Welcome to Our Community

Some features disabled for guests. Register Today.

InDoreXY

Discussion in '3D printers' started by Shortyski13, Apr 16, 2017.

  1. Shortyski13

    Builder

    Joined:
    Mar 28, 2017
    Messages:
    40
    Likes Received:
    9
    Shortyski13 published a new build:

    Read more about this build...
     
  2. Shortyski13

    Builder

    Joined:
    Mar 28, 2017
    Messages:
    40
    Likes Received:
    9
    I attached an excel sheet with my BOM and prices. I could use help reducing the costs. Does anyone see any place to save some money?
    One option I'm debating is instead of going for E3D V6 + Titan, go for the Aero Titan instead (save about $15 and it looks not as cantilevered), or I can go full cheap and buy clones and save $65. Still not even settled on bowden vs direct drive.
    Also, does anyone know a good place to find very large gt2 closed loop timing belts? Has to be over 2200mm in length (1100 teeth). SDP is the only place I can find that has one, but it's $12 + $10 for order being under the minumum cost ($30) + $10 shipping. This makes it tough to settle on. I could connect an open ended belt, but feel like that won't hold well or produce great results.
    I also still need to source the aluminum plate for my bed. I'm hoping recycle centers or junkyards will have something, anyone have any luck with such things?
    I was also looking at something cheaper than duetwifi to reduce cost, but not sure if I can with a nema 23 and going 24v (for future potential mods). Also i don't have to purchase a screen with it, which is nice.
    Anyone have spare nuts and bolts lying around you are trying to get rid of and clean-up the shop? ;-)
     
  3. Shortyski13

    Builder

    Joined:
    Mar 28, 2017
    Messages:
    40
    Likes Received:
    9
    Also, anyone have any other thoughts or concerns about this build? My biggest concerns currently are: A) how flimsy the V6 hotend looks in the Titan (looking at the model). How can this be tight and not wiggle? B) plastic brackets for connecting the bed frame to both the 3x12mm lead screws, and the frame. Though the bed isn't moving fast and only in 1 axis, so I feel it should be fine.
    Any thoughts on these? Any other concerns or comments or ideas?
     
  4. Shortyski13

    Builder

    Joined:
    Mar 28, 2017
    Messages:
    40
    Likes Received:
    9
    <3 Yes! Now I just to source that build plate. probably start looking around junkyards/recycle centers this week or next.

    Another thought I was having, in addition to the rest mentioned above, is that it would be easy to add an additional 500mm to the height of this. If I kept with 12mm lead screws, and 2040 vertical extrusions, it would probably cost an additional $35. However I'm kind of wondering if I should increase to 4040 extrusions and wider lead screws (which would then require larger pillow blocks, nuts, pulleys etc.), which would unfortunately really throw the price up there and leave me with my current dimensions.
     
  5. DracusMage

    Builder

    Joined:
    Jan 16, 2017
    Messages:
    4
    Likes Received:
    0
    If you haven't gotten too far in the build already, I have some suggestions. I'm designing a similarly doorway constrained printer, and I've got a couple of engineers who have been giving me free engineering help.

    First suggestion is GT3 belt instead of GT2 it will increase the cost, but greatly reduce the risk of skipping. Second, instead of a Nema-23 for my Z, they have suggested I use a Nema-17, and just gear it up a little. This both increases the accuracy of the bed while increasing the torque from the smaller motor, compensating for the size. It means it will travel slower, but how fast does the Z need to move? Additionally it means using a 1.8 degree for the Z is good enough, shaving a few dollars off the total.

    I would not scale up your lead screw, because if the screw is doing anything other than pushing straight up and down, there is a problem. I'm doing 2 8mm screws, one on each side of the middle bed with C channel to guide, but your rollers on the corners should be good too. If you are worried about the stability, longer roller plates (so the wheels are farther apart vertically, and possibly more of them) will help. It means more wasted space to the top and/or bottom of the bed, but it will be more dependable. I'm only going to have at most 29" length screws (including what is in the blocks), but that's already going to be mostly wasted space. The Z layers are the weakest points, so how tall do you want to print something? I'm doing a 20" square bed, and I'm having a hard time coming up with an object that I would need to print that's larger than 20" that I wouldn't want to cut into a few parts rather than printing it whole, let alone whole and vertically where the risk of failure is higher. How much plastic is ruined if your print falls over on the 3rd day of printing?

