Shop OBEX P1 Docs P2 Docs Learn Events
Prop-based CNC/3D printer - Page 12 — Parallax Forums

Prop-based CNC/3D printer

189101214

Comments

  • EffingWarrenEffingWarren Posts: 4
    edited 2011-12-03 06:00
    As I am more of a hardware guy...

    In sourcing parts for my CNC mill/3D Printer I stumbled across this site http://www.carving-cnc.com/cnc6040-series/cnc6040zf-router-engraver-drilling-and-milling-machine.html

    This is a well built CNC router at a low cost. Intrigued by this product, I contacted the manufacturer in China. They are more than happy to sell you a custom machine. I asked about ordering a machine as you see pictured, sans the computer interface, and got a price of around $800.

    While $800 is a lot of money, having sourced the parts they use here, you would find your self hard pressed to come up with the BOM at that price, even using eBay and other discount sources.

    Even if you are going to source your own parts, their precise design warrants study. Their use of linear rails and extruded components is a perfect example of a bot that is going to be capable of high precision motion, as well as sturdy enough to mill/route in most metals that are not hardened. IMHO this would be the ideal base, if we are looking at a machine that can both print and mill. With the addition of a laser tube and a few mirrors, we could even add laser cutting to the functions of this machine.

    Speaking of lasers, has anyone considered making a combination printer/laser cutter rather than attempting CNC milling/routing. With variable output or high travel speeds, you can even do 3D cutting with a laser. While the tube, supply, and mirrors might run a bit more than the cost of a Dremel or the like, if you can make your machine print, you can make it cut with a laser in no time, regardless of the rigidity of the machine. Even those of you who have a reprap already, could be cutting in a matter of a few hours.
  • prof_brainoprof_braino Posts: 4,313
    edited 2011-12-03 09:22
    price of around $800.

    laser tube and a few mirrors, we could even add laser cutting to the functions of this machine.

    Even those of you who have a reprap already, could be cutting in a matter of a few hours.

    Go for it on the Chinese CNC, I would be interested to see how it performs. It's a bit beyond my budget at the moment, but if the current project pans out I would like to go in that direction.

    I'm a bit skittish when it comes to lasers, its so easy to blind oneself. The TOF rangefinder project is supposed to be eye-safe and is being handled by an experience professional, but even that scares me. An acquaintance has an IR laser cutter that will not operate without the IR opaque shield in place, I would hope for this type of safety to be the primary consideration.
  • Graham StablerGraham Stabler Posts: 2,510
    edited 2011-12-03 09:42
    For a Co2 laser with a wavelength of 10.6um IR even perspex is opaque (and is used for most enclosures). Likewise the eye itself is opaque to IR, the light it not focussed by the eye on to the retina. Of course there are still dangerous as the cornea can be damaged but a 50W CO2 laser is safer than a 200mW red/green or near IR laser to the eyes.

    Graham
  • EffingWarrenEffingWarren Posts: 4
    edited 2011-12-06 00:30
    You also have to consider energy density, or amount of energy being pumped into a given area in a given time frame. A relatively low power laser can be used for cutting, if you are willing to invest the optics to form a collimated beam with a very small diameter. Inversely, a very diffused laser, even with high power is rather harmless.

    I can not name the company, but one I once worked for used a laser to expose film in a holographic process. The beam had the power to blow a hole in a penny when formed to a diameter of .8mm or less in only three pulses (operating pulse rate was over 120Hz on this particular machine) yet the same beam, when diffused across a larger area in a different section of the machine, (aprox 1cm diameter), could be applied to the film for more than 20 pulses before beginning to burn the film.

    Best advice concerning Laser safety.
    ALWAYS wear safety glasses for the frequency of laser light you are working with, if you use more than one laser, its going to cost more for lenses that are certified for multiple frequencies.
    NEVER place ANY part of your body in the beam path.

    But with a machine like we are talking about, a plexiglass enclosure that has a yellow tint to it will more than sufficiently protect your project, or rather your eyes from your project. For that matter, it doesnt even have to be a complete enclosure. Now that I am thinkign about it, many of our machines simply had shields in front, where the operator would be, to protect them. Again, a scattered, diffuse laser is not all that dangerous.
  • Graham StablerGraham Stabler Posts: 2,510
    edited 2011-12-06 02:45
    For a CO2 laser a yellow tint makes absolutely no difference. One reason perspex cuts so well it because it absorbs 10.6um radiation completely. Clear perspex is not clear to 10.6um and yellow is of absolutely no relevance to a wavelength 10 times greater than visible wavelengths.

