http://www1.atwiki.com/gingery_machines/ gingery_machines @Wiki description ... (under development orz) en 2006-05-13T08:53:32+09:00 http://www1.atwiki.com/gingery_machines/pages/195.html Netier XL1000 thin client This page will document the steps necessary to get Linux running on the Netier XL1000 platform, or at least how I did it. I thought these might make a good CNC/robotics platform because: no moving parts, rugged, spookily silent standard PC hardware including USB, parport, ethernet, IDE, VGA, PS/2, and serial ports (wow!) AMD K6 processor, and accepts standard low-profile SDRAM modules - i've got 256 mb of RAM in one right now. very cheap - about $15 on ebay small size: 9.5"x8.75x1.4" - a little under half a rack slot evidence that linux works on them, the question is whether RTAI will be happy running on it, or if we'll see glitches due to APM and onboard video like in most laptops cons: weird power supply connector - make sure that it comes with the power supply! however, one should be able to bypass this connector and go straight to the AT-style connector on the board. i'm not sure if the pinout is the same as an AT power supply. processor heatsink looks like it requires good ventilation. making it water/dust proof could be annoying. somewhat slow processor speed of 233mhz, If this is enough to do hexapod kinematics calculations at a reasonable rate remains to be seen. might be able to overclock it with a good heatsink. Strategies The built-in disk-on-chip is rather small at about 16mb, and some of the systems didnt have the chip in them at all. Out options for a boot disk are: boot off the network using PXE or similar use a 40X compact flash memory card ($30) hooked to the ide bus with a cheap ($6) adapter available on ebay. This requires a funky 44 pin 2mm pitch IDC header ($2) and ribbon cable ($1) use a usb stick ($14+) with DSL - may be slow boot off a usb CF card reader ($6) (see if this really works) soon to come - how to flash a boot image to a cf card 2006-05-13T08:53:32+09:00 http://www1.atwiki.com/gingery_machines/pages/179.html Hextatic mini hexapod The Hextatic will be a high speed 5 axis benchtop milling machine/router. It is intended as an experiment to try out various ideas before embarking on a larger project, the hexatonchires ?. At this point the rigidity of the structure is unknown, but my goal is to be able to rapidly mill aluminum and do light milling on steel. It will be lightweight, modular, extremely portable, and inexpensive. (well, inexpensive as far as servo CNC machines go at least!) Click here if you don't know what a hexapod is! Major components of the hextatic include: six iskewb integrated motor/drive modules (cost: $40 each,) which include:japan servo 127K9720 servo motor, irfz44n-based h-bridge motor power stage xl1000 thin client running linux with realtime extensions and emc2 the motherchip, a high speed parallel port interface/breakout board ($20) (DIY/open source) handles communications between a standard PC and up to 8 iskewb modules A 24V 10A unregulated power supply ($50-$100 bought) multiple spindle attachmentsa cheap harbor freight laminate trimmer with 1/4" collet ($20 p/n 44914-2VGA) an air powered 1/8" collet ($15 p/n 47869-2VGA) custom built planetary geared 3/8" collet spindle for heavy milling and drilling, with encoder position feedback What follows is an idea and part number dump. Pardon the mess. This page is for mechanics and geometry only, no electrical stuff here please. timing pulleys What I want: 90 or 120-groove MXL double flanged 1/4" belt width 8mm shaft size, with some sort of hub or way to lock the pulley to the shaft. MSC is being a pain and hardly has any timing pulleys in stock. All the other outlets have even less (mcmaster) or are ridiculously expensive (sdp-si at $12 each for plastic unflanged) At least you can check to see how many are in stock on the MSC item's page. (17 as of 4/29/06) I ended up ordering: 04894838 100 groove MXL plastic single flange no hub $1.48 qty 6 Well actually i ordered a bunch of different types to try them out and see which worked best, but that's all they sent me. Nuts. I considered hobbing my own out of plastic or aluminum stock but since I'm running out of time and have too many irons in the fire, I was forced to buy them. Too bad, because otherwise I'd have an unlimited supply of custom timing pulleys. Maybe I'll do it some time in the future anyway. Section view of the pulley assembly I hacked together. The pulley comes with a stub on one side that