Shop OBEX P1 Docs P2 Docs Learn Events
A New CNC Build - 3D Printer - El Cheapo - Page 3 — Parallax Forums

A New CNC Build - 3D Printer - El Cheapo

1356711

Comments

  • TtailspinTtailspin Posts: 1,326
    edited 2014-04-01 10:25
    Not everyone!,
    I know you can cut aluminium stock with a band saw, be sure to use a blade with lots of teeth, and set your guides tight to the blade.
    But, as for sand paper making anything within a couple of thousandths, that will take some 600 grit emery, and lots of patience.
    Maybe consider a well oiled honing stone for the fine tuning... :)



    -Tommy
  • ercoerco Posts: 20,244
    edited 2014-04-01 10:31
    idbruce wrote: »
    So what everyone is telling me that I cannot cut aluminum with a band saw and then true the ends within a couple thousandths on a disc sander? I say bull hockey. The widest width is only 2"

    Enough speculation, just do it! Remember that the bumblebee only flies BECAUSE the experts have shown that he can't.
  • idbruceidbruce Posts: 6,197
    edited 2014-04-01 10:43
    I am not talking about a smooth mirror finish, but to the top of the grain left from sanding.
  • LoopyBytelooseLoopyByteloose Posts: 12,537
    edited 2014-04-01 10:59
    Some of us might use an appropriate file and a machinist's square for aluminum, and do it nicely by hand.

    My alternatives are merely suggestions that I feel confident at giving me the best results. Everyone that is a craftsmans has personal preferences built on their own experience and confidence.

    It isn't that you can't. With a table saw, you can set up the blade to 90 degrees with a machinist square, than set up the end cut to 90 degrees with a framing square.

    Finally you can cut the ends of all the pieces of that same length as a stack.

    With a bandsaw you are likely to be cutting one piece to length at a time (more variance) and having less control of the the squaring of the ends.

    You may result in having everything cut without needing to finish the ends with a file or sanding. You may even get results of creating a set of frame pieces that are beyond 1/128" tolerance.. just because they are all square and cut at the same time.

    +++++++++++++
    I am really more interested in how the Propeller plays a role. I just thought the DIY tablesaw might be appreciated.

    You can forge your own steel if you like....
  • ellipserellipser Posts: 43
    edited 2014-04-01 11:55
    It's the distance between the rails that is critical, which would mean drilling accurate holes to within .001 distance apart. That doesn't involve a bandsaw.

    Can you do that in your shop?

    Like I said before, there are techniques to correct for alignment problems, but they require a lot of effort.

    Maybe you have a design that gets around this problem easily, if so, please explain as it would open up the possibilities for hobbyists everywhere that don't have access to a Bridgeport or jig borer. Do you have a design made up on some CAD of some sort?

    And, are we talking about the same thing? You want to build a CNC machine, right? Not just a head for an extruder.
  • TtailspinTtailspin Posts: 1,326
    edited 2014-04-01 12:44
    Being as this is an idbruce project, This made me smile...
    ... but they require a lot of effort.


    Build a crosscut sled for your band saw, and much of your cutting accuracy woes will vanish.:thumb:
    A tool/material holding jig, is a tool/material holding jig. how accurate the jig, equals how accurate your product will be.
    Often referred to in some circles as "Garbage in, Garbage out"... :)



    -Tommy
  • pjvpjv Posts: 1,903
    edited 2014-04-01 13:17
    Bruce:

    First, I mean no disrespect whatsoever.

    It isn't that I think you are incapable, or unresourceful, meticulous and stubborn. You have plenty of evidence to the contrary.

    What I do take issue with is your glossing-over of the complexities and believing, or at least stating that they are easy. They aren't easy. If they were, then all the hobby CNC and 3Dprinter products would be perfect, and they are FAR from that. Many of them unusable for real world applications.

    And, no, I believe that you will not be able to get the precision results you seek with such basic approaches as a bandsaw and disk sander. You will have to seriously relax your expectations if this is the best tooling you've got.

    Machining things to one thou tolerances (I know you have not specified) is difficult..... REAL difficult. I have a full sized commercial milling machine, and it can repeatedly position to one tenth of a thou. And even with that facility, and the rest of the machine shop at my disposal, it still is VERY difficult. The cutter deflections alone can throw you off way more than that.

    So, please don't trivialize the issues. It makes those who have more experience with the challenges feel like we must be dummies, because you say it is easy.