    The only way 10.6um gets through is by burning a hole.

    For lower wavelengths proper enclosure is best, add a webcam to watch the action.

    Scattered light from class 4 lasers of visible wavelength, is dangerous by definition though less so for 10.6.

    Graham
  • prof_brainoprof_braino Posts: 4,313
    edited 2011-12-10 13:44
    Rig is now calibrated. I think I know something about 3D printing at last. :)

    So far, except for details specific to my rig, the default settings in the tool chain are correct.

    The problem was figuring out which parameters affect details specific to my rig.

    The one that got me the longest was oscured by the myriad of SFACT (skeinforge) settings.

    I experienced problems with feed rate, and no amount of tweaking SFACT would fix it. When I finally got to checking e_steps_per_mm, I discovered that my feedrate issues was due to INCONSISTENT feedrate. The root casue turned out to be insufficient pressure on the filament idler bearing. All I had to do was tighten down the filament idler bearing screw such that the ballpoint pen springs are nearly completely compressed, and the feedrate problems have gone.

    The only big issue remaing is flakes of filament that accumulate, and need to be blown out of the filament feed on the extruder head. This may be due to the hand dremeled hobs being too deep in my threaded rod. (Before I tightened down the filament idler bearing springs, I thought the hobs were too shallow and getting clogged, and this might be the cause of inconsistent feedrate; so I swapped in a deep hobbed rod). If this becomes annoying I will replace the deep hobbed rod (dremeled deeper than the threads) with the original shallow hobbed rod (http://www.thingiverse.com/thing:9291)

    Of course, the machine will not be verified until another machine is built from parts printed on this machine, and feedback is recieved on the quality of the parts. But this final verification begins to excede the scope of this phase, so will have to wait until the next build.

    For now, I have an idea of what the baseline should be.
    1024 x 768 - 104K
    1024 x 768 - 81K
    1024 x 768 - 106K
  • Cluso99Cluso99 Posts: 18,069
    edited 2011-12-10 17:40
    Great work Prof. Nice to see it working at last.
    My frame still sits here for lack of time :(
  • EffingWarrenEffingWarren Posts: 4
    edited 2011-12-16 10:40
    Just found this... What is the possibility of porting for the Prop?
    http://grbl.tumblr.com/
    http://dank.bengler.no/-/page/show/5470_grbl?ref=mst
  • StefanL38StefanL38 Posts: 2,292
    edited 2011-12-16 11:14
    I don't know much about C. But as it is written in C and there is a c-compiler for the propeller. The logic of the sourcecode can be used.
    You will have to make adaptions for compiling for the target propeller-chip instead of arduino. And you will have to make adaptions to the different management
    of the IO-pins. How much effort this is for you, depends on your knowledge about the hardware-internals of arduinos and the propeller-chip.
    keep the questions coming
    best regards
    Stefan
  • prof_brainoprof_braino Posts: 4,313
    edited 2011-12-16 13:26
    Sal is talking about the possibility of doing CNC control on the prop. He is a master of embedded systems and assembly language. He already has support in place for stepper motor control and various sensor inputs. This thread has established a base line for a target implementation. The only major portions remaining appear to be control of the machine head and and G-code interpreter.

    So, a prop based CNC control is likely on the way, but it should not be ready for release (at least from this source) until say summer at the earliest.
  • prof_brainoprof_braino Posts: 4,313
    edited 2011-12-26 08:49
    Added this summary to the first post:

    Build log of a standard 3D-printer to be used as a baseline for this project

    http://reprap.org/wiki/PrusaBuildNotesBraino

    Based on this build, focus is a swap in replacement for Sanguino v1.3 electronics for a Mendel Prusa. This will allow an apples to apples comparison of cost, complexity, accuracy, functionality, etc between a prop solution and the stock solution. This investigation will use the same connectors and Pololu stepper motor drivers from the Sanguino. The Prop board and prop firmware will be the only change. The result is to be compatible with the PC side of the 3D printing tool chain:

