After a pow-wow with my right-hand man and sanity check, Mark Eijsermans, I’ve decided to go for the full split tower design, allowing the leg to jump back and forth in both directions. The suspension will behave quite differently when landing backwards, so this is a crucial thing to be able to test.
This is what we are building.
If the amount of time spent directly in front of the computer screen is any measure of CAD proficiency. My cat Leonard is surely a pro by now. Thanks Lenny.
It’s easy to build a machine that doesn’t break. What’s hard is to build a machine that feels sturdy and performs well under stress and dynamic loading. This is especially important when accuracy counts. A noodly CNC machine is about as useful as a broken CNC machine. That’s why I was so pleased when I finally welded up the tool holder and found it was like Gibraltar. With some expert drilling assistance from the long suffering Mark Eijsermans and a bit of thoroughly pleasant TIG welding we put the machine to the test. After a few absurdly heavy test cuts (including a straight plunge with a ½” bit!) it was clear that this structure will be amply stiff to hold the die-grinder and prevent it from chattering itself into oblivion, even under aggressive operation.
This removes one of the last remaining uncertainties in the design: structural rigidity. The last one is torque. A classic rookie mistake is not enough torque. We shall soon see whether or not the steppers have enough umph to hold this amply stiff structure firmly enough to do its work…
Project Alpha is the term coined to for the first phase of engineering development of Prosthesis. It is a 2:3 scale prototype leg mounted on a parallelogram tower that supports a platform where pilots in training will eventually sit. It will be mounted on the big trailer for mobility. The primary goals of Project Alpha are to serve as a test rig for the UBC engineering student projects, train the Prosthesis fabrication team and generally learn about making giant mechanical legs. Ultimately, if the engineering pans out, it will be put in to service as a pilot training rig – like learning to hop in a lunge position with one foot pinned to the ground.
These goals affect the design and engineering process. It is being over built and under engineered.The criteria for design are now: fast, cheap, easy to weld and sexy. This will not be a highly optimized structure. It will use thick walled mild steel, and lots of it. It will be heavy and robust and the leg will only be two meters tall, not three. This project needs to get under way fast if it’s going to be completed in time for the UBC students. The welding course I had planned for my team will now take the form of trial by fire as they are building Project Alpha. The CNC cutter will have to come together very quickly if it will be used.
Below are some basic renderings showing the overall geometry and scale compared to the full sized leg.
Last week saw a blinding execution of design-to-reality, made possible by the collaboration of UBC Fizzers, UBC Fizz Project Lab, Mark the miracle worker (maker of beautiful things from this website to the CNC cutter) and John Blake, who came out of nowhere and cranked it out all day Saturday.
I completed the design on Thursday.
The parts were cut on Friday.
And the die-grinder assembly built up on Saturday and Sunday.
We finshed the weekend with a test grinding of a piece of 3/16″ wall steel tubing by rotating the chuck by hand. No chatter, no vibration. Victory.
This weekend we attach the torch holder and hopefully reconnect the steppers…provided the new drivers come in on time.
This weekend was a solid push to cut our first CNC’s tube end. We assembled both axes and mounted the torch to the tower with clamps to simulate the actual set up. We then performed a manual cut on a 3/16″ wall steel tube by manually moving the drive belts, which proved that the basic set up is stiff enough and solid enough to hold the torch at the rigtht height and produce accurate cuts once the computer takes control. The next step was to hand over control to the computers. Thanks to the tireless efforts of Eng Phys student Evan Gillespe, we successfully moved the X-axis (the tool tower) and rotate the A-axis (the chuck) under computer control, which was a beautiful thing. Videos soon.
We then learned that if you manually turn a stepper motor when it is hooked up to an unprotected motor driver circuit, you blow the circuit because the stepper turns in to a generator. We all actually knew this, but we’re disappointed to find just how little spinning it took to blow the driver. A computer controlled cut was not to happen this weekend. Evan has returned to the UBC Eng Phys lab to spec more robust driver system and design of the tool tower continues. This picture shows the torch and die-grinder (which will be used to grind off the slag and heat affected zone after the plasma cut) mouted in their working positions…now all I have to do is connect the dots.
Engineering is a language. In the case of mechanical engineering, it is the language of machines. The laws of physics are the laws of grammar, geometry is the vocabulary, materials and mathematics are the alphabet and machines are the poetry. It is in machines that the creator is asked to shape materials in to specific geometries, within the constraints of the laws of physics to achieve a certain effect. If done well, it can be as beautiful and elegant as a piece of music or a poem. And like any other language, the more you practice the better you get. Examining and taking apart the work of other engineers is like reading what they’ve written; I read a lot as a kid. It started with Lego, that was like the Dick and Jane, the thick cardboard paged learner books. I then moved on to radio controlled cars, household appliances, and all sorts of ”forbidden works” that were never meant to be read by the consumer masses, but which contained deep, rich secrets. Once you find an appetite for it, and know what the machines are saying to you, you devour them like Harlequin Romances. Or in the case of a fine German engineered power tools, page through them like classics, careful not to miss a single word, not to miss a single message from the author who put so much care in to crafting the piece.
