Journal97 intro
From Simreal
Contents |
Pre-September Introduction
Though I thought about writing some kind of journal before now, I didn't start until the 25th of September. So, now I get to go through all my old receipts and notes and tell you...
The Story So Far
It was a dark and stormy night... well, actually, Oregon has had some remarkably good weather this year, so I guess that isn't really true.
Inspiration
I donut remember when it happened, or what I had for breakfast that day - but there it was. Robot Wars í97. I'm always keeping an eye out for cool stuff, and this had some promise. I looked through the pages, read about the í96 competition, the robots, the competitors. When I came across a blow-by-blow description of the contests I was hooked. That was my kind of stuff!
In the early (early!) ë80s I was a "stick jock" in the SCA - fighting sword and shield, face to face, sweating and grunting - real man stuff (for an 18 year old). Reading about the Robot Wars had that same adrenaline affect. I was hooked.
But what kind of robot did I want to make? A hard one. Something that would be a challenge. Something that would kick ass. Something really really cool. I would make a robot with legs. And then I would get extra weight allowances, too. Great! The robots with legs that competed in í96 were, to be blunt, terrible. Sorry guys. Those aren't legs, but glorified sticks on wheels. Same problem with the í97 legged robots, regardless of the wonderful design and engineering.
I wanted real legs. So I begin sketching up designs, ideas. Brainstorm on paper different shapes and morphologies. Looked at Team Delta's leg ideas. Getting closer.
Power
How do you power legs? They need to be fast - those wheeled robots are screaming all over the arena, and I need a responsive robot. They need to be strong - even with six legs, on a 300lb robot, standing on three legs they need to hold 100lbs each... or standing on 4, well, you do the math. They need to be robust - when I am rammed by a 175 wheeled demon, I donut want them to jam and fail. I donut want them to ever blow smoke. That is simply too embarrassing.
What does that give me? Air power. My early research into pneumatics was not promising - the pistons I found were very heavy, and would put me over limit. Then I found Air Muscles, or, the McKibben muscle. By June 11 I had purchased some surgical tubing and some small diameter bicycle tubing, some mesh ropes, and some this and that, to make trial muscles. Oh, and I had to buy an air compressor, too. Oh well, I always wanted one of those! I justified it because I could also use it on other house remodeling type projects. I'm sure Ill get back to those projects... eventually.
The theory of the muscle is that on the outside, you have a mesh of some kind that, when the center expands radially, the mesh contracts axially. I originally tried surgical tubing center with a hollow core nylon rope from the hardware store. Blam! The tube blew out - it would bubble in a weak spot and pop. So I tried the more compliant rubber bike tube. SUCCESS! The tube inflated the rope, and it all shrank. But not enough.
So I researched some more. My local Parker pneumatic supplier (Air-Oil) had some samples of a protective mesh covering from IcoRally. This was great stuff - strong, flexible, and it had a good weave angle. It worked beautifully, developing a strong pull and good contraction. I also found a lightweight Parker piston from Air-Oil and got that, to compare technologies.
Structure
About the same time as I was designing air muscles to test, I was thinking about the physical structure, the framework. I donut have any kind of metal working tools - no mill, no lathe. By September I had bought an abrasive disk cutoff saw, but until then everything was done by hacksaw. I learned to hate hacksaws.
In fact, I not only had no metal tools, I had next to no metal experience. I did a little bit of hand forging, for jollies. I had started working through the Gingery books and had cast a few pieces of aluminum. I wanted a metal shop, but didn't have one yet. I had a decent jewelry shop, because I had taken some classes. I made some body jewelry for a local piercing shop - the simple stuff, nothing with threads (you try drilling and tapping a 1/4" ball, and drilling and tapping a 12ga stainless steel wire, without decent tools). In fact, this is how I am paying for the entire project -- $100 a week at a time (with occasional influxes of other money as it comes along).
But back to the robots structure. It needed to be fairly large - I calculated that stored air wouldn't hold up to the task, so I was putting a compressor on board, and that took space (and a bit over 100 lbs weight). It needed to be light. It needed to be robust. So... an open framework of steel tubing. The 3/8" round tube is fairly light, fairly stiff, and really cheap. So on June 23rd, I sauntered down to the local Steel Shack and got a couple 21í lengths of 1/8 schedule 40. Whatever that means.
