Part 1 of 2:
During the development of an all terain robot project, as with most robot projects, specialized mechanical parts had to be fabricated. One pair of these parts is the inside and outside protective wheel hubs. The parts are shaped like a pie tin with a 4"OD and ~3/4" draw. At first glance the solutions seemed easy, machine the protective wheel hubs out of 3/4" thick 6061 plate stock on a milling machine. As the saying goes, "when you've got a hammer in your hand every problem looks like a nail." My friend, Dan, suggested looking into a process called metal forming (also called press forming or drawing). I have heard about this process and even seen it in action many times but I have not had hands on experience with it. The project to build the protective wheel hubs seemed like a fun way to begin developing that 'hands on' experience while learning a lot along the way. During this project additional input came from friends Ross and Jay and my hat is off to them for the invaluable insights and technical pointers for information and materials. In the beginning of the project and testing phases I purposefully overlooked some of their input to gain a better understand of the material and process failure modes in an attempt to better understand the journey.
Part 1 of this article delineates just a few of the steps and test performed on the way to the design of a final tool-process-material schema to produce the desired part. Part 2 of this article will follow in a few months with detailed information the production of the forming tool, results, and lessons learned. The article is structured as a photo essay with comments. There are many, many samples and tests not depicted here, as I have chosen a subset that I feel adequately represents the work at hand.
Note: Many of the tools, processes and materials depicted in this article can cause sever bodily harm if used incorrectly or by those inexperienced with their operation. Make sure to read all warning labels and adhere to tool usage instruction. Proper safety equipment, face shields, welding gloves, etc must be employed. Of most importance, seek the advice of a seasoned machinist or someone knowledgeable in the operation of these tools to avoid the risk of tool damage or personal injury.
The wheel assembly required for the robot is shown in this first CAD rendering. The parts that this article will focus on are to the bright blue wheel hubs to the far right and left of the assembly.
Before we start take a quick look at the eventual outcome of the planned work outlined on this page.
These inside and outside wheel hubs provide two functions. The inside wheel hubs protect the electrical connections to the back of the motor and the optical encoder from damage cause by mechanical collision with rocks, curbs, etc. The outside wheel hubs protect the gearbox, but also server to connect the output stage of the gearbox to the drive wheel assembly. The output gear is attached to the protective wheel hub by the smaller 6-position bolthole pattern in the middle of the plate.
Although there are a lot of photos indicating physical work here, a large investment in project time came in the form of materials research. I have collected data from many sources relating to many materials (far beyond that listed below) however the concise delineation presented below covers most of the materials that an amateur robotics enthusiast is likely to use.
|Material||Alloy composition||% elongation for 2"
||Aluminum 1100-0||99%Al||60% elongation
||Aluminum 2014-T6||90%Al - 4.4%Cu - 1%Mg, Mn, Si||13% elongation
||Aluminum 3003-0||98%Al - 0.12%Cu - 1.2%Mn||30% elongation
||Aluminum 5052-0||97%Al - 2.5%Mg - 0.25%Cr||25% elongation
||Aluminum 6061-T6||1%Mg - 0.6%Si - 0.28%Cu, Cr||17% elongation
||Aluminum 7075-T6||90%Al - 1.6%Cu - 2.5%Mg||11% elongation
||Commercial Brass||90%Cu - 10%Zn||45% elongation
||Red Brass||80%Cu - 20%Zn||50% elongation
||Yellow Brass||65%Cu - 35%Zn||64% elongation
||Free Cutting Brass||61%Cu - 35%Zn - 3%Pb||60% elongation
||Phosphor Bronze||91%Cu - 8%Sn||65% elongation
||Manganese Bronze||89%Cu - 11%Mn||40% elongation
||Nickel Silver (coins)||70%Cu - 5%Zn - 25%Ni||45% elongation
||Steel (low carbon)||98%Fe - 0.3%C - 1%Mn, Si, Cu||20% elongation
||- Martensitic: 400 - 500||92%Fe - 1%C - 10%Cr||10% elongation
||- Ferritic: 405,430,446||20%Cr - 0.2%C - 1.5%Mn||20% elongation
||- Austenitic: 201,301||18%Cr - 0.1%C - 8%Ni||68% elongation
||- 302,304,310,321||26%Cr - 0.03%C - 22%Ni||50% elongation
Specimens are tested for elongation due to tensional forces in a test set-up configured as that depicted below. During a typical elongation for materials with good elongation characteristics like brass, stainless steel or 1100 series aluminum, the material will narrow by 70% during the drawing process. The actual length of material draw is determined through a projection of the slope of the line through the linear region on the stress vs. strain to the right by 4%.
