---

This chopper
kicks your (editted).

---

1. Concept

While most 340-2 design projects in the past yielded functional or kitschy ornamental items, we wanted to create something that was challenging, fun, functional, and that looked distinctive. Most designs for ultra-short-timeframe projects such as this are created from protrusions and intersections of primitives, but we wanted to do something that was more organic in nature that also incorporated key elements of assemblability, fine features, and moving parts. A fine-scale model fit the bill, and everybody likes vehicles, so we went with a helicopter, which has at most two moving blades but also several fine features and a very organic shape. Primary design goals were that it had to:

• Be incredibly good-looking (modeling phase)
• Fit together (design, assembly phase)
• Allow blade rotation (design phase)
• Be impervious to surface sinks due to shrinkage (design, molding phase)
• Have excellent surface finish (machining phase)

Here are some of the initial design sketches:

2. Modeling

Modeling on the workstation was done in a free sculpting style, where the final shape was formed over several iterations. The intent was to have good weight distribution so that the fuselage would sit well upon three wheels when the model was assembled. In the modeling stage, the fuselage took on sort of bulbous form, which has sort of an endearing quality to it. The blade section would be a simple hub with cylindrical protrusions that taper into airfoil protrusions for the blades.

Rough blend model:

Final design schematics shown below. The fuselage is a 4-layer smooth parallel blend of 1/8" average thickness with semicircular and rectangular protrusions and cuts for detail. The prop blade cross-section is an inaccuracy on our part - helicopters actually have symmetrical airfoils - but an asymmetrical airfoil will reduce machining time. We would have loved to add a rear prop, but 3 pieces total with one rotating part is more than enough for us to handle. Besides, lots of choppers are designed to maintain stability without that rear stabilizing blade, so this must be one of those... We will drill the shaft's 'clip nubs' by hand using the tip of a 1/32" drill bit.

Final assembly scheme:

3. Mold design

Molds were designed in the usual subtractive way. Models were aligned to datums and cut out of the mold parts. The finished molds (3.5" square, ~1.5" thick) to be CNC-machined:

4. Machining

After considerable wasted (sort of) work making very inefficient cut paths with too many tool changes, we redefined our tool paths and cutting styles with much-needed help from Milos. The things to keep in mind:

• End mills have a higher surface-to-volume ratio, hence they remove material more efficiently than ball mills.
• Spiral cuts work well on nearly spherical volumes (i.e., bulky volumes created via blended surfaces), but not particularly well with long troughs or fine-featured volumes with several regions. Using spiral cuts on the latter is very dangerous because the software is not very intelligent and will introduce a ridiculous number of superfluous time-wasting cuts.
• Step-over and step-depth are very important parameters. Step-over determines your lateral resolution. If you specify a very large step, you will get large ledges and not very good resolution. This will also stress the tool more because it must remove more material per step. Using larger steps will speed up the process, but will kill resolution, so you must tailor both to your particular needs.
• We have found that the surface milling mode (rather than volume mill) is the easiest and most efficient. However, the material to be removed must be sufficiently thin, so it is best to use this type of cut on shallow relief features no thicker than the radius of the tool, or use surface mill to clean out volume mills (rather than use inefficient local milling operations).
• If you have a feature that is exactly the tool diameter, you need to reduce tool diameter by a couple thousandths (~0.002) or else Pro/E will not cut the feature.
• If you're like us and anal about nice surface finish, you'll be filing, dremeling, and polishing till the cows come home. The NC machine gets you close, but it's still raw manual labor that yields the finished product.
• Efficiency is key. In the real world, time buys you money. In our microcosm, time buys us less pissed-off TA's.

Here are our optimized tool paths. Each mold part has at least two sequences; the chopper bodies had the most. We have optimized each tool path sequence considerably to save the most time, and we tried to minimize tool changes. The fuselage mold takes the longest time at 70min, while the blade hub mold takes less than 5min. All four mold halves can be cut by the CNC mill in under 2.5 hours.

Sample cut path for the bodies_left mold:

Mid-NC_Check:

Sadly, the molds did not turn out exactly as we had designed them. There were a series of debilitating complications, the first of which was an alignment issue. Upon creating the molds, we had trusted the intent manager to perfectly align the mold to the datums. While they looked symmetrical in Pro/E, all molds were offset by 0.05" in one axis. Consequently, the mold halves had an effective offset of 0.1". We had to redesign the molds and cut two extra right molds to correct for this, costing us several days. We saved on raw materials cost by putting one of the new blades_right on the opposite side of the bodies_right mold. Problems such as this that one tries very hard to prevent during the painstaking design process will always bite one back very hard in the end. The moral of this story? Never trust the intent manager!