possibly could be drilled/tapped for a setscrew? But it didnt look very sturdy and had to be trimmed off with the lathe because it was in the way of the three 8-32 mounting screws for my custom aluminum setscrew hub and flange. A job made easier by gripping the pulley inside the rim with the 3-jaw chuck. I'm using a largish flange with a notch cut out, in conjunction with an optointerrupter as a primitive 1 line per rev optical encoder. This will enable precision homing without any modification to the current incremental-only encoder assembly. Pulley housing The motor mounts are coming together. The blue styrofoam will be cast into aluminum using the lostfoam method. Timing belt McMaster-carr p/n 7887K85 1/4" wide 165 groove neoprene/fiberglass MXL timing belt, $3.57 ea min qty 1 - Should work.. 608RS ABEC7 skate bearings, qty 2 to support radial and thrust loads on the screw. Ordered from VXB bearings on ebay.com - $0.50 each or so in qty 100 --these might be a bit wussy in the axial direction. I'd prefer angular contact bearings but they cost a heck of a lot more and you cant reliably find them in the dumpster either :) Anti-backlash nuts Will be cutting these out of acetate or some nylatron that I have laying around. Acme threaded rod can be made into a tap by turning a taper and grinding flutes into it. Basic design is a split nut that is squeezed closed by a rubber O-ring. Someone suggested the o-ring might stretch over time and lose its elasticity, so I may go with a 2-piece spring and tapered collar design. __ __ O___ -------- |_| | | /\/\/\/\/\/\/\/\/\ O-ring anti-backlash nut \/\/\/\/\/\/\/\/\/ _ | __--------__|O|__| ____ spring ___ |/\/\/\| / ____ |______|// | | /\/\/\/\/\/\/\/\/\/ sleeve/collar a-b nut \/\/\/\/\/\/\/\/\/\ | ________ | | |\\____| ____|/\/\/\|_\ need to get springs to make this work - msc part numbers of interest while i'm ordering from MSC: 02312395 0.688" OD 89lb/in stainless spring - $0.71 ea in qty 6 jmk suggests using wave washers instead of springs 02566628 0.588 ID 0.734 OD rusty steel wave washer $0.34 min qty 25 struts stainless 316 tubing from scrapyard - qty 2 dimensions 10ft OD 1" ID 0.835" aluminum tubing crusty 9ft OD 1.25" ID 1.125" - has holes 18" from either end 1/2"-20 threaded rod, cut into 18" pieces - qty 8 (2 6 foot sticks) -- 18" was chosen because it divides nicely into 3 foot and 6 foot sections, is reasonably large for a tabletop sized machine, and isnt too long. (I was worried that 24" screws might have excessive bending when cutting tougher materials.) Bearings and the drive pulley will take up approximately 3 inches, leaving us 15" of screw thread. The nut will be another 1" or so, and if I size all the tubes and slide bearings correctly I should end up with 14" of usable travel. Will I ever even approach critical speed? read critical speed section (critical speed is rpm at resonant frequency which can lead to severe whipping of the leadscrew) http://www.nookindustries.com/acme/AcmeGlossary.cfm#ScrewCritical critical speed calculation for 1/2-10 acme rod 15 inches long: 1.47*4.76*10^6*0.33/15^2=10262 rpm this value double-checks with the chart provided 10262 rpm * 10 tpi = 1026 ipm max (434mm/sec), not too shabby at all. However I will only be going at a maximum motor speed of 2500 rpm with a 30/100 gear ratio, so my real leadscrew speed will never be over 750rpm == 75ipm (30mm/sec) linear elongation speed. So, the answer is no. What is the maximum compressive force the struts can withstand? Formula from http://www.nookindustries.com/acme/AcmeGlossary.cfm#ScrewColumn Calculation for 1/2-10 acme rod 15 inches long with a 1 inch nut, fully extended: 14.03*10^6*2*0.33^4/14^2 = 1431 pounds of force max before the rod exceeds its elastic limit and bends permanently. This corresponds with the chart but seems to suggest that I'm approaching the maxmimum practical length of the screws. Hey that means I'm doing something right! The struts can carry much more load when only partially extended (as will be the case for most work) so we shouldn't worry about these numbers so much. Also the chart tells us the maximum load for 1/2-20 screws at any length is 4000 lbs, but I expect something else will break before that. 2006-05-13T08:22:58+09:00 http://www1.atwiki.com/gingery_machines/pages/194.html Motherchip The motherchip talks ieee 1284 protocol with the EPP parallel port, receiving control commands, motor torque commands, and receiving position, state, home offset, and general i/o. The ieee 1284 protocol has some