    I do encourage you to pursue your dreams.... that's what makes it all worthwhile, and you will learn a ton to boot. Probably even turn out to be a reasonable machinist.

    Cheers,

    Peter (pjv)
  • garyggaryg Posts: 420
    edited 2014-04-01 15:41
    Idbruce
    I'm thinking that a photo of the 1st part of your framework, cut with bandsaw and sanded to be square would help to slow down the You Can't Do That
    comments.
    I'm very interested in your project.
    I think that solving the sanding machine being square and true will be the 1st largest challenge.
    Thank you for this message thread.
    I look for updates/comments each day.

    Gary
  • idbruceidbruce Posts: 6,197
    edited 2014-04-01 15:44
    Okay, temporary results are in....

    After ripping off the old table, and replacing it with my own, I have made my first attempt to true both ends of some 2" aluminum tubing. After truing, I measured all four corners end to end, and here are the results.
    Corner #1. - 5.467 in.
    Corner #2. - 5.465 in.
    Corner #3. - 5.485 in.
    Corner #4. - 5.487 in.
    First attempt... 0.020 in. maximum difference. Not too shabby, if I do say so myself. Time to add a little shimming. :)

    EDIT: Or rotate the table a smidge or both.
  • idbruceidbruce Posts: 6,197
    edited 2014-04-01 16:06
    garyg

    Thanks for the words of encouragement.

    The ends of my aluminum tubing never looked so nice :) Looks like it has been milled with a fly cutter, only the arcs get progressively smaller
  • idbruceidbruce Posts: 6,197
    edited 2014-04-01 19:02
    Second attempt...
    Corner #1. - 5.442 in.
    Corner #2. - 5.443 in.
    Corner #3. - 5.458 in.
    Corner #4. - 5.458 in.

    0.016 in. maximum difference. Getting closer :)

    Definitely a shim issue, instead of rotation.
  • idbruceidbruce Posts: 6,197
    edited 2014-04-01 19:34
    Third attempt...
    Corner #1. - 5.438 in.
    Corner #2. - 5.438 in.
    Corner #3. - 5.445 in.
    Corner #4. - 5.443 in.
    0.007 in. maximum difference. Getting her all dialed in

    OH YEA BABY! :):):)
  • idbruceidbruce Posts: 6,197
    edited 2014-04-01 20:37
    Fourth attempt...
    Corner #1. - 5.431 in.
    Corner #2. - 5.431 in.
    Corner #3. - 5.435 in.
    Corner #4. - 5.435 in.
    0.004 in. maximum difference.

    Just a couple more thousandsths to go!!!!
  • msrobotsmsrobots Posts: 3,701
    edited 2014-04-01 20:52
    Persistence. Exactly.

    The same is with those printers. You need to fiddle around until it works.

    But isn't this half of the fun when building something?

    Enjoy!

    Mike
  • Paul K.Paul K. Posts: 150
    edited 2014-04-01 22:01
    Trying to square any type of material repetitively on a piece of sand paper is not the best way. As you try to square your first piece of stock on a fresh pad it will wear the paper unevenly. As you move along in production your results will vary main reason is that you stock will vary and your paper is not square. And it will vary. Your band saw blade will dull and it will pull to one side after repetitive use. Now if you change pads and blades constantly then your good. Your going to spend a lot more money in sanding discs and bandsaw blades. It should work.

    If a milling machine is out of the question. Bridgeport is a great choice as mentioned above. The next solution might by a chop saw with a carbide blade made for aluminum cutting. My supplier for small aluminum tubes uses a saw. Nice cut.
    Another option is a grinding wheel. You can always dress the wheel so that it leaves a clean grind.
  • idbruceidbruce Posts: 6,197
    edited 2014-04-01 22:03
    Thanks for the encouragement Mike.
    Persistence. Exactly

    Yea, just a little bit more time on the sander and it will be all dialed in. Next, I will be rewiring a very old motor for the band saw. It is only a 1/4 H.P. motor, and the old manual for the saw recommends a 1/3 H.P. motor. I just hope that it has enough power to cut through some 3/4" stock, because that is the thickest stuff I need to cut. Then all I need is a cross-cut slide for pushing the stock throiugh the saw, and I should be ready, or at least I hope. I certainly don't want to be chasing down motors and bi-metal blades.
  • idbruceidbruce Posts: 6,197
    edited 2014-04-01 22:07
    Paul

    I understand your point very well, but I do not plan to go into production with these tools, I just want to create a nice prototype.
  • LoopyBytelooseLoopyByteloose Posts: 12,537
    edited 2014-04-01 22:17
    Well, the discussion benefits all that might want to DIY... not everyone has the same shop. And others might want to acquire different tools to achieve the same goals.