In addition to the print references at the end of this article, there are numerous resources available on the Internet for reading about materials strength testing. The two images referenced above come from one such web site.
The image right depicts the author's 20-ton metal working press. This medium-light duty hydraulic press uses a 4.5" throw bottle jack and some pretty heavy gauge "C" channel steel to apply pressure.
The reader may be wondering how one uses a hydraulic press to apply tensile forces to a material sample rather than compressive forces? That is a good question, and the answer is that the tensile force comes from the configuration of the setup fixtures placed into the press. The image 2 frames below highlights the area of material in direct tension as well as those in tension from material draw.
Only one of the materials used in the exploration outlined on this page comes in a convenient size for use. The remainder of the materials comes in the form of sheet stock that must be cut into appropriate sample sizes prior to working. Pictured here is the author's 4-ton hand notcher used to make short work of this process in operation.
One of the variables that did not at first seem particularly important was sample size. As it turns out, based on visual observation of the draw characteristics of materials under test, this may have been a bad initial assessment; however initial samples were eye-balled between 5" and 6" square.
The first samples will be prepared and then set up on a 20-ton hydraulic press in the configuration shown, right. Later changes will be made to this set up to modify various parameters during the materials test.
Materials Samples Testing:
The first forming attempts are made with a 20 ton hydraulic press by merely forcing the press mandrel into an open ring of sacrificial material with the test sample in-between.
6061T652-Aluminum extrusion 0.125" Other samples of this material had similar (expected) failure modes.
5052T6-Aluminum sheet stock 0.060" (Questionable lineage)
The first two trials in 5052 come from material of questionable pedigree. The later replacement from Storm Steel, just south of town has known lineage and the weights, surface characteristics, and drawing characteristics are similar.
5052T6-Aluminum sheet stock 0.060" (questionable lineage)
This sample, and a few others through out the experimentation phase of the project, has had the raw sheet stock trimmed to a geometric shape with 8,12,16 or more sides approximating the shape of the form (round). An interesting note, even with the edges of the raw sheet stock trimmed the drawing characteristics of the material still left four distinct quadrants of folds, with a round form and a round male mandrel.
The observation here is that the crevasse formed by material drawing process always seems to line up in the quadrants inline with the shortest stock. Cutting the raw stock to even out the material did nothing to change the observed characteristic.
5052T6-Aluminum sheet stock 0.060" Known good material.
In the photo the reader should observe that the creases from the material draw propagate up the side of the part into the elongation region, which unfortunately, is a portion of the finished part. After the next material type, this defect will be addressed by changing the mechanical set up of the test configuration.
304 Stainless Steel #18 gauge
Although this test sample may appear to be a catastrophic failure on the surface, it does highlight the fact that this sample has a much deeper draw than those processed prior and it did not tear. Originally, stainless steel was ruled out as a material choice candidate for weight considerations. This promising test result will however cause the author to re-evaluate and perform strength vs thickness calculation to see how thin the material can be while still performing with adequate tensile strength overhead after forming.
Sooo... After looking at the image above, it became quite clear (as if the images further above were not enough) that some sort of top clamping plate was required to keep the material from wrapping up. A clamping plate was added to keep the material from wrapping up to far around the mandrel. The forces applied by the clamping plate had to be strong enough to keep the material from warping and not so strong as to keep the material from drawing.