    PRINTRUN/PRONTERFACE python gui on the PC
    SFACT/Skeinforge on the PC

    OpenSCAD will be used to create model on the PC, and export these as STL files. The STL files will be processed on the PC to generate G-code. The G-code will be processed on the Prop to control the device motion.
  • StefanL38StefanL38 Posts: 2,292
    edited 2011-12-26 10:59
    Hi Prof. Braino,

    I tried two examples from the OpenSCAD-software-package. From what I've seen there OpenSCAD-Software seems to be just a script-language to generate 3D-objects.
    When I downloaded it I thought it is a freeware 3D-CAD-system. I mean drawing konstructions in 3D with drawing functions like draw rectancle, circle, extrude to 3D etc.

    Of course modern CAD-systems have scripting capabilities. But where is the classical drawing aspect of OpenSCAD? Are these functions hided anywere in OpenSCAD or is the name just misleading?

    best regards

    Stefan
  • prof_brainoprof_braino Posts: 4,313
    edited 2011-12-26 12:00
    As far as I can tell, openSCAD is a only a scripting language. We compose little program-ish chunks of code to define and object, example
    cube (size=[5,6,7])
    

    make a cube 5 unit in X, 6 in Y, and 7 in Z. The program displays an image of the items defined, and we can move, rotate, and zoom the image on the gui display. We CANNOT click on points with the mouse and add points and lines etc. It takes a little getting used to, but not so tough once we get going. So the name is true; it does use a Computer to Aid Design, but its not AutoCAD.

    I needed a filament guide for my extruder, the filament kept twisting and sliding off the hobbed bolt. So my first part is this, it took about a day starting from a blank file.

    http://www.thingiverse.com/thing:14960

    Next, I am making a case for an android device. Many of the details on the device will be close to the tolerances of the printers. If I can make a case the fits and doesn't fall off, I think I will know something about designing using openSCAD.
  • StefanL38StefanL38 Posts: 2,292
    edited 2011-12-26 13:24
    Maybe I have this opinion just because I'm not used to create 3D-objects with script-languages.
    But I guess creating complex 3D-objects with a CAD-software is much easier than with a scripting language.

    Ok a simple cube, or sphere is done very fast. But how about a PCB-case with with a lot of holes and mounting-holes etc.?

    For 2D I use QCad (which is mentioned on the OpenSCAD-site. In 2D for me it has the right size between complex functionality and ease of use. Anyway I wish QCad would be 3D.
    Ten years ago I used MegaCAD as I worked at an house-facility planning company (don't know the real english word). It was not as ovrloaded like AutoCAD or even CATIA or ProEngineer.

    As I'm already writing about CAD-Software does anybody know of a FREEware with the same concept as QCAD / MegaCAD: two-letter commands typed on the keyboard to start drawing
    or zooming or whatever using the mouse to set start and end-points of the drawing object? So to say some kind of google-sketchup but with linetypes as usual in professional CAD-software?

    best regards
    Stefan
  • eiplannereiplanner Posts: 112
    edited 2011-12-26 18:19
    CONGRATULATIONS!!!! Way to go Prof_Braino on seeing this project through. Very impressive to say the least.

    I just started reading this thread earlier this afternoon and have followed along eagerly. (Reviewing all the various links provided as well as darting off on a few tangents on my own.) I would have loved to have followed along and built my own right beside you guys. Too bad I am just now finding this thread. However, it's never too late to get started. I have a high interest and fascination with CNC but very little knowledge on all the steps to make it work. I would like to first say THANK YOU!!! to all of you guys involved in this project for all of your valuable knowledge, time consuming research, bench time, software development time, hardware development time, experimenting time, testing time and all the other personal time involved in putting all this information out here in one place for guys like me to find. It's all very much appreciated. Thanks guys!!!

    Now, on to me getting my own 3D printer.......YAY ME!!!!

    OK, in the name of due diligence, (@ Braino) do you have a final tally of your expenses? I read your post with all your notes and it shows costs scattered throughout, but I can't get a complete figure on what it cost you to build. I realize the knowledge gained is priceless but am wondering if you have kept track of your total expense. I want to have my own 3D printer as bad as the next guy but am only willing to build myself if the cost savings is fairly significant. Which I believe was one of the original goals of the project.