    I have a ton of studying to do if I am ever to sort out CNC control on a Propeller.... so I suspect no more tool suggestions.
  • idbruceidbruce Posts: 6,197
    edited 2014-04-01 23:10
    Latest results...
    Corner #1. - 5.419 in.
    Corner #2. - 5.419 in.
    Corner #3. - 5.421 in.
    Corner #4. - 5.421 in.
    0.002 in. maximum difference. :)

    I could probably get it a little tighter, but most of the stock will be much thinner than 2", which will drastically cut down the difference, so I am happy for now. However, I may need to make it better for the screw driven actuator (Z axis), since the motor will be mounting to the end of the tubing, instead of on the side, like the belt driven actuators (X and Y).

    Anyhow.... I am now off to rewire a motor, if possible. The cord and wiring is very old, and I am sure that it could end up being worse then a can of worms that has set in the sun for two days.
  • LoopyBytelooseLoopyByteloose Posts: 12,537
    edited 2014-04-02 06:06
    Have made a bit of progress with sorting out G-codes by backtracking into the history.

    Wikipedia has a listing of all the EIA RS-### standards, of which the RS-274D is one (along with the resistor color code, RS-232 ports, and others).

    I have been trying to figure out an entry point. So the big question is where to find a good exemplar of a RS-274D or the later RS-274X that applies to 3-D printers.

    Historically, there was no file format and this was all done of paper punch tape. But, likely the first adopter of a RS-274 file format was Gerber for the electronics printed circuit boards.

    The only problem with Gerber is it was intended for light exposure via a camera and for only an X-Y solution. It just doesn't seem quite right.

    The real solution is to simply use a 3-D CAD front end to produce an example in .STL or .DXF files and convert with existing conversion utilities.

    One is STL2GER or STL to Gerber, another is STL to reprap called Slic3R.

    I have OpenSCAD in Linux Debian with plenty of example files to export. So I should be able to get an adequate idea of what the normal RepRap file is like.

    With that knowledge, a G-code front end can take data from a file on an SDcard and feed it into the Propeller. It may be possible to modify Mike Starkey's OBEX object to load these files and do reprap, though it is about 50/50 that there just isn't enough hubram or cogs. 3-D printers use a 4th stepper to feed the print head plastic.

    For now, I will just try to focus on a good file format for a front end, and confirm that all the codes are normally accepted in an Arduino Uno device. Then it will be necessary to write a RS-274 parse for the Propeller.

    It is a beginning, seems like it will be a bit time consuming unless I can 'cut and paste' some stuff.

    Attached is what seems to be a good EIA RS274 reference, NOT a Gerber RS274 reference (which gets weird).
  • LoopyBytelooseLoopyByteloose Posts: 12,537
    edited 2014-04-02 06:32
    Just a quickie up-date.

    I loaded Slic3R into my Debian via the binary zip file without any problems, even though the site seemed to indicate that only Ubuntu 12.04 might be supported.

    Having gotten beyond that hurdle, there is a install menu that requires another snag. One has to choose which g-code file format your particular G-code interpreter board desires.

    We have the following:

    3-D printer file formats offered by Slic3ri five alternatives and only need ONE for a Propeller solution.

    RepRap (martin/spirnter/repetier)
    Teacup
    MakerWare
    Sailfish
    Mach3/EMC (EMC is now LinuxCNC)

    And so, I will do some more investigation as to which may be best. I am leaning toward the EMC/LinuxCNC format as it would allow the PropellerCNC to take advantage of a rather nice user interface for MDI (manual data input) if one can figure out their Hardware Abstraction Layer.

    Of course, it might be easier to do something else that just reads a file, does its job, and doesn't require and status reports going the other direction.

    Think, think, think.

    The attached g-code conversion (to RepRap file) is a proof-of-software using

    1. OpenSCAD
    2. Slic3r
    3. Debian 3.4.2

    So, there is a free software suite to take a design from scratch to your DIY Propeller g-code interpreter, the rest is programing the parser and g-code interpreter.

    And the best news is that Slic3r does NOT require a reinstall to switch file format outputs.
  • idbruceidbruce Posts: 6,197
    edited 2014-04-02 06:51
    Loopy
    It is a beginning, seems like it will be a bit time consuming unless I can 'cut and paste' some stuff.

