New Machine Build Fast Cartesian 3D Printer


Page 1 of 2 12 LastLast
Results 1 to 20 of 30

Thread: Fast Cartesian 3D Printer

  1. #1
    Member
    Join Date
    Jul 2011
    Location
    UK
    Posts
    73
    Downloads
    0
    Uploads
    0

    Default Fast Cartesian 3D Printer

    Hi All,

    This actually started in a thread where I asked for advice specifically on stepper motor choice. However the discussion evolved in to more general details of the machine so I'm moving the chat over here. I'd welcome constructive input. Although I have already purchased most of the drive parts, I have still to build the frame and assemble it.

    So far this is my design. Over-all size is about 550mm^3.

    Frame is welded 40x40mm steel with concrete and SBR (latex) admixture core for damping.
    1mm Steel panelling (not pictured) welded on all sides for bracing except front and top.
    Gantry is 30x30mm two aluminium tubes with a constrained (layer damping) PU core.
    16mm dia. 20mm pitch ball-screws with 4-start thread.
    MGN 12mm profile linear rails on X and Y.
    X and Y drive is Nema 23 Leadshine 57CM13 with 60V TMC2160 drivers.
    Bed is 10mm or maybe 16mm cast ali tool plate, face machined for flatness.
    Z rails are 16mm round rail with ball-bearing slide blocks.
    Z drive is dual Nema 17 0.42Nm with integrated T8 lead-screw.

    Fast Cartesian 3D Printer-3d-printer-4-jpg

    I'll respond to a few existing questions below.

    Quote Originally Posted by joeavaerage
    Did I read that correctly, 4 starts at 20mm pitch so 80mm linear travel per revolution??
    One rotation produces 20mm linear travel. However there are 4 threads wrapped around each other. Visually it looks like a 1605 ballscrew but it's not quite. I didn't spec it like that, but that's how it came! AFAIK it just means there are more balls in the nut for support and redundancy. I'm not sure if it improves backlash??

    Quote Originally Posted by peteeng
    Are these [Z axis] plain bushes or linear bushes with ball bearings?
    The Z axis uses ball-bearing bushes. So stick-slip I hope won't be an issue. 3D printers don't need to handle any cutting forces or significant weight on the z-axis so 'surface-on-surface' ways or oiled bushings are not commonly used unlike in mills or presses. Many printers only support the print surface on one edge! Horrible IMO.

    Fast Cartesian 3D Printer-2-moment-3d-printer-build-chamber-jpg

    Quote Originally Posted by peteeng
    The concrete you are adding provides mass for damping (which is a valid approach) but it does not dampen via friction or viscous damping or material hysteresis.. The gist of it is that in these sort of structures the strain delta from the steel to the core is so small (due to the steel being so stiff and being on the outside of the structure where the strain is, the core does not internally deform enough to contribute to hysteritic damping.
    That's interesting to know. I was going to use concrete with SBR which is a latex addative used to improve damping (and water-proof) in concrete structures. I got the idea from here - https://www.researchgate.net/publica...city_of_cement

    This paper 'Principles of Rapid Machine Design' is super informative on damping beams (see captor 4).

    Although it is investigating layer damping, it does mention the loss factor of polymer concrete.

    Fast Cartesian 3D Printer-untitled-1-jpg

    If, as you say the internal damping of the concrete makes little difference when cast inside a steel beam, so be it. Maybe I can through some damping sheets in there with the concrete to constrain the damping sheet. However I'm conscious of how practical it will be to add the damping materials to the frame once constructed. Capping off the ends and pouring is pretty easy, but individually adding damping to each beam before welding sounds tricky.

    I actually work mainly in acoustics, so have the means to measure and test damping and make comparisons. Maybe I will do a small test run to compare basic concrete fill vs. concrete with SBR addative (and possibly 10mm granite gravel?)

    Similar Threads:


  2. #2
    Member peteeng's Avatar
    Join Date
    Jul 2018
    Location
    dum dum
    Posts
    6248
    Downloads
    0
    Uploads
    0

    Default Re: Fast Cartesian 3D Printer

    Hi Sash - If the structural element is all concrete then the structural hysteresis is significant as the load and strains are born by that material. One paper I read found by mistake that the filling they used had uncoupled from the steel wall and this produced better damping then the filling that had stayed coupled. ie the friction between the filling and outer was the significant damping factor. This is how leaf springs damp. I think damping a structure is a very complex subject. I do modal analysis on structures and just try to remove pesky low freq wobbles. This means making walls of tubes thick (thin metal pants and oil cans), use triangles where ever possible, use known damp materials (but they have to be in the load path) , use bolted joints vs welded joints, do not cantilever or have free edges. Make asymmetric structures eg if adding braces use an odd spacing because if you use an even spacing you create perfect conditions for vibration.