rather tight timing requirements (must pull ack low within 10usec of data strobe) so it handles comms while the attiny13's in the iskewb are busy counting encoders. The motherchip has 8 (expandable!) slave select lines to talk to 8 iskewb's or other peripherals. Maybe will throw in some kind of power supply monitoring as well.. Should be able to squeeze it into a tiny card, but will fit into half a 1U rack slot if nothing else.. Parts: atmega48 microcontroller talks to a PC or thin-client (such as the netier xl1000 running emc2 via the ieee1284 protocol across a standard EPP parallel port, handles a small amount of I/O, and drives the SPI bus to control each individual iskewb ps2505-4 quad photocoupler, qty 4 to protect the computer and parallel port from voltage surges and protect the motherchip from static discharge during handling - cheap insurance, Jameco P/N 320725 $1.29 each min qty 1 -- might be too slow, rise time 5us fall time 3us HCPL2530 2 channel high speed opto .4us rise time .4-1.5us propagation delay Digi-Key $1.94 ea tlp2630 2 channel high speed opto 0.03us rise time 0.075us propagation delay, note: inverts the output digi-key $2.25 ea 4n33 Opto Isolator qty 8(?) t- goldmine-elec.com p/n G13385, $0.20 each min qty 5 pcb mount RJ45 jack qty 8 side entry low profile Jameco P/N 524461 $0.55 ea min qty 10 OR ES1607 shielded PCB mount RJ45 jack from excesssolutions.com $0.25 each min qty ?? 2006-05-13T07:39:32+09:00 http://www1.atwiki.com/gingery_machines/pages/178.html Japan Servo 127K9720 This is a little 24V brushed DC servo originally intended for canon copier machines that I plan on using in the prototype hexapod. These little guys have a 4.1" square base with mounting holes in a 3" square bolt pattern. It comes with an oddly shaped piece of metal attached to the motor with a pivot hole and slots (4" pivot radius) used for adjusting belt tension. The motor body is in two sections, one made of stamped steel, the other of cast aluminum or pot metal. The stamped steel section is 0.625" tall and just under 4" in diameter, and made of 0.059" sheet steel. There is a raised nub on the side the wires exit the motor that could cause problems if mounting the motor flange from underneath. On the pulley side (bottom) there is a 1.4" diameter 0.165" tall nub with another 0.825" diameter by 0.125" tall nub on top of it, which holds the shaft bearing. The shaft has press-fit on it a 0.75" diameter 0.54" tall die-cast 30 tooth MXL timing pulley with 0.943" diameter by 0.054" thick flange. The cast section is 0.5" thick and 3.23" diameter. It is attached to the stamped section with three machine screws and is centered by three raised arc-shaped nubs. A red and a black wire come out of a rubber 0.225" x 0.493" x .940" translucent plastic connector with two D-shaped (like an ATX power connector) sockets. A tamiya connector might work for this, if filed down a bit. On top of the cast section is a 1.062" tall 2.20" diameter black plastic encoder cover, secured by 3 radial (#4-40?) philips screws to a 0.35" tall 2.125" diameter raised hub on the cast section. This cover is rather fragile; I broke it when I accidentally knocked the motor off my desk. The encoder cover is marked: DC SERVO MOTOR DF10BE22-01 DC 24V 127K9720 No. 201 JAPAN SERVO CO., LTD TOKYO JAPAN The encoder ring is glued to a set screw hub bolted to the 0.236" shaft. The encoder sensor is mounted to a round circuit board that just connects it to a 4 wire cable. The cable has a connector that is exactly like a CD-ROM audio connector. There are 4 gold pins on 0.1" centers in a row. Starting with the pin with an arrow pointing to it, we have: Vcc (white), A (green), B (red), GND (black). The encoder cable has a rubber sealing grommet that fits in a 0.316" slot in the plastic cover. The outer diameter of the grommet is about 0.45". There is also a 1.12" long by 0.55" OD by 0.25" ID ferrite bead surrounding both the encoder and motor power wires about 1" from the connectors. The poewr wires are 5" long and the encoder wires are 5.5" long and 0.2" thick (due to shielding?) and the discrepancy gives the two cables a nice bend of about 90 degrees. The motor coils are about 10 ohms. From the catalog: Rated at 24 VDC, 2700 ±270 RPM, 22 watts, 0.9 amps, torque 11 in-lb. 12V DC, 1300 RPM, torque 6 in-lb. CCW rotation. Optical encoder: 2-phase, 500 pulse/rev/phase; TTL logic; requires 5V DC supply @50 mA. Includes wiring diagram. Shaft size 0.225 dia. x 5/8" L with press-fit aluminum gear, easily removed. Motor