    Very time consuming.

    Please allow me to give you a different perspective.

    One way or another, data has to get to the chip to instruct the machine what actions to perform, G-Code being the most well known and accepted code. Regardless of anything else, machine code must be present, which is all that is needed. Without embedded 3D modeler or CAD software, the original design must take place on a PC or other type of computer. And unless the embedded solution has internet accessibility, an SD card will be needed for storing either an STL file or G-Code. The most likely choice would be to just add an SD card, but I believe there are 3D printers out there with internet connectivity, however I am unsure whether it is embedded or PC. I am now assuming the use of an SD card, and as I stated earlier, data must be on that card. Since the computer has already done much of the heavy lifting, why not have it complete the rest of the heavy lifting, such as coverting the 3D model or STL to G-Code, and then just transfer the G-Code to the SD card? From there a G-Code interpreter and machine controller should be suffice.
  • idbruceidbruce Posts: 6,197
    edited 2014-04-02 06:55
    From what I have read, all the big boys in 3D printer software, each have their strengths and weaknesses. I suppose the one to use would be the one with the best reviews.
  • LoopyBytelooseLoopyByteloose Posts: 12,537
    edited 2014-04-02 07:08
    My thread above resolves getting a g-code onto an SDcard.

    Unless you have time to figure out how to make the Propeller convert STL to g-code, it is likely you will have to depend on Slic3r or an equivalent to do so.

    G-code will provide instructions to the print head one slice at a time, then move up to the next slice.

    The only great drawback with G-code is all dimensions are in floating point in inches or mm. It might be wonderful to have yet another software convert G-code into entirely integer representations and release the Propeller from the burden of doing a floating-point calculation for each X, Y, and Z dimension in real time. I am pondering if this is feasible as a means to accelerate the printer speed to the max. Of course, there would be a speed limit set so that you don't shake the printer apart.

    I have no idea of how to sort out reviews and fear that would be fickle. Just make the Propeller version compatible and better.
  • ellipserellipser Posts: 43
    edited 2014-04-02 07:26
    idbruce wrote: »
    Latest results...
    Corner #1. - 5.419 in.
    Corner #2. - 5.419 in.
    Corner #3. - 5.421 in.
    Corner #4. - 5.421 in.
    0.002 in. maximum difference. :)

    I could probably get it a little tighter, but most of the stock will be much thinner than 2", which will drastically cut down the difference, so I am happy for now. However, I may need to make it better for the screw driven actuator (Z axis), since the motor will be mounting to the end of the tubing, instead of on the side, like the belt driven actuators (X and Y).

    Congratulations. You'll be the first person in history to build a functioning 3D printer using your grandfather's rusty old bandsaw.
  • idbruceidbruce Posts: 6,197
    edited 2014-04-02 07:30
    Unless you have time to figure out how to make the Propeller convert STL to g-code, it is likely you will have to depend on Slic3r or an equivalent to do so.

    My point exactly.
    The only great drawback with G-code is all dimensions are in floating point in inches or mm. It might be wonderful to have yet another software convert G-code into entirely integer representations and release the Propeller from the burden of doing a floating-point calculation for each X, Y, and Z dimension in real time.

    Now that is a very good point.
    I am pondering if this is feasible as a means to accelerate the printer speed to the max.

    That would all depend upon just how long it takes to do each floating point calculation, the stepper driver's capabilty, and the software providing instructions to the stepper driver. EDIT - Providing the actuator can move the given load at max speed.
  • LoopyBytelooseLoopyByteloose Posts: 12,537
    edited 2014-04-02 07:51
    If this were simply a 2D stencil cutter, maximum speed might be very appealing.

    When you add in printing in 3D, the print head feed rate dictates the flow and solidifiction of the material and what is the reasonable rate of everything else.

    So I worry more about floating point bogging down the quality than the ultimate speed. Just trying to widen the target for a sweet spot.

    I suppose you can turn down the heat and the feed to hit a sweet spot, but it would be nice to have it print fast as well as with good fidelity.

    Nobody can say for sure how long the floating point conversions might take without testing. It might free up one cog for other uses. Whatever the Propeller does, it certainly won't be a complete clone of the Arduino solution.

    i suppose that some short g-code routines to diagnose and calibrate the printhead performance will be necessary -- something small and just 4 or 5 layers.