    My questions would be:

    How much did it cost?
    How did the precision turn out on your machine in comparison to other comparable complete DIY kits on the market?
    Now that you have done it this way, would you recommend others (noob's like me) copying this approach or going with kits on the market?

    Keep up the excellent posts. There really are those of us out here that are grateful.
  • prof_brainoprof_braino Posts: 4,313
    edited 2011-12-27 07:47
    eiplanner wrote: »
    CONGRATULATIONS!!!! Way to go Prof_Braino on seeing this project through.

    Thanks eiplanner. But notice, project is almost a year so far, and will likely go another year.
    Phase 1 was forming the concept.
    Phase 2 research and data gathering.
    Phase 3 was establishing a baseline. This is the Arduino base Reprap Prusa
    Now we are in Phase 4. The task of this phase is to swap out the arduino and replace it with a prop. The remainder of the tool chain should remain the same. The hypothesis is that having one cog per stepper driver (etc) might show a measureable improvement in positional accuracy, etc.
    This phase also includes improvements to the design of the printer, for lower cost, fewer parts, easier build and maitenance, higher accuracy, bigger print volune with lower desktop footprint. This is the current investigation, we don't know if any of these things are possible. Hopefully they are not, and somebody smart will come along and do it anyway, as typically happens on this forum. :)
    It's all very much appreciated. Thanks guys!!!

    The best thank you would be to try to follow the notes, and give feedback if possible. The goal is to make and instruction for an interested passer-by ( starting with grade school kid and parent) could reproduce a unit without losing fingers. You are now the critical component in the development chain!
    wondering if you have kept track of your total expense. How much did it cost?

    This cannot be tracked too closely or revealed at the development phase, or the boss would cancel the project. But the doulbe secret total is around $600, because I made mistakes. The instruction and BOM list parts that are not used, I got too many extras, I did not get enough of other and had to place and addition order, (shipping was a killer), I was impatient and bought pre-assebled subsystems instead of building from scratch.

    BUT we should have an accurate list soon, and if we take advantage of quantity purchase, the price for a complete kit (seems to be) as low as $500 (based on printrbot, but that is like of a temporary, intro machine, we would have to constantly upgrade the design and parts, which is a cool alternative and might be a good way to go; but not my path today).

    IF you want to be a guinea pig, PM me and I will walk you throught the build.
    How did the precision turn out on your machine in comparison to other comparable complete DIY kits on the market?

    Warp during cooling appears to be the biggest factor in the precision of the parts themself. This can be addressed, but is not well documented (to the point where a beginner can get it addressed right of the bat).

    Trueness of the machine is the other big factor, this depends on care during assembly. This also can be addressed, but also has not been well documented. (My machine is great, about half the guys can get theirs to run straight, they did stuff different). This is being addressed in the next couple week (I hope).

    The single core arduino may also be a place where improvement can be found. Replacing the arduino (one core tending everything in series) with a multicore solution may evoke improvements (many cores each dedicated to a single entity in parallel). So far, the prusa is as good or better than previous designs, but not quite and rigid and expensive as MendelMax http://www.thingiverse.com/thing:12645

    The Prusa has a good cost benenfit for extrusion, but an improved MendelMax class would likely be better for Milling etc. (that is a later project if I live so long).
    Now that you have done it this way, would you recommend others (noob's like me) copying this approach or going with kits on the market?

    Prusa looks like the optimal design AT THIS POINT, but development is in flux, so this should change every couple months as we have seen. Printrbot is a complete kit, but is not been releases to the field yet, and you have to wait till March. Prusa you have to find your own parts (want kits?).
    Keep up the excellent posts. There really are those of us out here that are grateful.

    Again, if you want, PM me and we can try to arrange to walk through the build, and I can update the notes as we go along. It actually quite easy, once you get past the uncertainty of never having done it before. This is open to anybody that wants to contribute to the community.
  • StefanL38StefanL38 Posts: 2,292
    edited 2012-01-03 06:02
    Hi,

    I would like to get an overview over the status of this project.
    Therefore I would like propeller-hats working on things that can contribute on this to post a short description of what they do and of estimated milestones when a certain part is finished.

    Then I would like to ask who is willing to compromise on details for the win of dividing the effort for developing all parts (Software and hardware.