    Re "granite" spent a lot of time in other forums on this sort of thing. Concrete shrinks and will micro crack improving damping by allowing internal movement and creating friction at interfaces (if its in the load path). So also look at plastic or metal fibres as additives. The more materials in the mix the damper it will be. How will you determine dampness via acoustic methods? Sounds interesting... Peter

    The way they test mechanical damping is to make a cantilever of the material and vibrate it. They measure the frequency decrement and calculate the "zeta" of the decrement. I'm about to make a rig that uses a large steel ball bearing dropped onto the material and measure its rebound. The restitution ratio should be proportional to the internal damping of the material. Peter

    Attached Files Attached Files


  3. #3
    Member
    Join Date
    Jul 2011
    Location
    UK
    Posts
    73
    Downloads
    0
    Uploads
    0

    Default Re: Fast Cartesian 3D Printer

    Hi Pete,

    In the acoustics field we are greatly concerned with testing the damping of materials and structures. I would free-hang a test beam, then place an 'exciter' (like a speaker without a cone, that attaches to a surface) on one end and an accelerometer at the main nodal point. The decay of energy in the beam can be recorded across a wide frequency range and plotted in various ways such as a water-fall plot.

    Fast Cartesian 3D Printer-waterfall-plot-showing-decaying-mode-50-hz

    I would plan to use expanding concrete to fit tightly in the beams. The concrete would be part of the constraining layer really, while the latex additive, granite gravel, rubber chippings or butyl sheet etc... would be the constrained damping component.

    Simon



  4. #4
    Member peteeng's Avatar
    Join Date
    Jul 2018
    Location
    dum dum
    Posts
    6248
    Downloads
    0
    Uploads
    0

    Default Re: Fast Cartesian 3D Printer

    OK that's effectively what the lab does to the std just uses a cantilever vs its natural vibration shape. Its the decay your interested in. Expanding concrete. Hadn't heard of that before. very good. Peter



  5. #5
    Member peteeng's Avatar
    Join Date
    Jul 2018
    Location
    dum dum
    Posts
    6248
    Downloads
    0
    Uploads
    0

    Default Re: Fast Cartesian 3D Printer

    Hi Sash - some time ago I made this model. Left to right. All tubes same outside diameter 100mm. thin tube 6mm thick thick tube 13mm thick. They are supported as a cantilever and I ran a modal analysis on the set of tubes. Generally with machine structures we try to push the vibration modes to higher frequencies. I placed 30kgf at the bottom of the tubes. The thick tube is stiffest and deflects 0.49mm, the next stiffest is the filled tube at 0.69mm and then the thin tube is least stiff at 0.87mm. So I'd pick the thick tube over the filled tube from ease of making and cheaper etc and stiffest. The interesting thing is the vibration modes...

    LHS steel filled with concrete total weigth 40kg
    middle steel tube same size as LHS weight 20kg
    RHS steel tube 40kg (thickness picked to be same weight as filled tube)

    Now the first mode is 33htz and it's the filled tube. The second mode is 38htz and is the thin tube and the third mode is 42htz and is the thick tube. The filled tube being the lowest freq is a surprise. But using this info the thick tube wins all round. Now damping is difficult to model as we need to know the damping factors for the concrete and the friction at the interface etc. Plus it has to be solved as a transient dynamic model which takes a bit to set up an solve. Been meaning to get this done when I have some time. But this info sort of fulfills the "beware of unintended outcomes" saying.

    If I look at the stress/strain in the filled tube the steel carries 8MPa at the max stressed spot and the concrete under this carries 0.78MPa which means its strain differential at the interface is tiny. So there's not much happening in the concrete thats why I think conceptualising the concrete improves damping is fallatious. Plus this cantilever deflects 1mm in this condition and we do not expect our machine structures to deflect that much in practice so strain ratios are even less, especially internal to the tube.

    That's why I think its the mass addition only that improves "damping" but people do this with no before and after testing. Plus this sort of modelling shows you may lower the freq of the structure vs increasing it which is not good.

    So some small scale tests on damping would be good, that's why I went with the dropped bearing test and hopefully you publish some info with your tests....I intend to make some carbon fibre machine parts and glass fibre parts which will be very damp...

    What I have learnt from designing many machines and analysing them is 1) Being involved with lightweight structures for many years usually drives me to thinner and thinner structures but this does not work in machine design (panting, oil canning, lozenging, edge flutter, local wobbles, shear wrinkles etc all are there 2) so large thick structures do better cheers Peter

    Attached Thumbnails Attached Thumbnails Fast Cartesian 3D Printer-mode-1-jpg   Fast Cartesian 3D Printer-mode-2-jpg   Fast Cartesian 3D Printer-mode-3-jpg  


  6. #6
    Member
    Join Date
    Jul 2011
    Location
    UK
    Posts
    73
    Downloads
    0
    Uploads
    0

    Default Re: Fast Cartesian 3D Printer

    I do get your point. However it's worth considering that if the thin tube and thick tube each have a similar resonance frequency, it will be easier to damp the thin walled tube. It has less inertia and mass, so the same amount of damping is more effective. To put it another way, the thicker walled tube has higher mass and so the resonance becomes higher Q.


    Fast Cartesian 3D Printer-2-jpg

    I wish I had some various tube thicknesses here so I could do the test. I'd have to buy them just for the sake of trying it.

    Given what you say about the steel skin of the beam carrying most of the strain, it makes sense to line the inside of the tube with a damping sheet, weld it all together, then fill with concrete to constrain the damping against the tube.

