Journal97 sep
From Simreal
Contents |
September 1997
Thu, 25 Sept 97
Finally got around to welding the pivot flanges on the leg. Had a bit of hassle getting it all to stay in place, but finally wrestled it down. Had to take my 1/2" rod down .001" to get it to fit the holes in the flanges smoothly, but still be snug, so everything stays in alignment while I weld it. Without a lathe, that meant going with a light touch on the grinder, lots of sandpaper, and some time on the polishing buff. Anyone want to donate a lathe?
While adjusting things after tack welding, I forgot the GOLDEN RULE - just because it doesn't glow, doesn't mean it's not really really hot.... <sizzle>, my fingers fail the smoke test <ouch!>. Dump the hand into my water bucket (ahhh!), run inside, squooge "Soothe-Rite" on them. Still hurts.
I finish that one joint, pound the alignment rod out of the flanges (*that* was too much work! Darn crusties on the rod), and call it a night. Now I'm cranky, with burnt finger tips.
Fri, 26 Sept 97
Woke up - my fingers are doing great! Iíll finish the other joints today, and have a complete leg articulated and ready to attach power to! Now, to get through all this real work that pays the bills...
One thing I thought about, though - the tab on the side of the flange is a dopey idea. I need to have two notches, and a central flange. This I can stuff inside the tube, it will align better, hold better, and weld easier. Yeah! Next time. And then twice the hacksaw cuts. Oh well, not everything can be easy.
I also keep discovering that cut-off saws leave razor-sharp flaps of metal attached to the edge of the piece. The slices in my fingers are healing nicely, thank you.
Sat, 27 Sept
Yesterday I welded the joint flanges on the rest of the frame. Then I cut-out and assembled the dummy body section where the leg will attach for testing.
Today I welded the joint flanges on the body section. NOW I need to get more bushings - nobody here carries them, so they are coming in from CA... by next Thursday. Oh well.
I think I will push the six bushings I do have into place, and do a trial assembly. Then I need to do a final determination of how I want to mount the pistons. Nothing like have a chunk of steel in hand to help visualization!
Instead, I spend my entire evening writing the catch-up dialog that precedes this journal. Tomorrow I have off (I work Tuesday-Saturday, so today was a work day. I find having Mondays off to be great! Everything is open), so Iíll do the bushings and assembly tomorrow.
It's 9:30 pm now, and wuss that I am, I'm getting tired.
Sun, 28 Sept
When I had welded the flanges onto the hip section, I accidentally (well, I wasn't paying attention) welded the two that go to the body backwards. No biggie! Iíll adjust as I go! Nope. Get out the hacksaw (argh!), whack them off, and re-weld.
However, I stuff my 6 bushings into half the flanges, cut some 3/8" pins, drill them, grind them, and stuff them into the holes. With my modification of the hip flanges, the spacing is off there - but it will be okay for the test. Half of the holes are, of course, 1/2" waiting for a bushing, so the whole thing is a bit loose, but it is great to finally feel it move again!
The 3/8" rod I am using for pins was hot rolled. Which means it is crap, and I have to get some nice cold rolled steel. Hot rolled stuff seems to have a poor overall shape, rough texture, and is covered with a tough, gritty oxide that is both hard to remove and messy. It seems that if you don't ask for exactly the right thing, you get exactly the wrong thing! But that is why I am making so many tests - because I generally have no idea as to what I am doing.
Now, with the leg assembly in hand, I try to get an idea of how I will place the pistons. I have the general concepts already worked out, but I didn't want to get too crazy until it was there in front of me. The body/hip junction should be easy - there is a lot of room to work with there. The hip/thigh might go high above the leg, but I am still aiming to fit it inside. It looks like I might want a 9" stroke for this one! With the mechanics the way they are, I can get a huge range of motion plus extra strength by fastening a longer piston in there.
The leg/thigh junction is funky... I think I will mount that piston high and outside the leg. Though I may try it inside with a low connection... but that puts the piston itself closer to the enemy robots.
Tomorrow I will weld in a bunch of attachment points, and check with Air-Oil on getting a piston with longer strokes.
Mon, 29 Sept
Today I was going to attach the pistons to the assembled leg. But first - I need to get sensors together to detect leg motion. So I call Images Company to try out their bend sensor - the cheapest and most direct sensor I could find for this task.
And believe me, I looked. There are nearly an infinite variety of sensors in the world, except for low cost ones. I found some excellent linear motion potentiometers, with all the correct packaging layout - for a mere $100 to $200 each. And I discovered LVDT sensors - wonderful devices - for closer to $50 each. Shaftless rotary encoders looked fairly simple to hook-up, but even these relatively simple optical devices went for $100 each! And if I took my basic $50 pneumatic piston and got a position sensing version -- $300.
So I was left with two cheap options.... Some form of shafted encoder or transducer (there is a nice 128 ppr absolute encoder in Allied for $8 each, and potentiometers are cheap and easy) that would be difficult to attach, or the Images flex sensor that should be easy to attach.
I called Images.
They were back-ordered for a month! So much for getting it installed this week.
So I went back and re-designed parts of the leg joint mechanism, cut out the pieces, and welded some parts in place. Then, bop out to the Shack, get some 10K pots, and cut down the shafts. Iíll attach them tomorrow, ëcause it's late tonight. Now I don't need those other six bushings - the sensor assembly fits that need on half the joint attachments (by rigidly fixing the pin on the outside flange), and I use a rigid mount on the other side, too, for symmetry.
Oh, and I cut out some bracket pieces for attaching the tail end of my pistons to the frame. Need to do some welding on those, too.
One step back, two steps forward! Onward into the dark...
Tue, 30 Sept
Took it easy tonight - watched "Liar, Liar" with Jim Carey. Some of his best work! Of course, some might think that isn't saying much, but I enjoy him...
Finished up the new pins for the joints - the ones I am going to attach potentiometers to. Fun fun! These are actually 3/8" SS tubes, with a roughly 1/4" inside diameter. Where the pot isn't, I place 1/4" SS rod, for strength. Had to grind the rube diameter down 1/1,000th or so, so they would run smoothly in the bushings. Almost as much fun as the hacksaw! Grind grind grind with the big wheel. Grind grind grind with the little hand-held bit. Sand sand sand so it isn't too rough. Buff buff buff until it is mostly shiny.
Are you sure you don't want to donate a lathe?
Tomorrow - piston mounts!
Oh, yeah. Every day I spend some time wandering around on the ënet trying to find that last magical missing piece that will make it all easy. Didn't find it tonight, but I followed up a lead from comp.robotics.misc that lead to Clippard, and that led to a discovery of low power (6 or 12 volt, .67 amp), high speed (5-10mS) pneumatic valves! And they are cheap ($20) too.
The only drawback is that they only push through about 0.6 SCFM... which is about 20 times less than I am using now.
The good part is - there is in integral booster ($10) that cranks it up to 6.1 SCFM, which is only half of what I am using now. I ran the numbers, and with my 1.5" piston, it will be able to perform one full 6" extension and retraction per second, with full pressure equalization and everything.
The bad part is, that the booster lowers the response from a best speed of 5mS to 13mS - so I can run the pulses a maximum of 70 or 80 times a second (unless I have my milliseconds confused again). But my previous tests were at half that, and they worked fairly well.
The end result is - it should all be plenty enough to scare the other robots, and two sets of these 3 way valves and boosters costs roughly the same ($60) as my current 4 way valve ($50), and apparently do a better job.
Of course - if anyone knows better, let me know!