    So I start with myself
    I got two mills for low.
    One with a mechanic and really small NEMA14 0.8 A steppermotors 200 steps per rev.
    I plan to re-use the stepperdrivers (based on L297 and L298) with Step/Dir-Input.
    and a still small mechanic but with stronger steppermotors 4A mill without any electronic driver.
    Just the mechanic and motors.

    I started coding a G-Code interpreter but I'm at the very beginning. My interpreter expects G-Codes as strings like "G01 X100. Y200 Z 300 F250" etc.
    The status of the interpreter is really very at the beginning. Right know I can just extract parts of the commands like "X100" "Y200" etc.

    I have a two axis PASM-stepperdriver creating step/Dir-pulses. This has to be extended to minimum three axis.

    estimated milestones:
    - G-Interpreter finished 9/2012
    - 3-axis PASM-stepperdriver finished 7/2012
    - 8A stepper-driver finished don't know

    I'm somehow involved in the development process of an Opensource Stepper-controller for the OpenSourceEcology-project
    http://opensourceecology.org/wiki/Open_Source_Stepper_Motor_Controller

    Me personal I'm willing to compromise a lot of things about details within the following frame:
    - Well documented and modular code
    modular here means strictly dividing between
    - User-Interface (beeing TV, VGA or serial)
    - G-Code-interpreting
    - Calculating start and end coordinates which are the input parameters of the stepper-driver-software

    modular PCBs which means:
    - G-Code interpreter-PCB that use the DIP40-housing of the propeller
    - Power-stage-PCB using a logic chip that allows currents up to 3-4A
    - Power-stage-PCB using a logic chip that allows currents up 8A (which means discrete MOSFETs)

    Even if you disagree with some parts of this frame please post your opinion and the things you are working on.
    I may change my mind about the frame described above.
  • Cluso99Cluso99 Posts: 18,069
    edited 2012-01-03 14:04
    StefanL38:

    I am working on the hardware.
    * I have a passive motherboard pcb 1.8"sq that houses 4 Pololu compatible stepper drivers based on A4982/A4984 (my pcb) and A4983/A4988 (pololu and others). It has the shunts and caps and brings out the connector pins to a 1x14 header.
    * I have a Pololu compatible pcb using the TSSOP version of A4982 or A4984.
    * I have a prop development pcb 1.8"sq that connects to the stepper motherboard via 1x14 header.
    * WIP to design a heater controller pcb that will drive the MOSFET(s) and sense the thermistors to control the temperature on the extruder and heated bed.
    * I have two pcbs 1.8"x1.2" that do either VGA or TV plus keyboard & stereo out

    All these pcbs can be (designed to be) horizontal headers so the pcbs are horizontally adjacent, and plug together (no cables so cheap). In addition, they will fit into cheap Hammond 1551 R/S/F/G boxes.

    My concept is to make thing modular so that the parts that users tend to destroy (stepper driver chips, MOSFETs) can be replaced simply and cheaply. At the same time, keep it simple to reduce cables required.

    As a basis, use the RepRap open software on the prop. I currently have a prototype micromendel built (frame only).

    Hope this helps.
  • StefanL38StefanL38 Posts: 2,292
    edited 2012-01-03 16:30
    Hi Cluso,

    thank you ver ymuch for replying. I found your thread with all the pictures. Nice work.
    I like it beeing modular for the same reason you mentioned.
    If I see right you are using the QFP-chip. The A948x-chips are TSSOP
    How hard is it to solder the QFP-Chip and the TSSOP by hand with a standard 15W solder-iron when the hot end has a diameter of 1mm?
    what kind of solder-iron would be better suited?
    Do plan to make this work openspource?

    As I have taken a closer look at the PCBs are they doule-sided? Do have a lot of SMD-components on them?

    I can see you want to have it compact and small. Therefore double-sided PCBs, QFP, TSSOP, SMD is the way to go.
    But this makes it much more complicated to manufacture it with minimal equipment.