Okay, I had some nice designs with lots of triangles (triangle = strong), like those giant framework, triangle filled crane arms. I continued to create and refine leg designs up through the end of June, and I start experimenting with the tubing. Since I have an oxy/acetylene welding kit, I was going to weld it all together. Welding is fun. Everyone should try some, at lest those pyros out there.
Now, with some good designs, all I had to do was dress the ends of the cut tubes. Lessee--- tube notchers have some kind of circular cutting bit in a drill press with clamp. I could do that! Uh.. no I couldn't. Can't get those core cutting thingies in 3/8".
Okay, Ill set punch and drill with a normal drill bit. Uh, nope. Dances off the tube like a frog on a hot skillet. Should have known, but had to try anyway.
Okay, Ill get an end mill and use my cross slide clamp to mill the end on the drill press. Almost! But no cookie. The chatter is unbelievable, even clamping close and tight, at any speed my drill press possesses. And then the chuck flies (er, wobbles) off. Agh! Okay, it wasn't on tight. Floop! Uh, again? Floop! Okay, this isn't working. Neither did the 3/8" cutting bur.
Square tubing. Yeah - modify the designs a bit, get rid of the compound angles, and square tubing is great. Oh, but it doesn't come in 3/8". Okay, 1/2" is fine. Back to the drawing board.
By the end of July, I have redesigned the first leg, painfully cut it all out by hand with the evil hacksaw (actually it's a nice hacksaw, but it doesn't have a motor on it, so I hate it), and weld it all together. I make the joints like a door hinge, long and sturdy. I keep the accuracy of the hinges by placing rods into the hinge tubes when welding any of it, and yup, with a little grease afterwards, the joints work wonderfully. I have a fancy knee with a tie bar up to the hip, to keep the lower leg at a constant angle to the ground as the thigh moves (not necessary, but neat-o).
Somewhere in there, I come into some extra money ($1,200! Heaven!) and buy an oscilloscope (for the electronics down the road), some electronics training kits, some books, and a MicroBench starter kit with all the trimmings. I also get lots more metal, more of the woven stuff from IcoRally for muscles of different diameters (I'm so sure they will work right), a filter element for my air, a nice Parker valve and fittings, and a stack of electronics stuff from various sources. Oh, and a soldering iron. Mine was getting a bit old, carried forward from somewhere in the ë70s. Soldering irons had gotten expensive! So much for the money...
Brains
Right up front, I knew this robot was going to need extensive electronic and computer support. Software is no problem - I've been writing code of many different types since before I was interested in girls. In fact, I found puberty to be a terrible nuisance and it interfered with other existing projects. I worked with electronics and robots back then - but haven't touched a chip in fifteen years!
Time to study. So I bought books, "learning" packs, and spent my spare time at night wading through the materials at hand and getting a refresher course. Well, actually, I did all this while watching TV. Sometime in August, however, I stopped even watching TV except for rare occasions... scary. But my wife still knows my name, and the kids recognize me, so I guess I'm doing okay.
About electronics learning packs. They are okay. But you are better off getting all the Forrest M. Mimms books, or some other beginner project book, and a stack of discrete components. Cheaper, and better for future reference. I got soaked on educational stuff.
Then I started cruising the ënet in search of a nice little computer. Ouch! Things have changed since 1979! There are tiny computers of every imaginable size, shape, and capacity - at any of a dozen price levels. I became instantly confused. So I bought a really nice packaged system through MondoTronics (a good source for many things), built by MicroVention. It has a PIC16-something-or-other chip, a nice "control panel" plug-in, a nice proto-board plug-in (to supplement my existing proto-board), and nice support software. It was all very nice.
I also got some discrete logic chips, a few digital power MOSFETs, and started working on a valve controller. My first one was run by a dinky little set of switches simulating TTL input. I made a nice logic circuit to prevent me turning on both sides of my 3 position valve at once, and then I tested it on the valve. Nice! It worked!
I also did some programming on the MicroBench. Made it twiddle LEDs, listen to switches. Timed stuff. Cool! It worked too!