    8mm trapezoidal screws are more than capable of running the 50lbs of bed and plastic (around 20lbs for a .375" aluminum plate a little larger than your printable dimensions, plus a sizable margin for the extrusions, rollers, glass sheet, heater, and whatever you are actually printing) that you are likely to need to move. Plus you get to divide that between the screws you are using, so for your design we are looking at less than 25lbs per screw. Charts I've found for column strength for acme screws show one screw of that size can do around 200lbs with a plastic nut, but only at around 15" between bearings - but these charts also don't usually bother going below 100lbs. Pretty clearly a 3d printer is using these screws at such a small portion of their capacity that industrial manufacturers don't even bother testing it. In fact when I suggested that maybe I should scale up my rod to 12mm just to be safe, the engineers laughed. A 12mm trapezoidal screw fixed at both ends with quad bearings is rated for over 600 lbs with a plastic nut and almost 3 feet between bearings, or over 1,000 lbs of dynamic load with a bronze nut, per screw; so 3 screws with bronze nuts are good for over a literal ton of dynamic load without becoming unusable due to compression if the screws are around 33" long.
     
  6. Shortyski13

    Builder

    Joined:
    Mar 28, 2017
    Messages:
    40
    Likes Received:
    9
    You bring up a lot of good points. I'm pretty set on a 1500mm height, 1230mm Z-build height. This will allow me to print wind turbine blades in 1 go for a diy wind turbine. If it seems a bit shakey, I can screw it to a wall. I am actually already planning to use 3GT belts but am having a bit of heartburn on the price.
    How does using a Nema17 instead of Nema 23 improve the accuracy of the bed?
    Up-sizing lead screws isn't to allow it to handle higher loads without breaking, its to minimize any deflections while under load, that could push the bed or frame a little bit and cause banding in the print.
    I haven't updated this post yet, but will shortly I presume with new pics and specs.
     
  7. DracusMage

    Builder

    Joined:
    Jan 16, 2017
    Messages:
    4
    Likes Received:
    0
    There isn't anything about a 17 vs a 23 that improves accuracy. The improvement is from gearing the motor - if you do a 2-1 gearing it means that a 1.8 degree motor (of whatever size) is effectively a 0.9 degree. If you do a 3-1, a 1.8 degree performs like a 0.6 degree motor. Plus gearing the motor means higher torque, so you don't need as much motor to do the same job. So a 20 tooth gear on the motor and a 40 tooth gear on the screws means that you can use a Nema 17 1.8 degree with the same gradation as the Nema 23 0.9 degree with the same gears on the motor and the screws. Slower speed, but the same granularity with the cheaper commodity 17 that you wouldn't have to under power from the controller. A Nema 17 0.9 would become an effective 0.45 degree motor with the same 2-1 gearing. A skipped step is a smaller problem, and a microstep in't

    8mm steel is pretty strong, but I would be anxious if they were more than a meter with only end supports. Especially with a printable 1230, you will want probably want at least 1400 for overhead, adjustments, build plate, etc, that's going to be a lot of screw flapping around. You could look at using the C-channel extrusions with a platform and nutblock to add a support on the screw that moves with the bed, but then you are just hoping that the C is straight. You could look at replacing your stronger corners with Cs around your screws and weaker stuff at the corners, but either way that's going to cost a lot more. 12mm screws will be stronger, but it's still a long span.

    As far as shaking goes, I'm less concerned about the printer frame, and more concerned about the part. I have a delta with a fixed bed and I don't even print things more than a half dozen inches tall just because a part acts as a lever against the build plate, so a little snag can exert a lot of force to pop itself free or snap itself in half, let alone a moving bed (even one that is just moving with gravity). It's not terribly likely, but a little curling or a little ooze could wreck a day's worth of printing - I've failed prints only 3.5" tall because a corner lifts and the nozzle knocks into it. Might you be better off making a printer that is longer in the X or Y? It's a lot more space required to put the thing down in your house and the printer bed would cost more, but then at least you would be able to print the turbine with the strongest layers along the length of the blade rather than the weakest ones. Your screws would be much shorter, but you would probably need more than 3. Another thing to consider is the material you will use to print the blades. If it's ABS, that's going to be a long print, giving it lots of time to curl free of the bed or between layers, and enclosing a 1.5 meter printer is either going to be expensive (acrylic would be around $400-600 US from Home Depot for single sheets large enough to prevent a seam) or ugly (plywood and/or trashbags). It seems sad to cover a shimmering aluminum monolith in plywood, but not enclosing it will mean some materials are out of reach.