    How much effort would it be to design a PCB for a TB6560 (HZIP25-package) that is the same way plugable like the other PCBs?

    best regards
    Stefan
  • prof_brainoprof_braino Posts: 4,313
    edited 2012-01-04 07:38
    I am working on mechanical at the moment. Inspired by printrbot, I am examining an earlier goal of 10x increase in accuracy and 10 reduction in cost. I've made some progress using fewer, cheaper parts. It might not reach 10x, but we'll see how far I get. I'm looking to allow for upgrading the unit for milling etc. So far it looks like the basic chassis will be OK for things like PCB drilling, milling wood and plastics; but probably not not so good for milling steel (go figure).

    I'm planning on using cluso99 electronics hardware as possible.

    I plan to use Sal's software. He's said that the motor control should be straight forward, and using the forth interpreter for the G-code interpreter should make for many fast development interations. In addition, it will be easy to modify and extend the G-code finctionality. Sal optimizes the bottlenecks in assembler, so it might be pretty fast execution.

    We think we want to limit scope at this time to the printer itself, so there won't immediately be any user GUI on the printer itself. But there can be easily added, as all the basic module exist already.
  • Cluso99Cluso99 Posts: 18,069
    edited 2012-01-04 23:19
    Stefan: The prop is QFP44. I currently hand solder all parts except the TSSOP A4982/4 because this has a ground plane pad under the chip which must be soldered. So I am making a temp controlled oven for this. I also have an IR Led soldering station that can also be used for this and BGA.

    IIRC the TB6560 is quite simple but does not provide all the advantages of the A498x series.

    There is nothing special to my hardware (anyone can duplicate it but I am not making the gerbers available) and the software hopefully will be a group effort.
  • wjsteelewjsteele Posts: 697
    edited 2012-01-05 06:41
    Cluso99,

    If you have a via hole in the center of the solder pad under that A498x, you should be able to solder that spot from the bottom side and not need the oven at all.

    Just a thought.

    Bill
  • PartDaddyPartDaddy Posts: 2
    edited 2012-01-11 18:33
    Hello everyone, I'm a big Propeller fan!

    Anyway, I would love to see the Propeller running our 3D printer my company just started selling. I designed, tool, injection mold, and manufacture my version (the H-1) of this open hardware machine.

    Google SeeMeCNC H-1, http://seemecnc.com

    ~PartDaddy
  • Cluso99Cluso99 Posts: 18,069
    edited 2012-01-12 01:11
    PartDaddy: Welcome to the fabulous world of the prop.

    I took a look at your site. Nice set of injection parts. I have bookmarked so I can follow your design.
  • LeonLeon Posts: 7,620
    edited 2012-01-12 02:35
    There are some interesting low-cost 3D printers at CES:

    http://www.bbc.co.uk/news/technology-16503443
  • prof_brainoprof_braino Posts: 4,313
    edited 2012-01-12 08:40
    Hey PartDaddy, cool rig! Can you PM me with info on any units in the Chicago area?

    Are the electronic and stepper motors equivalent to the pololu parts commonly used on the Repraps?

    Would you be interesting in testing on your model when we something ready?
  • wjsteelewjsteele Posts: 697
    edited 2012-01-13 09:25
    Hey PartDaddy, I've already got one of your kits. I'm assembling it now! I love the injection molded parts... but there are a ton of screws!!!! :-)

    Bill
  • PartDaddyPartDaddy Posts: 2
    edited 2012-01-13 15:02
    wjsteele, Yes, an insane amount of hardware. Now multiply your kit by 500. For our first machine, John & I liked the open hardware route. Enjoy! ~Steve
  • lonesocklonesock Posts: 917
    edited 2012-01-24 11:24
    <Plug value="Shameless">

    My slicer site went live (http://www.kisslicer.com, and the free version should work fine for printers with a single head). If anyone is developing a G-code interpreter on the prop, please drop me a line, so I can make sure that KISSlicer supports your firmware!

    thanks,
    Jonathan

    </Plug>
  • idbruceidbruce Posts: 6,197
    edited 2012-02-13 07:41
    Hello Everyone

    It has been a while since I participated in this thread, and I have not been following along. I know, shame on me. Anyhow, I now have another PCB driller designed and ready for assembly, so I have decided to turn this x, y, z table into a 3D printer. By looking at the photos, what do you estimate the remaining cost to be?

    Bruce
    1024 x 768 - 51K
    1024 x 768 - 48K
    1024 x 768 - 65K
    1024 x 768 - 46K
Sign In or Register to comment.