Testing it Together
By now it was into August! I assembled my initial test leg. Greased the sticky knee joint. I dropped some muscles into the knee to drive the lower leg. First discovery: wire cable is no fun to work with. It is hard to cut, it doesn't terminate cleanly in the short space I have, and it is springy. In fact, it caused my first panic - I couldn't get a muscle into the space I had allotted! Too long! So I built a new set of muscles using the new information about wire cable. Okay, that is mostly in place.
Now to put on the piston. Hmmm... didn't design for pistons. It interferes with the structure at the hip, so I just chop out some structural members, weld in a flat piece of steel, and generally torture the hip-thigh joint into submission. Then I attach my 6" stroke, 1.5" diameter piston.
The excitement is building! I can hardly eat.
I drag my fake valve driver outside, wire it up. I plug my air compressor to the filter, the filter to the valve, the valve to the air muscles. Flip the switch, the compressor chugs up a good head of air.
I make sure the leg is firmly clamped in my 5 ton vise. I position it so it wont smash into my head or other body parts while I work the controls. I adjust the air pressure to 80lbs.
I flip the first switch. Twitch, the muscle contracts. Not enough. Too bouncy, too compliant. Not nearly as promising as the first solo tests. I flip back to the other. Twitch. I fiddle some more, but it is becoming clear that my design and these air muscles were not working well together.
Okay, I let out the air pressure with a hiss. Reconnect the valve to the piston at the hip.
Pressure, 80 psi.
Standing back, I reach out to the controls and...flip, hiss, SLAM! The leg leaps into the air and smashes into the back-stop! Wow! That was fast! Flip again, hiss, SLAM! The leg smashes down to the bottom stop.
Whoa.
I could pull, oh, five complete cycles a second at that speed! Much better! I call everyone in to see. SLAM! SLAM! SLAM! Yeah, now I'm cookiní with gasoline! I feel the power reach its tendrils into my skull. SLAM! SLAM! Bwah-hah-ha-haaaa! SLAM! Hmmm, that looked a bit odd. Oh well. "Nikolas! Come see my new leg!" SLAM! SLAM! Hmmm, definitely wobbly.
I disconnect everything. I grab the leg before loosening the vise - and it wobbles. That's wrong. I move it around a bit, and the vise is loose. Yeah, that's it. Check the bolts. Nope. Ack! It broke my new 5 ton vice! There is the crack, right in the base.
Fortunately, the store I got it from has a nice return policy (sorry, guys). I felt bad, so I bought up to their 6 ton vice.
I think this is going to work, but I have got to get that leg under control.
Feedback and Computers
I donut know when this is, probably still early August.
I did a hack job on the thigh joint and attached a 10K potentiometer to it. I had experimented a bit with the A/D on the computer, and it seemed really nice.
I then sat down and discovered I had no idea what my program was supposed to do! So I designed a bit, and came up with a nice test program. I had to call the factory at one point - could I cycle their valve kinda fast, you know, 30 Hz maybe, to get some kind of PWM control of air flow? They must have thought I was nuts, but didn't see why not. I checked specs some more, discovered that the timers on the PIC were all exactly the wrong speed to be useful at the range I wanted to work in, and improvised.
Then, I hooked it up to my fake TTL interface to the valve, and surprise! It worked! I had a few switches this time - set upper limit, set lower limit, go up, do down, and something else. Oh yeah, speed control, which changed the pulse spacing and duration. I tested and tried everything on the bench, and heard the gratifying clicks from the valve, and flickers from the signal LEDs. Time to go outside again.
I hook the pot up to the thigh joint. A bit loose, but good enough. Electricity, wires, computer, switches.... Air. Okay. Raise the leg, set upper limit. Lower leg, set lower limit. Set speed to slow. Hook up valve to piston. Switch - go up. Yeah! Jerky, but slow. Go down. Click-click-click-smack. Okay, missing the limits.
I set different limits, try different speeds. It all mostly works, but I get a lot of overshoots, and the motion is really rough. I need to try a faster modulation - and maybe a faster valve. Later, I research and find some really nice valve technology, but it all costs three to six times too much.
In the trials, I got really inconsistent results. Sometimes perfect, sometimes it would just stop halfway. It wasn't too hard to figure out, though - the computer, sensors, and valve all operated on the same power supply. I put capacitors and surge diodes on the valve (inductive load and all that), but it ran at what? 100, 150 ma.? Ill have to check my notes. Anyway, the power bus was getting hashed and confusing the electronics. Make a note - separate power supplies.