    All that said, I'm an IT professional, not an engineer (although I know some engineers), so you shouldn't do anything based on my say-so alone. I'm enjoying thinking about the problems of a printer I'm not likely to have a need for so I'm happy to keep chiming in, but if I'm making you sad or you otherwise don't like what I'm saying, let me know and I'll keep it to myself.

    I will, however, defend to the death [note: not actually willing to die over 3d printers] that you should use 3GT belt over the GT2. It costs about $0.75 per foot more, but that extra cost is nothing compared to slipping a belt tooth; not just because it ruined whatever you were doing, but also because then you have a damaged belt that you get to replace and may not have a spare on hand. I'm still looking for a good place to buy cheap closed 3gt belts that are long enough for the Z, but I'm sure they are out there.
     
  8. Shortyski13

    Builder

    Joined:
    Mar 28, 2017
    Messages:
    40
    Likes Received:
    9
    "All that said, I'm an IT professional, not an engineer (although I know some engineers), so you shouldn't do anything based on my say-so alone. I'm enjoying thinking about the problems of a printer I'm not likely to have a need for so I'm happy to keep chiming in, but if I'm making you sad or you otherwise don't like what I'm saying, let me know and I'll keep it to myself."
    I love brainstorming like this and getting different opinions/ideas/suggestions from different people! Keep at it! I appreciate the concern though, haha.

    I'm actually already gearing the Z-axis up from a 20T drive pulley on the motor to 48T pulleys on the screws, so yes a nema 23 is probably overkill, especially since it is a .9 deg/rev one on both resolution and torque. I'd kind of like to keep it though for if I convert it to a cnc milling machine for wood (and hoping 2 nema 17's on x/y working together will be sufficient). This means I could have some heavy hunks of wood on the bed. That resolution just has me drooling though too. Also, I'm currently planning on a 6mm GT2-2mm belt for the Z-axis because there won't be any quick back-and-forth motions like on x/y so once it's tensioned, it shouldn't really stretch more, or at different rates throughout the print, and it's slow. I'm debating changing it to 9mm, but I have to see what there are for pulleys for that as well as belts, and I feel it is unnecessary there. Now that I think of it, I think I was only able to find a 6mm closed-loop belt for that length. For X and Y, I agree that I should probably look at a 3GT, as I said before. The biggest thing I'm concerned about is that I read it has more backlash than GT2-2M. Have you had any experience or knowlege comparing the backlash of both GT2-3M (3GT) and GT2-2M?
    You are deadnuts on point with your "1400mm" length, and I had picked that out, but it seems like the vendor that makes it no longer does that length so I'll be going with 1500mm length. 12mm diameter.
    I don't believe that the part would shake with the bed only moving in the z-axis; remember, this is a CoreXY design. To your point, a curled up piece could still definitely get snagged on the nozzle. That is a risk I'll have to weigh after having more experience with the machine. worst case, I print large parts in pieces and glue them together. Best case I'll have the capacity to do it in 1 go.
    You bring up a good option that really had me thinking: to essentially turn it sideways. One problem I do have is space in my basement (and even less if I decide to leave it upstairs out of the higher humidity), so that definitely is an issue for me specifically. Also, imagine how much longer those x/y belts would be and the increased cost of those. It would probably add another 2 meters to the length of each, adding cost and inaccuracies. For these reasons, and mostly the first honestly, I'm going to keep with the tall one. To your point about the turbine blades being loaded parallel to the layers, aka the weak direction, I may have to add a thin sheet or two (or a rod?) of metal inside of them to prevent them from deflecting too much. I'll figure that out when the time comes.
    The enclosure-on-a-budget is a concern of mine, but I've mostly figured it out. I'm keeping an eye on craiglist and waiting for someone to put up a metal filing cabinet for free on it, in my area. There was one 2 months ago I should have grabbed :-(. I just got an angle "grinder" with metal cut-off wheels (I'm not attributing this to the cost of the project since I may likely use this for other stuff in the future and was still only 45 bucks), so I plan on cutting off the side panels of those metal filing cabinets and using them on 3 sides of my printer to enclose it. This will also help hold it structurally too. I also could use glass from doors on craigslist too, but I feel like it, while looking quite a bit better, it wouldn't hold the heat in nearly as well. The front, I'm debating using just straight up glass from one of the many glass doors on Craigslist, or I can go plexi-glass and pay an arm and a leg but would be easier to work with and be better for keeping the heat in. I haven't figured that out for the front/door. Any suggestions there?
    Also, for outdoor parts (like wind turbine parts), I'm currently planning to use ASA, which from what I've read, is essentially ABS that's more UV stable and prints just like ABS. I may need to add in a heating lamp for larger parts to keep the chamber warm.
    I just posted a pic of the 3D model of the 1500mm version of the printer (still need to update the pulleys to be all for 9mm belt). It's in the Files and Drawings section, which apparently is where I should have put the rest too...oops. I'd post my new BOM, but it's a mess right now with Alternates listed for pretty much everything and numbers everywhere; I have to clean it up.