size 4" dia. x 2-1/4"L with 4"sq. face plate. These motors are available from Jon Elson here and also from an ultra-top-secret source known only to a select few.. :) 2006-05-13T05:43:39+09:00 http://www1.atwiki.com/gingery_machines/pages/177.html Motor H-bridge The specifications of this h-bridge were mostly dictated by the small Japan Servo 127K9720 servos I am using. They have a fairly common encoder resolution but were designed for 12V operation, which is fairly low. Maybe this will be an advantage rather than a handicap. Since these motors are only rated 20W continuous and 100W peak, I am using components with appropriately small ratings in order to reduce cost to a minimum. I am sure most of you will appreciate that, as there seems to be very little in the way of economical open source motor control designs out there. I am aiming to keep the whole design below $20 per controller, but without sacrificing efficiency or reliability. This is my first power electronics project, so suggestions and comments are appreciated. -fenn Components IRFZ44N N channel mosfet, 35A peak 60V, used for switching power to high and low sides of the bridge. $0.37 each in qty. 50 from digikey, part no. =IRFZ44N -- oh no! it seems the reason these mosfets were so cheap is that digikey is trying to close out the product line, so get them while they're still there! IRFZ44E would work well as a substitute. stupid.. little.. gray pads!! grrr! the mosfet heatsink tab is connected to the drain terminal, so i need to isolate it from the heatsink.. digikey p/n 345-1079-ND $0.17 in qty 10 freewheeling diode, may not be necessary due to the relatively fast reverse recovery time (95ns) of the IRFZ44N. :other candidates:| SB560-E3/1GI-ND 60V 5A schottky $0.54 ea qty 1; 1n5400 or 1n5400-t, 3A 50V standard and fast recovery, $0.20 and $0.41 each, respectively 1N4148 digikey part no.. ? $0.07 ea. - exactly which version was jmk talking about? 1N4004| general purpose diode, 400V 1A, $0.069 in qty. 10 from digikey, part no. 1N4004-E3/23CT-ND -- recovery time of 2us is way too slow to work as a freewheeling diode; forget it. clamp diode qty 2 30V 1W zener diode to keep inductive kickback and back-emf from destroying the mosfets. digikey p/n 1N4751A-TPCT-ND $0.10 each min qty 10 power capacitor to supply power at the switches, and reduce EMI on the +24V supply lines CAP,ELEC,SNAP-IN,4700UF,50V,105C,20%,25X30MM, Jameco p/n 331897PS $2.19 min qty 1 (chosen for its small size) CAP,MYLAR,.22uF,100V, Jameco P/N 26972PS $0.19 each min qty 10 gate driver transistors to convert 5v logic from the micro/parport to appropriate levels for driving the mosfets. need to source _and_ sink the gate charge to turn the mosfet on/off? candidate MPSA06 80V 500mA transistor, $0.17 ea min qty 1 from digikey current sense resistor 0.1 ohm 5W wirewound (? looks like a sand resistor) from futurlec p/n R0001R5W $0.20 ea; this will give me 0.5V at 5A, so will have to diddle with adc voltage reference to pick up such small voltages. current limiter an op amp to amplify the voltage across the current sense resistor and cut off voltage to the mosfets if above a certain value. need to latch it so it acts like a breaker - how to do this? is this necessary with fuses? oh well a little redundancy never killed anyone. digikey p/n NJM#4558D-ND $0.36 each qty 1 power supply connector 3 pole pcb mount screw terminal for 24V 36V and GND- goldmine-elec.com p/n G12942 $0.20 each min qty 5 motor connector I had a really hard time finding connectors that fit the 127k9720, but eventually I figured out that if i chopped up an atx power connector into columns and whittled one of the plugs down to a "D"-shape, it would work! shown in the background are the tools used: a dremel with saw blade, and hobby knife. Since there is no retaining clip, the connector tends to fall out, so I will probably use some heatshrink tubing over it. 2006-05-13T04:48:01+09:00 http://www1.atwiki.com/gingery_machines/pages/86.html machine#current_designs: hexapod Hexapod The portable machine tool of the future. News! 4/17/06: I have begun work in earnest on a small hexapod milling machine codenamed "hextatic" Why a hexapod? The variable length struts all point directly towards the end effector. All force vectors are directly along a strut. There is only tension or compression, and no bending anywhere in the (ideal) structure. This leads to a very rigid and lightweight structure, similar in function to a geodesic dome. The total error is averaged over all six actuators, instead