That was fun.
The most fun Iíd had in ages!
Time to take the muscles out, make my notes, and decide that I liked pistons. Sorry McKibben muscle.
Then, I grab my test leg and lean on it. Twist. Oops. Weak spot. Gotta go. I torture it some more, then pack up my tools and take a break.
New Designs
With my new-found knowledge, I break out the old graph paper and pencil, and start designing a new leg. I am still using the open framework design, but I eliminate the weak points. The fancy knee goes - Ill just have to compensate the leg angle with power and not design. Some non-triangulated areas go - a more robust thigh. The hip design gets reduced to ultimate simplicity. I begin a body design, to fit trial air compressor sizes.
Now, that is the real hard part. Ever try to find a gas powered air compressor that fits a mobile platform, and weighs in at about 100lbs? Try it. It doesn't exist, as far as I can tell. So I am going to build one up from components - a nice gas engine doesn't weight too much. The compressor stage I need is in the Graingerís catalog. Hmmm.... They have some nice hot-dog tanks, too. Okay!
A friend (Hi Wayne!) suggests that I take a 2-cycle chainsaw motor, and tie it to another one - use one as power, the other to compress. Or maybe a 4-piston something, and use half as a compressor stage. But I feel really conservative here - I donut know much, so I'm going to use parts designed for the task. If anyone has some good ideas, I'm more than happy to entertain them! I want about 100 to 150 psi delivered at 10cfm, preferably with two tanks, one for each side, to minimize pressure drop and line length.
The new leg has a different joint design. I will weld a 3/16" thick, 1" wide steel flange on the ends of the framework. It will be drilled to fit a bronze bushing, and then I can use a 3/8" steel rod or pin as the pivot. This can be terminated close at either side of a flange pair, or pass all the way through both flange sets in a joint. This will give me more options for internal space, look nicer, and give really smooth motion. The problem is, 3/8" bushings are hard to find in Eugene! Oh well, I can order them.
I am going with pistons all around, keeping the nice Parker 6" stroke, 1.5" bore. I donut have a valve I like yet, so I will experiment with the one I have more, and maybe order some more. They donut stock the low power variety, so it will take a week or two once I decide.
I procrastinate a lot (work has been busy, and I've been kinda burnt out). Finally I finish the design and drafting for the new legs. Since I am using this flange approach, I need to make the leg ends nest inside each other (or at least the flanges at the ends). For this, I need more math - to make it all fit! I then finish the calculations and drafting.
I buy a cutoff saw. I really hate hacksaws.
I buy more metal. And some large drill bits. And some bushings.
Mascot
I also buy a striped knee zebra tarantula and nam it Boris. I watch her (most of these bugs are female - the males donut live very long, so donut get sold by reputable dealers, unless you want one) for motion studies. She doesn't actually do very much, but when she does, it is really cool.
The project is actually named after a song I heard once -- a dopey little ditty that goes something like:
Creepy, crawly, creepy, crawly, creepy creepy, crawly crawly, creepy creepy, crawly, crawly....BORIS THE SPIDER." From Pete Townsend's rock opera "Tommy"
Building the New Leg
I am well into September now.
The first leg took something like two weeks to cut out and build. I cut out all the pieces, except for the joint flanges, in under two hours. Yeah! Love that cutoff! Good clean cuts, decent accuracy, angles up to 45 degrees. Fast. Easy. And best of all, it blows an amazing tail of sparks across the bench. Oops. Better move those papers, or the whole shop will be in flames.
I start welding them together. The thigh goes together like a dream. When I am finished, though, the ends are different sizes by about .25". Darn. I forgot to be really careful and measure the placements as I went. Can't just trust that things to be right, you have to measure everything. That's okay. The other pieces build around the thigh, so Ill just adjust as I go along. For the final legs Ill have to be more careful. And I have an idea that will force the accuracy, regardless.
The lower leg is sweet, too. I need some cross bracing in some places for the final, but Ill skip that now. I want to test again this month.
Hmmm... hip, easy. Body can wait for later.
I cut out all the pivot flanges now. The cutoff saw starts it, but I am taking a deep notch out of the backside, to make a tab for alignment and welding. That I cut by hacksaw. Darn. Oh well, I have some nice new blades at least.
