with varying blade gaps (first gap large, second gap small, third gap medium)
  

The time and costs involved in each step of making the molds are shown below.

Some observations were made from these first molds. One was that layer shifting occurred when the molds were built. The layer shifting was primarily attributed to high viscous sheer associated with the very close second blade sweep. This affected not only the outside dimensions of the mold, but also the alignment of the ejection pin holes.

Another observation was that there was some undesirable material formation between the lettering in the keychain. This resulted from the jump speed being set too fast. By slowing down the jump speed, this problem could be corrected.

Crowning also occurred in these molds. It was determined that the best way to compensate for this was to increase the z-wait time so the resin could level-off better. If there was still a problem with crowning, the molds would need to be sanded to level the mold face.

A final observation was that the ejection pin holes were designed too large. It was attempted to fill these holes with epoxy so they could be drilled, but it was not successful. Because of this and the layer shifting, it was decided to not design the ejection pin holes into the model in the future and have them machined after the molds are built.

The use of the VanDorn machine was requested by the industry members. By designing molds for the use of this machine, the results would be more effective for them. A disadvantage of using this machine is the lack of knowledge of the students on using this machine. In order for the students to become familiar with the operation of such a machine, a great deal of training would be needed.

3. Built mold for desktop injection molding machine.

The mold design for this machine was similar to that for the VanDorn machine. The keychain was centered on the mold face and oriented in the same fashion. The difference was the overall dimensions of the mold block. For the desktop injection molding machine the mold dimensions are 5 inches wide, 4 inches high, and 5/8 inch thick.

The mold halves were then designed using Pro-E. Ejection pins locations were not designed in these molds because they are not needed on this machine. The holes needed for the bolts that attach the molds to the machine were machined. The files containing the Pro-E model of the molds are called mikecore.prt and mikecavity.prt and are located in the directory usr/people/bvanhiel/molds on the SGI.

This model was then prepared on Maestro and build on the SLA machine. The resin used in the build was SL-5170. The following parameters were used in building the molds:

• Aces build style
• 3-mil layer thickness
• 15 second z-waits
• 3 sweeps with varying blade gaps (first gap large, second gap small, third gap medium)

The time and costs involved in each step of making the molds are shown below.

These molds were better than the first molds because of the lessons learned in building the first ones. Despite what was corrected in the build parameters, crowning still occurred in these molds. The crowning was removed by using sandpaper to level the mold face.

One reason molds were designed for this molding machine was that this machine is easier to operate. The time needed for students to learn how to operate this machine was much less than that needed for learning about the VanDorn machine. Also, this machine could be operated without supervision once a student was trained how to use it.

Another reason was that by running parts on this smaller machine, a great deal of information could be obtained for running parts on the VanDorn machine. Because temperatures and pressures can be controlled on this machine, more can be learned about the process in a shorter amount of time. The information obtained from this machine can later be applied to the VanDorn machine.

The major disadvantage of using the desktop machine is that it is not something that is used in industry. For example, ejection pins are not needed to remove parts from the mold of this machine, whereas they are needed on the types of machines used in industry. Therefore, by using the desktop machine, not all of the issues in using direct tooling can be addressed.

4. Ran parts on desktop injection molding machine.

Using the molds designed for the desktop injection molding machine, keychains were run on this machine with low density polyethylene. The results of molding the keychains are explained below, followed by some conclusions from this experiment.

Results
First Part: Temperature 400 F, pressure 30 psi (estimated leverage on pressure is 92X, therefore estimated injection pressures of 2760 are anticipated). Heavy leakage around the top of the mold at the nozzle seal resulting in a partial part. The top halves of the mold do not appear to be 100% flush. Try to rework nozzle seat on the high side mold half.

Second Part: Temperature 400 F, pressure 30 psi. Heavy leakage again around the seal of the mold halves and the injection nozzle, resulting in a partial part about the same size as the first part. Rework nozzle interface again and try to open the runner gate with an Exacto knife to try to achieve better flow.

Third Part: Temperature 400 F, pressure 30 psi. Nozzle leakage remains. Broken piece of the mold near the top by the nozzle interface has occurred. A third partial part occurs because too small of a charge of plastic.

Fourth Part: Temperature 400 F, pressure 50 psi. Partial part (length of runner only). Not large enough plastic charge. Still nozzle leakage, so more rework of the nozzle seal on the high (moving mold half) side.