of accumulating error as in a stacked slide (milling machine) or serial rotary joint mechanism (robot arm). Similarly, wear is evenly distributed. All six actuators work together to move the platform, providing higher power/weight ratios since less time is spent idle per actuator. The high rigidity and low mass allow for very high acceleration rates, making feasible techniques such as high speed machining and active vibration compensation. The machine is composed of multiple identical parts, greatly simplifying manufacture, design, and repair. It is easily disassembled, shipped, and reassembled on-site. Machines with an enclosing frame (such as the ingersoll octahedral hexapod ) do not require heavy foundations or precision leveling. Under construction. Sorry! Well, would you rather have me work on a real-life model, or sit here and write about what other people are doing? item vote work on a real life example (18) write about what other people are doing (1) pros cons workarounds build test reiterate some ideas: use an airmotor (or tesla_turbine ?) as the spindle drive motor, saving on mass, space, and heat buildup per spindle horsepower tube-shaped ballscrews ? with internal tension and viscoelastic damping layer delrin ACME ? anti-backlash nuts for the cheap prototype For a small prototype, I'm planning on using a rollerblade bearing u-joint for 2-dof linkages similar to what's found on the FANUC F-100i robot Fanuc F-100i robot The F-100i is capable of a static payload of 140kg in a vertical direction and 75 kg in a horizontal direction. It seems to be less than one meter tall, and weighs somewhere around 100 kg. It's not in production, unfortunately. It seems to have been eclipsed by the much less aesthetically pleasing f-200iB: http://www.fanuc.co.jp/en/product/robot/lineup/f200ib.html The F-200iB has a functional work envelope in the shape of a cylinder, with diameter 500mm and height 300 mm, but the full work envelope is a complex lenticular shape about 1 meter in diameter and 500mm in height. Repeatability is within 0.2mm. The robot is 800-1300mm tall, 630mm in diameter, and weighs 190kg. It is capable of moving a 100kg payload, has a maximum speed of 1.5m/s, and is primarily intended for welding and assembly applications. Just to give you some idea of what the run of the mill hexapod is like. Hexapods are also sometimes called (for your googling pleasure) Stewart Platforms Parallel Link Manipulators Parallel Machine Tools Geodetic machine tools geodesic machine tools octahedral hexapod A hexapod that uses flexible cables instead of ballscrews or hydraulic cylinders is called a robocrane, at least that's what it's called by the guys at NIST. links: Pictures of the NIST/Ingersoll prototype octahedral hexapod: http://www.mel.nist.gov/photos/photo/phexa99.html hexapod cad drawings, inspiration, hexapod design engineering software, code bits: http://www.i-way.co.uk/~storrs/lme/LMEHexapodMachine.html double ball bar procedure and calculations for a hexapod: http://precnt.prec.kyoto-u.ac.jp/mmc/research/para/ishida/pa_ishida.html parallel mechanisms info center. history, comparison, nomenclature http://www.parallemic.org/ forward kinematics translation equations. fortunately we dont have to do this thanks to EMC (I think): http://www.enee.umd.edu/medlab/papers/robautforsem97/robautforsem97.html EMC. Enhanced Machine Controller. Linux realtime motion planner, cnc controller, step sequence generator, live/preview tool path , and more: http://www.linuxcnc.org/ Hexel owns the trademark for "Hexapod." Fuzzybunnies. Thanks for raping the information commons, guys. Send them your hate mail: http://www.hexel.com/ Most folk ignore "Hexapod" as a patented/trademarked word much the same way we ignore the "Windows" thing when we put clear panes of glass in our homes. If it becomes a problem with cash down the line, call it a Stewart Platform. That is well beyond the experation date. University of Wyoming hexapod: http://wwweng.uwyo.edu/electrical/research/hexapod/ University of Florida hexapod: http://www.mae.ufl.edu/SAMM/Research/Manufacturing/manufacturing_main.html Homemade DIY hexapod: http://www.foxkid.net/cmm/platform/project-notebook.html Another DIY hexapod: http://robofac.sourceforge.net/screenshots.html I am carefully avoiding any mention of "that other hexapod robot" that is really stupid and serves no purpose whatsoever. Also avoiding any terms that are commonly used to describe it, so that people searching and doing a NOT boolean operation won't