During the design process for the final form tool consideration will need to be paid such that an even, consistent clamping pressure is applied between samples and the associated opening / closing of the tool. This will also require compensation for variable material thickness, which is quite variable (with-in a tolerance range) across most aluminum sheet material.
This image shows the application of the plan, above.
And into the press it goes...
Two different views of this image are included to help the reader more clearly visualize the work at hand. Of note in these two images is the excessive gap between the top and bottom forming rings. The Kant-Twist clamps in this photo are 2.5" clamps and were fully tightened before the forming process began. The pressures exerted by the drawing forces over pressured these clamps forcing them to spread nearly 3/8".
Photo #2 depicting clamps in overpressure condition.
304 Stainless Steel #18 gauge
The next move of course is to apply a greater number and beefier style of Kant-Twist clamp. Those shown in the photo, right, are 5" Kant Twist's with a substantially heavier wall cross section
304 Stainless Steel #18 gauge
This piece is approaching the desired shape. The depressed portion needs to proceed deeper. Additionally, I notice that with this configuration, no matter how well I center up the fixture prior to starting the pressing operation, it will grow one direction more than the other, "walking off center".
The clamping pressure issue has been correctly addressed here, however a new question now lingers in the back of my mind. How is this affecting the draw characteristics of the material, which clearly needed to pull though the clamping region? It would seem that some sort of slot formation is required in the final tool to address this problem, giving maximum clamping pressure to a slot around the material allowing the material to draw but not deform vertically.
304 Stainless Steel #18 gauge
Note that the creases formed by the drawing effect stop completely outside the clamping rings (compression band) for the stainless steel samples and for the remaining aluminum samples, the creases caused by the drawing effects are nearly obscured.
My friend Jay suggested annealing the material prior to press forming it. From the very beginning of the project I had looked into purchasing annealed material (-T0, or H32) but it just isn't available here in this small town. What I didn't understand was how simple the process of annealing really was. This first attempt is at an inadequate temperature, but without the proper set-up to generate the appropriate temperatures, it was at least, worth a try. When I get the chance I'll grab a torch and generate some "proper" heat levels. Results, as expected:
5052-T6 0.060" not even close to an annealing attempt.
So it is time to pull out the propane heating equipment. Yep, this is about as cheap as one can go, but for $40 it works and works, repeatably.
The Tempilsticks, pictured, right, are quite a simple way to monitor accurately the temperature of the material one is heating. Rather than hoping for a good seat on a thermocouple, or even if a thermocouple is not handy, drop a bit of Tempilstick on the metal you are heating. It will melt from chunky to liquid with-in 1% of the temperature rating on the stick. Tempilstiks come in a paint form for specific mixtures. They can be ordered one at a time or in kits such as this. There are also smaller kits for just high temperature, low temperature, etc...
5052-T6 0.060" Annealed for ~4 minutes at 600-deg-F
The failure mode hides a bit in the picture, but the split material shows around the bottom edge of the crest.
5052-T6 0.060" Annealed for ~4 minutes at 700-deg-F
Note in this attempt the clamping force on the upper ring was lessened to encourage material draw vs. tearing due to the weakened material structure and shortened crystal growth. Alas, to no avail. This picture is representative of a several samples tired at different clamping tensions...
6061-T6 0.125" Annealed for ~6 minutes at 650-deg-F
I did not particularly expect this trial to be productive but gave it a go anyway, given the inadequate heating applied to this piece.
5052 0.060" annealed for ~4 minutes at 750~780-deg-F
Failure modes are highly evident when the material looses most of its strength. Right above 770-deg-F the surface characteristic of the material changes drastically with discoloration, texture and even minor bubbling. The material became visibly weak and bent under its own weight. These effects are clearly visible in the photo to the right.
Notice also that in previous failure modes tears, even those that did not follow the male pattern form are unidirectional and uniform. In this photo the tear can clearly be seen to be jagged and change directions many times.