Fifth Part: Temperature 400 F, pressure 40 psi. Another partial part. Partial parts appear to be getting smaller on average. Nozzle leakage remains, try to open runner gate up to the depth of the part wall.

Sixth Part: Temperature 400 F, pressure 40 psi. No part again. Not enough plastic. Leakage obviously occurring. Try to readjust the mold half alignment. Adjust fulcrum lever to try to counteract the mechanism "lift". More sanding and grinding at the nozzle seat.

Seventh Part: Temperature 400 F, pressure 40 psi. Heavy leakage again. Close inspection of the fixed side mold half under the nozzle reveals heavy erosion has occurred sometime in subsequent molding trials. It is obvious this is where the increasing pressure loss and leakage is occurring. Halt experiments and remove mold halves.

Conclusions

The failure mode was the heavy erosion occurring at the top of the fixed half of the mold. Suspect that with the very small runner gate opening, the pressure built in the top portion of the mold looking for the "path of least resistance". Due to either the actual presence of a crack from poor mold half mating or poor nozzle seating on the fixed mold half size, or a combination of both, the hot plastic charge escaped through this opening and enlarged it in the process. This would explain the decreasing size of the parts yielded, although it was not noticed during the testing. It is not clear from the "enlarging" or "erosion" process whether actual pieces of the epoxy mold were torn and carried away by the escaping plastic stream or the mold actually melted at the interface edge and was carried as a commingled gel with the escaping plastic.

Some of the failure characteristics are particular to, or aggravated by, the way the desktop injection molding machine is designed to operate. In other words, a similar type of failure might not occur in a production injection molding machine where tolerances would be tighter and leakage is less likely. However, it is clear that the runner gate should be larger for direct tooling epoxy molds than for a normal metal tool. A larger runner gate will increase the flow of plastic into the mold at lower pressures. Other suggestions that would increase the flow of plastic include using a fan gate and designing vent channels for the air in the mold to escape.

In the future the face mate between the mold halves and the nozzle is a critical design parameter for the desktop injection molding machine and must be taken into greater consideration. One method to achieve this is to design a tapered alignment pin (pyramid shaped) in the epoxy mold halves, fix them relative to one another via the aligning pin, and machine the nozzle seat of the clamped and aligned mold halves. Another method is to bore oversize fastener holes in the mold halves, allowing adjustments when attaching the mold halves to each side of the machine. The use of an alignment pin designed into the mold will aid in this alignment.

5. Built mold insert using SLA machine and back-fill with aluminum-filled epoxy.

The mold design was similar to that for the first molds built. The keychain was centered on the mold face and oriented in the same fashion. The mold size was the same. The difference was the back of the mold was hollowed out. This would allow for the back of the mold to be back-filled with another material, such as aluminum-filled epoxy. The reason this type of design was considered was because it would require less resin and less build time to make the molds.

The next step was to design the mold halves using Pro-E. Ejection pins holes and sprue passages were not designed in these molds based on what was learned before from building the other molds. However, the runner system was designed into these molds. Any characteristics, such as ejection pin holes and sprue passages, would be machined.

These molds have not been build on the SLA machine, but have been partially designed on Pro-E. The reason these molds were not build was because there was not enough time to run parts with these molds. However, if this option is still being considered in the fall, the mold designs can be completed and built on the SLA machine.

Future Work Directions

Based on the work done above and at other places, the use of direct tooling for injection molding has a great deal of promise. It allows for molds to be produced for prototype injection molding in a relatively short amount of time. The design of molds for full-size injection molding machines should be continued and tested on the VanDorn injection molding machine. Back-filling molds with aluminum-filled epoxy or another material should be pursued because it will decrease the amount of resin needed to build the molds and reduce the build time on the SLA machine.

Project Co-Leaders/Members
Michael Harrington, Graduate Research Assistant, RPMI (MSM/CIMS, June 1996)
David Hartkopf, Graduate Research Assistant, RPMI (MSIE/CIMS, December 1996)

Primary Project Advisors
Allen Brand, Process Technician Specialist, Motorola (RPMI Founding Member)
Thomas Graver, Director of Operations - RPMI, Georgia Institute of Technology

Secondary Project Advisors/Information and Technical Resources
Paul Hafner, Senior Applications Engineer, 3D Systems (RPMI Founding Member)
Marshall Barrash, The Coca-Cola Company (RPMI Founding Member)
Reggie Ponder, Laboratory Manager - RPMI, Georgia Institute of Technology
Dr. Colton, Professor of Mechanical Engineering, Georgia Institute of Technology