cancel out this page. mostly ideas generated by fenn 2006-05-12T07:29:07+09:00 http://www1.atwiki.com/gingery_machines/pages/193.html ISkewB Industrial Servo Brick Gotta love marketing eh? Really I just needed a name to refer to in other projects. This will be an integrated motor module - motor, sensors, power electronics and data acquisition, all housed in a compact, rugged, waterproof heatsink/enclosure. Plug it in and go! Target budget is $40 each.. damn that makes it sound expensive :( It actually is turning out to be quite a challenge to get it that low.. The initial revision will use an attiny13 to PWM an h-bridgeand simultaneously count encoder pulses, talk SPI via cat5 cable with a motherchip, and read/write i/o for limit switches and whatever else. (thermistor? cooling fans?) RJ45 interface pinout 5V GND SS SCK MISO MOSI Fault (open collector-ish) Parts list: h-bridge power stage 127K9720 servo motor pcb mount RJ45 jack top entry Jameco P/N 524488 $0.39 each min qty 10 OR pcb mount RJ45 jack side entry low profile Jameco P/N 524461 $0.55 ea min qty 10 OR ES1607 shielded PCB mount RJ45 jack from excesssolutions.com $0.25 each min qty ?? cable assembly,cat 5e,3ft jameco p/n 523021, $1.33 ea min qty 2 Standard Opto-interrupter, fork style, qty 2 forhome and index pulse goldmine-elec.com p/n GP44, $0.14 ea min qty 18 (or so) (i know bad choice of components, i'll find something on jameco or digikey later, something like Jameco P/N 114091) thermistor goldmine-elec.com p/n G14610, $0.33 ea, min qty 3 encoder read board: attiny13 cheap powerful microcontroller to read encoder signals and PWM the h-bridge CAP,RADIAL,33uF,50V, 5mm x 11mm power filter for attiny13, Jameco p/n 93745PS, $0.05 ea min qty 10 CAP,CERM,DISC,.1uF,50V power filter for attiny13, Jameco p/n 15270, $0.15 ea min qty 10 cd-rom audio connector for connecting the encoder lines, maybe this should go in iskewb..Jameco P/N 100802CK $0.25 min qty 10, and jameco p/n 152733CK (maybe?) $0.45 (need pins too) jameco p/n 181673 another $0.44 each. also maybe molex 50-57-9004 would work, it can be obtained from digikey. 2006-05-11T23:25:05+09:00 http://www1.atwiki.com/gingery_machines/pages/173.html machine#current_designs: robocrane RoboCrane (*1) Much work in the field of parallel kinematics has to do with cable-based systems. I will refer to these by the name RoboCrane to differentiate them from hexapods. RoboCranes have the advantage that a large percentage of the frame volume can be utilized, since the travel of a cable and winch mechanism is higher than that of a ballscrew strut assembly. Cables are also lighter and more easily transported, especially when considering an actuator with hundreds of meters of travel. RoboCranes are quite easily controlled using EMC, and there are multiple papers written by the original authors on this subject. (see below) Fig. 1: the hanging platform cannot rotate beyond the point where the center of mass is directly below a joint (or through any kinematic singularity) While a RoboCrane is capable of being positioned in six degrees of freedom just like a hexapod, its range of rotation is restricted to much less than 90 degrees, since the cables have no compressive strength. This limitation also applies to hexapods, however a hexapod can still function once it has passed through the kinematic singularity because the struts can provide compressive support. The amount of force a robocrane can exert in any direction is equal to the amount of preload. (verify this) Preload can be accomplished by using redundant cables to tension the structure or by adding weight to the moving platform. Applications of robocranes include: Robotic greenhouse tending Contour Crafting, Excavation, and traditional construction Material handling, pick-and-place, automated component assembly Photography, inspection, and digitizing large objects Hedge trimming and ice/snow sculpture In the next year or so, expect to see some examples on this page of research on the topic of incorporating robocranes into geodesic domes. Links: Models of various arrangements to accomplish the robocrane concept. http://www.isd.mel.nist.gov/projects/robocrane/gantry.html Alex Joni's 3 cable system, to be used as a plasma cutter XY table with torch height control: http://dsplabs.utt.ro/~juve/emc/mytoy/ 2006-05-07T21:30:12+09:00 http://www1.atwiki.com/gingery_machines/pages/190.html nothing here yet 2006-05-07T02:58:16+09:00 http://www1.atwiki.com/gingery_machines/pages/189.html nothing here yet 2006-05-07T02:57:25+09:00