    In sum, elimination of each and every dimension and position having a floating point conversion will increase the top speed potential. But realizing that potential depends on the balanced performance.
  • PublisonPublison Posts: 12,366
    edited 2014-04-02 08:09
    Floating point calculations can be off-loaded to a dedicated FPU processor.

    Parallax still carries the 32 bit FPU

    A
    64 bit FPU in now available here or here.

    There is Propeller code available in the OBEX
  • idbruceidbruce Posts: 6,197
    edited 2014-04-02 08:29
    Congratulations. You'll be the first person in history to build a functioning 3D printer using your grandfather's rusty old bandsaw.

    Actually it was quite easy to get the sander where I wanted it, and just in case anyone wants to set their sander up the same way, this is the steps that I took:
    1. Mount the sander to a secure wooden surface with the appropiate size bit for 3/8" lag bolts, leaving about 10" of room in front of the sanding disc.
    2. Position a 8" X 8" X 4" electrical junction box (make shift table) in front of the disc for layout purposes, place a 4" long piece of 2" X 2" aluminum angle on top of the J-Box, so that the angle fits in the upper left side of the disc, with about 3/16" - 14" of room above the top leg of the angle and the same at the corner where the two legs join.
    3. Using a square determine how true the angle is to the disc, both vertically and horizontally, and note any discrepancy, because J-boxes are not created square.
    4. Drill (2) 13/16 holes at the rear of the j-box and (2) at thte front for securing the angle with 10-32 screws.
    5. Drill out the angle with 5/32" holes that match the centers of the holes you made in the j-box, and tap these holes to 10-32 threads.
    6. Drill to (2) 5/16" holes in the bottom of the j-box, in the center from front to back, and approximately 1" inside from the outter sides.
    7. Drill (2) holes that match the centers of the holes in the j-box, into the wooden support surface, with an appropiate size bit for 5/16" lag bolts.
    8. Sand the top of the j-box down to bare metal where the angle will be fastened, with about 1/2" bare metal of each side of where the angle will rest.
    9. Scuff up the lower outter edge of the leg of the angle with sandpaper and also scuff the back corner, for adding JB Weld, so that it will be more rigid.
    10. Using appropiate size washers as shims, as determined by the discepancy noted in Step 3, fasten and adjust angle with washer shims to closely square the angle to the disc vertically.
    11. Apply JB Weld to the outter edges of the angle, making sure that you do not get any epoxy on the inside of the aluminum angle. This helps to ensure a semi-rigid platform, just in case the shim washers rattle loose from vibration. So try to get it under the angle where there is sufficient clearance.
    12. Place several strips of paper under the j-box for shims, dependant upon vertical alignment discrepancies and attach the j-box to the wooden support surface with (2) 5/16" lag bolts, fender washers, and lock washers. Snug the bolts down and check vertical alignment, then loosening the bolts and removing paper strips to achieve the desired tolerance.
    13. Once vertical alignment has been achieved, simply loosen the bolts again and rotate the j-box left or right to achieve the desired horizontal tolerance, providing their is enough slop in holes you drilled to allow a minute measure of rotation.
    14. Firmly retighten bolts.
    EDIT: Actually I was wrong about the first step. The first step is to remove the rickety table, so that it may be replaced with a solid and true structure.
    EDIT: You also want to remove any knockouts from the top and the bottom of the j-box and then debur the knockout holes to remove any imperfections that may cause misalignment.
    EDIT: This procedure was performed on a Craftsman combination belt/disc sander, with a 6" disc and a 4" X 36" belt. Your size and brand of sander may vary, so alter the plans as necessary.
  • LoopyBytelooseLoopyByteloose Posts: 12,537
    edited 2014-04-02 08:52
    Publison wrote: »
    Floating point calculations can be off-loaded to a dedicated FPU processor.

    Parallax still carries the 32 bit FPU

    A
    64 bit FPU in now available here or here.

    There is Propeller code available in the OBEX

    Not sure about this at this point, but will certainly keep it in mind. It does look powerful and feature rich.

    I am just trying to get my head around the RepRap file header. They general claim is g-code files don't have a header, but that is obviously no longer true.

    There is a of of comment lines with ; and aside from the expected G-codes and M-codes there are E codes and other letters to run down. Details, details, details, oh my ! ! !

    I get the feeling that I am going to have to code from scratch. I might even learn something.

    It certainly was worth the effort to run down an actual g-code' file and everything that Wikipedia tells you is general.

    ***********

    I just am beginning to dread that one of these file formats is copyright material. Please tell me if someone knows.
Sign In or Register to comment.