I’m still an avid reader. As with the written word, the messages are all around you. Some are hidden in riddles, with tiny screws, imperceptible molding part lines and ingenious concealed mechanisms. And some are like witless banner adds, warning you of how not to make your point, like cheap knock offs and the mountains of disposable consumer goods that overflow our land fills. But now I am also writer, a maker of machines, both in a professional context, where I design microscopic brain implants, and in the pursuit of my own passions like Prosthesis. The language is the same, but the stories are very different. In both cases though, the quest for elegance in design is paramount. A well designed machine will tell its own story; it will just work properly. And just like in writing, you need an idea first, then you choose a tone and an audience and begin to craft the story. In the case of Prothesis, the idea is a machine that a pilot must be able to make walk by engaging his entire body. The tone is uncompromized design, that is to say the the normal constrians of mass producability, market appeal and cost, cost, cost are not in charge. Prosthesis is a poem, not a narrative. The audience is whoever wishes to take the time to read it. Like you.
The CNC tubing cutter hit another milestone in its development last night: the headstock and spindle assembly was put together and worked beautifully! The design came from Kevin, one of the Eng Phys students on the CNC team, with the help of Berhard, the indispensable Project Lab technical supervisor at UBC. It is an unconventional combination of huge, German engineered FAG bearings from Schaeffler Canada and a clever, low profile, easy to assemble clamping structure. Schaeffler rep David Hansen is shown on the right helping to fit the precision bearings.
The spindle that supports the massive 20kg chuck is a work of art from the engineering physics machine shop and the whole assembly is optimized to take up as little of the precious travel of axial the stage (also running on a beautiful INA linear bearing from Schaeffler) as possible. The result was a silky smooth, very solid, easily aligned heavy-duty chuck, ready to hold and rotate tubes up to 3.1” in diameter while leaving both ends to be cut by the plasma torch or die grinder without having to re clamp the piece, as shown below. Brilliant!
The next step will be to re-mount the x-axis slider with the new, reinforced tool holder, assemble the tool positioning motor and drive assembly, mount the steppers, tune the drive belts and bring in the electronics to make it all dance!”
Today was an auspicious day indeed. Thanks to the hard working team of UBC engineering students helping design the hydraulics (Scott and Matt shown here with me kneeling), I now have proof positive that the mechanical feedback system I designed will actually work! This video shows the system in operation. Notice not only the power of this very scaled down systems, but the precision of control. The powered arm follows Max’s hand movement very smoothly and with feather control. If his hand stops moving, the arm stops moving. He has to keep moving his hand if he wants the arm to keep moving and in doing so. This behavior results in his hand position matching the arm position. Voila, positional feedback! There is another video of the system in the gallery where it’s operation can be seen in closer detail.
This simple prototype clears the way to proceed with this fundamental design of a totally mechanical positional feedback system. The next step is to build a more adjustable prototype in a configuration more like what will be experienced by the pilot and start to develop the ergonomics. The next prototype will also have the powered arm set up so that it moves the operators seat, which is what will happen in the real machine. You move your arm, and your whole body gets moved.
I often say the difference between Engineering and design is the math. The math and the physics (which ultimately boils down to more math) are what put the “science” into “applied science”. And nothing says math like FEA. Finite Element Analysis. It gives me shivers just to say it. FEA is the Big Math Stick. When used correctly (and there is an art to this math) it can be used to beat design uncertainty in to submission, or pound a clever design into a highly optimized solution. FEA is a brilliant method used to determine the stresses in mechanical systems that result from a given loading situation (like trying to make a 2500kg walking machine jump for example). Its brilliance lies in its simplicity. It literally breaks the immense and mind-bogglingly complex problem of three dimensional stress analysis in to hundreds of thousands of tiny, tiny little problems that can then be solved by the incredible number crunching power of modern computers. The result is a nearly supernatural insight in to your design, and how close it is to failure or perfection. The output is pretty rainbow pictures.
The primary caveat, however, is a well known computing adage, garbage in: garbage out. If you don’t have numbers that accurately represent real-world conditions (like just how many G’s do the hips of a 2500kg walking machine feel when it lands at 30km/hr?) and a well constructed mathematical model of your parts, then you won’t end up with answers that represent real-world solutions. So you still have to be smart, and you definitely need someone who knows FEA at the wheel, otherwise all you’ll end up with is a bunch of pretty rainbow pictures…and broken parts. Lucky for me, I have that guy, Tony Hewett. He’s and old friend from my Eng Phys days at UBC and he’s in Germany doing FEA for Nokia (ever notice how tough their phones are?). The picture above is the first of many beautiful such images and it shows the stress state of a simplified version of thigh of Prosthesis under worst case impact conditions. As the project progresses, and the design is refined, the pictures will look more and more like real parts and we will cover all sorts of scenarios that are impractical or impossible to test, like impact scenarios and the effect of fatigue over time. The results will come in the form of more pretty rainbow pictures, but these ones will have more than just gold at the end, they will have the truth. Thanks math.
All contents © Copyright 2010 Jonathan Tippett. All rights reserved.