Unfortunately, this was not our only problem. In the flurry of activity during the last week of machining, Milos had removed several fine features from our manufacturing assembly, including the all-important alignment pins (!). The chopper would still be an assemblable model, but unfortunately no longer a convenient snap-together kit. Among the features removed were the rear wheel cores, making the rear support a continuous cylinder rather than two discrete wheels, and the rotor shaft core, which were simply unnecessary. A further problem was gouging, as some of the molds had been run at higher speeds than we had intended using old shoddy blades, resulting in slight voidage in a few areas in the mold that would cause some imperfections in surface finish on the final product. After machining, we did considerable work cleaning out the molds, rounding out features, and polishing surfaces using a trusty Dremel^® tool. A comprehensive polishing step was crucial for obtaining the superior surface finish we desired. Given all the complications, the end result was quite satisfactory.

5. Injection Molding

The parts were to be injected in a Cincinnati single-screw injection molder. The unit is shown below:

The molds were set inside two cavities inside the injection molder. During injection, a hydraulic ram presses the two plates (each holding one mold half) together. The melt is injected through the right plate and flows down the channel in between the two plates into the mold. The left plate with the bodies_left mold inserted is shown below:

We injected the blades with a pressure of roughly 600psi, cooling time of 8s, with a shot size of 0.75" to avoid flash. All other time settings were defaulted to 6s. The parts mated well and we obtained very good-looking blades with little shrinkage. Extraction was slightly difficult if we used both mold halves due to some imperfections/burrs that we simply could not remove; using the top mold half resulted in a thinner hub that turned more easily on the spindle anyway, so our 'production' runs used only the blades_left mold.

The bodies were a little more difficult to work with due to the non-optimally-placed sprue channel that fills the chopper from the tail (not our fault). The result was some shrinkage at the nose due to lack of pack pressure. We had originally intended for the sprue channel to come from the other direction and inject the molds from just behind the blade shaft. Nevertheless, we were able to work around this by injecting at very high pressure (800-900psi) with equally high pack/hold pressure. To minimize the problematic nose shrinkage, we extended the cooling time to between 20s and 25s. The results were quite good with these settings and the gap between fuselage halves was reduced to roughly 0.2mm, quite acceptable given the conditions we designed this thing under.

The final fit'n'finish is quite excellent. For some of the fine features, the CNC mill was not able to cut as deep into the mold as it should have, which was to our benefit, since the thinner diameter of the spindle facilitates freer blade spinning. There is not much flash, better than many commercial products out there. The inside of the blade hub needed to be cleaned up considerably, as did some tiny bits of flash about the spindle but overall assembly was very easy. The model glues well only with cyanoacrylate adhesive, since polyethylene is so inert. LDPE is not the greatest material for making a fine-scale model from; perhaps if we had polystyrene available, the gluing operation would be much easier. Nevertheless, it works, it fits, and most importantly, it looks good. Don't forget, this chopper kicks your ass.

6. Metrology

Our primary goals were to create something that fit together, functioned, and looked good, and in the end that's all that matters. However, for completeness, we should evaluate how well the final product matches the initial design. There are a few key areas of the mold that were not machined to spec due to limitations of tools we were using. Wherever we could, we designed in safeguards to ensure fit even if machining was erroneous. An example of this was to maintain symmetry in the mold design so that one error is compensated by a corresponding error in the opposite part. As one can see below, there were some serious errors in machining (the tools were in awful shape), but the parts still fit together perfectly. Here is a list of critical dimensions and their errors:

Since the rotor spindle was a little short and the hub slightly fat, we abandoned the blades_right mold and injected our production units with only the blades_left mold. This resulted in a better-spinning rotor due to the smaller contact surface, as well as a better-proportioned blade. Some care must taken during assembly to avoid burrs that will block rotation. A bit of graphite powder on the spindle decreases friction immensely.

7. Epilogue

The production run consisted of a whopping 15 units with continuous production lasting roughly 45 minutes. The final units featured clip nubs (machined into the mold with a 1/32" end mill) to hold the blade permanently on the spindle, excellent surface finish, and very little shrinkage. Production units were claimed by friends and family and provided immense satisfaction to our consumers.