    P.S. I've always wondered why acrylic/lexan is so expensive. I just can't figure it out; I've needed it for other projects too.
     
    #8 Shortyski13, May 31, 2017
    Last edited: May 31, 2017
  9. DracusMage

    Builder

    Joined:
    Jan 16, 2017
    Messages:
    4
    Likes Received:
    0
    I use GT2-2M on my Delta printer, but it's only a few feet tall with about a 6" diameter build area, so it's tiny potatoes. I have helped fix a 3d printer that drove the nozzle into the bed hard enough to skip the belt (the motor drivers were turned up too high so they kept spinning rather than skipping) and the belt was a loss - the skipped teeth were wrecked. Some were partially ripped so the tops would flap a tiny bit and a few looked like they had been torn downwards on the tooth. I wish I had taken a picture. This is a rather extreme case, and if the drivers are tuned correctly to stall under too much load you will probably never see it, and it may have been deficient belt contributing, but it can happen.
    GT2-3M is used on a system that goes extremely fast and precisely back and forth with multiple independent carriages. So fast that it's designed so that if the system loses track of one of them or is just slightly wrong about the position and doesn't slow one down in time and they crash into each other, it shatters a purposefully weak and easily replaced piece rather than destroying the whole (milled aluminium) carriage. 2M is probably sufficient for our actual needs, but 3M is so much better.
    As far as indexing accuracy I've not had any comparable experience - most uses for timing belts don't need to be accurate within .1mm to be good enough. I did have a carriage on GT2-3M belt where I could touch it a couple weeks ago, and it had no play that I could notice when I wiggled. But in the end what does that mean? It might have a half millimeter of wobble and I'm not sure I would have noticed it, but it certainly didn't move much if at all. It wasn't powered so I didn't have the belt being held tight, but I didn't notice any slack in the movement. It's also a thicker belt, and I understand less likely to stretch, so even if it does have a tiny bit of wiggle that the 2M doesn't, it will still have the same tiny wiggle in a couple years, while the 2M might have another eighth inch of belt per yard.

    My design is going to have about 18' of XY belt, so upgrading the belt for that to the better stuff is less than $15. So maybe a $20 premium gets me better belts everywhere, and in exchange I don't have to worry about the belts failing essentially ever under the loads I'm going to be putting them through.

    As I've been designing mine I've slowly settled on the realization that I would rather have a printer that has $50-100 worth of over-engineering (or as my grandfather called it, "engineering failure") and not worry about my belts, or my frame, or whatever. If I were designing this for mass production, saving a few dollars in parts to add a 1-5000 risk of failure would be a great trade-off, however I'm making my printer just for me, and if I save $50 but I turn out to be the 1-5000, I just lost weeks of time and I may still have to upgrade the part to make it more stable. My grandfather died when I was young, but he told me something that stuck with me: Whether an item doesn't do it's job or does it's job too well, it's an engineering failure - if you have to pick one though, which do you want? The Mars rovers have all been engineering failures because they either didn't do their jobs, or they didn't meet the mission requirements (mostly a couple weeks mission), but would you rather have one that never communicated, or one that has been running for more than a decade after it was supposed to stop?