Leading up to this point in the project many more samples and materials than are depicted here have been run in varying test configurations. The materials seem to fail reliably with-in 10% of the value given in the table at the beginning of this document, which is encouraging. The only material that did not perform as expected was the 1100 series aluminum. The material sample had to be special ordered by a mail-order supplier and the sheet that was delivered is suspect to not be 1100 by lack of marking, bend characteristics and draw depth failure.
The Form Tool:
The last portion of this article will review the go forward plan for the eventual realization of a form tool that predictably reproduces quality parts time after time:
The next graphic identifies two concepts for the design of the press form tool. If the reader compares them closely, the difference will reveal itself through the application of the pinch band. The possibilities enumerate between a constant clamped pinch band, and one that is integrated into the male pattern of the press tool. Note, the one on the right uses the pressure of the press to apply the pinch band pressure, where the concept on the left relies on the clamping pressure of the pinch band.
The full implementation / CAD model is illustrated to the right.
Close inspection of the CAD model representation, will show that concept #1 (left from the illustration above) with a slight / small modification was choosen for implementation. If the reader looks at the middle press / pinch plate, in red, they can see that a 0.050" raised boss has been added to cover the pinch band and apply pressure only to that area (and slightly beyond) to keep the wrinkling due to draw outside of the part.
Since the forming tool will need to apply a substantial amount of pressure to the retaining ring / pinch band combination, 8 socket head cap screws were selected to engage the ring. Washers are used under the heads of the socket head cap screws to keep binding with the form tool material to a minimum and allow compression to an even amount. To aid in the even pressure and repeatability, Jergen's steel thread locks have been designed into the bottom plate, so that there will be no binding upon repeated screw insertion / removal. Add a tiny bit of light oil, and a micrometer torque wrench and consistent / repeatable clamping pressure will be applied each time. Further this keeps the problems with flaking and material shedding with high torque screws in raw aluminum without coatings from impacting consistency.
The large 1/2" diameter alignment pins and oil impregnated bronze bushings are hardware features that are meant to keep the tool aligned during pressing, insuring an even part each time. Additionally, the tool will be used to locate the through holes in the part while the formed material is still clamped in under the socket head cap screws. The multi-function aspect of the forming tool will allow the part + tool to be moved to a simple drill press for a 2'nd operation to put in perfectly aligned holes with zero set=up overhead. The decision to make this a second operation and drill the holes rather than to punch them right on the press at the same time was made in an attempt to keep the complexity of the tool down. In this example I am only making 24 parts and it is a hobby project, so there is no need for a mass production tool that automatically punches the holes at the same time as forming the part.
A additional clamping ring (not picture) is installed in the tool stack up for 3'rd stage operations. These operations involve moving the form tool and part over to the mill to cut the outside profile onto the part after pressing and 2'nd operations are complete. In between these steps the part can be removed and re-inserted into the tool after the alignment holes have been added to the part profile. This facilitates running batch operations before moving & set-up overhead.
And here are the components used in the CAD model, above to build the compression tool. Take particular note of the Jergens steel thread inserts. This will allow the insertion and tightening of stainless steel socket head cap screws into aluminum and repeated removal without crushing galling or wearing out the threads.
The project has been quite educational leading up to this point, and it is the authors' sincere wish that in this writing some of this knowledge can be distributed out to the amateur robotics enthusiasts spread across the United States and possibly open up additional methods for robot construction that the reader may have overlooked in the past.
If the reader wishes to see some photo shots in advance of part 2 of this article or to track the rest of this robot project, feel free to drop on by Max's Little Robot Shop at the link below where regular progress updates and additional information is posted.
 Link: Mat Web Material Testing http://www.matweb.com
 Book: Machinery's Handbook, 27th Ed., Erik Oberg et al, Publisher: Industrial Press, 2004
 Book: ASM Handbook Volume 14: Forming and Forging, Publisher:ASM, 1988
 Book: Handbook of Materials Selection for Engineering Applications, George Murray, Publisher: CRC, 1997
 PDF: Tempilstiks Data Sheet
 PDF: Jergens Insert Data Sheet
 Link: Max's Little Robot Shop