    It looks like you have a couple of your screws outside the frame of your printer. That will make it harder to enclose because you will not be able to use the frame on that side for the direct mounting of panels. If you're going to use metal, you will need to worry about the heat conductance and thermal mass. Probably not issues with steel, but you may want to consider putting some insulation in to keep it from being a problem. The best reason to use acrylic is that it's extremely non thermally conductive. At room temperature, Acrylic is 0.2 W/(m K). Stainless steel is 16, Carbon steel is 43. Glass is about 1. For perspective, Aluminum is 205, and glass wool (insulation) is 0.04, and Plywood is 0.13. (I love The Engineering ToolBox: Thermal Conductivity of common Materials and Gases). A roll of $20 reflective insulation with a little duct tape to seal any air leaks would fix it though, and could be slapped on after the fact if it turns out the steel is letting the heat go too quickly. At least the steel sides will keep the printer looking sexy and ominous.

    If you use glass you should consider using a lot of silicone to both trap the air and keep the glass from chattering against the frame. The quick XY motions will cause some vibration, and the last thing you want is for a 4 foot piece of glass to shatter because it bonked an aluminum extrusion one too many times.
     
  10. DracusMage

    Builder

    Joined:
    Jan 16, 2017
    Messages:
    4
    Likes Received:
    0
    I'm not sure how your day has been, but mine has been boring enough that I started reading white papers and a single college thesis (not all of it, only about 20 pages) on backlash in belts, then a bunch of scribbles, notes, and a little math (so you can imagine how the day before this has been).

    GT2-3m belt is still the way to go on the Z. I don't think it's ideal for the XY though.

    Page 7, figure 10 "Comparison of Positioning Errors of Various Belts" (Timing belts | Belt (Mechanical) | Transmission (Mechanics)) does a direct comparison of 2mm GT belt to MXL, and 3mm GT to 3mm HTD. That's not particularly interesting for us, but it does allow us to see a vague comparison of the positioning error between 2mm and 3mm GT belt. It's not a perfect comparison, because the 2mm belt is wider and a much lower tension (1.8 lb vs 6.6 lb) and the gearing is different, but it tells us that the backlash for the 2mm belt is less than 0.025mm (way less, like a quarter of that). The 3mm belt however is above 0.25mm, and closer to 0.38mm which might be due to the higher tension on the belt in the test, or it could be an inherent design tolerance issue of the belt. With a cartesian printer that might be an acceptable margin, but with an XY printer, both motors control every direction change, so it's a lot more back and forth and small issues that can be extremely hard to fix, and I don't know of any firmware that supports backlash compensation for CoreXY (and only Repetier supports it for cartesian, as far as I know). Fixing it in the slicer will just move the issues away from the perimeter to other areas.

    For the Z however, no amount of backlash is too much that it can't be corrected for with a tiny amount of configuration. If your printer can park the nozzle(s) of the hotend(s) off the build surface, homing to a minimum endstop that is past the nozzle then dropping the bed back down to the correct start print height can be used to fix any normal and even some rather egregious amounts of backlash, pre-loading the belt and screws in the correct direction. If you have perfect information about the printer you can even make it preload in the correct direction without moving the build plate (just "moving" it enough to eat the slack, not to actually change the position), but that assumes a level of familiarity with the hardware that few are likely to have short of hours of testing. It's possible that the backlash is slowly released because friction with the belt is enough to turn without full tooth engagement which means you can't fix it before printing, but if it's slow enough, we are still only talking about a backlash of between 0.025mm and .06mm of Z height total (2mm leadscrew pitch, 2-1 gearing up to the screws). I don't know about you, but if my printer is accurate enough that a part squished by less than 0.06mm of Z in total is an unacceptable amount of inaccuracy then I'm going to be too busy doing cartwheels instead of walking and high-fiving strangers I meet on the street to have a chance to fix it.
     

Share This Page

  1. This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
    By continuing to use this site, you are consenting to our use of cookies.
    Dismiss Notice