    Attached Thumbnails Attached Thumbnails Fast Cartesian 3D Printer-modalsim-jpg  


  7. #7
    Member
    Join Date
    Jul 2011
    Location
    UK
    Posts
    73
    Downloads
    0
    Uploads
    0

    Default Re: Fast Cartesian 3D Printer

    I suddenly thought, maybe a piece of information that is missing from this discussion is the frequency bandwidth of excitement. If we consider a steady state source such as a spindle rotating and cutting chips at a fixed 10,000 rpm we have a single frequency to deal with and we be certain that it is either above or below the resonance modes of a structure like the machine frame.

    However, it's important to realise that an impulse, such as a mass changing direction (moving gantry accelerating or decelerating) includes a broad frequency range. Theoretically, an instant change of direction is a perfect impulse and will include all frequencies from 0Hz to light-speed. In practice, the upper bandwidth of the impulse will be limited by acceleration/deceleration parameters, but it will still include all excitation frequencies between 0Hz (constant motion) and the upper bandwidth.

    This means it may not be so easy as it first appears to keep a structures FS above the operational range.



  8. #8
    Member peteeng's Avatar
    Join Date
    Jul 2018
    Location
    dum dum
    Posts
    6248
    Downloads
    0
    Uploads
    0

    Default Re: Fast Cartesian 3D Printer

    Hi Sash -The two main inputs into the structure are 1) the motor "steps" as these are impulses. In another thread spent a lot of time working through this in an extruded frame design and found that the ends of the extrusion and the small features in the extrusion where vibrating. The solution was to put screws into the ends to close out those free edges and to use a plastic isolator under the motor mount. 2) the tool cutting impulse, chatter. This is a chicken and egg problem as well. Is the chatter due to inadequate stiffness? Most likely or is the chatter due to inadequate damping? I think the damping is a second order issue here.

    It is a complex area. But there are many steel and aluminium machines out there that are good. Re damping the tube. The tube may never vibrate in a fashion that affects the machine. That's my point, people do all sorts of things to second guess vibration prevention and have no clue if it was successful or not. Only solve problems that are real, make machines much stiffer then you think. No-one has ever complained that a machine is too stiff... Peter



  9. #9
    Member
    Join Date
    Jul 2011
    Location
    UK
    Posts
    73
    Downloads
    0
    Uploads
    0

    Default Re: Fast Cartesian 3D Printer

    I had a look today and I have some thin and thick(er) wall 40 x 40 steel tube. I'll cast some concrete in each of them and see what the difference is. I'll report back in a couple of weeks hopefully.

    In my own CNC router (not high end) the main cause of artefacts on the cut is a sudden direction change of the gantry, say cutting a square corner at 5,000mm/min. ~50Kg gantry dumping all its inertia in to the frame as fast as possible gives quite the energetic burst for the frame to burn away. I do have concrete with latex additive in the 4 main columns, but I don't actually know how much difference it makes since I didn't do comparisons. I did find using a modal sim helped me determine the best bracing which was very beneficial (I re-iterated it after the first build and saw a big improvement).

    It's interesting what you say about the stepper impulses. They should not actually step when driven by a powered signal. The waveform *should* be a sine-wave and lead to a smooth rotation between each magnet pole, or partial pole if micro-stepping. This is not always the case in practice. Things might have changed in the ~6 years since I built my router, but I don't think much effort is made to get a good smooth waveform from the drivers to feed the steppers - rather many focus on efficiency to reduce heat from the driver. In comparison with 3D printers I guess it just isn't so critical because the high mass of a router / mill will kind of 'low pass' noise on the motor drive. Going high-end of course AC servos are used which I would assume have a nice clean drive.

    When we talk about 3D printers, I think the steps of the motors and drive signal cleanliness is more apparent in the finish of the printed item, probably because the machine can more directly respond to small fluctuations from the steppers.

    This is a big reason I was put off using the affordable closed loop steppers from leadshine etc.. because I've no idea what the drive quality is like. I went for Trinamic drivers becasue they have shown their capability in the 3D printer world at reducing all kinds of noticeable steps and noise from the motors. You may find this short video interesting (and impressive!) -

    By the way, I hope it doesn't come across like I'm arguing with you regarding damping. I'm just throwing ideas in the air to see what comes out since it seems like we both come at this complex puzzle from a different background.

    Simon

    P.S. Regarding your point 2. The tool will produce sharp impulses when the tooth hits the material - chatter or none. Chatter is feedback from the frame unable to dissipate the energy and it just gets worse! Whether chatter is caused by the sharp impulse of tooth impact with the material, or by the sustained drag on the tool as it moves through the material, I've really no idea. Having built my own semi-pro router I marvel at how deep a cut some of these top end machines can take and still get a mirror finish cut! That's voodoo enginearding!

    Last edited by ssashton; 06-08-2020 at 03:03 PM.


  10. #10
    Member peteeng's Avatar
    Join Date
    Jul 2018
    Location
    dum dum
    Posts
    6248
    Downloads
    0
    Uploads
    0

    Default Re: Fast Cartesian 3D Printer

    Hi Simon - All discussion is good here I forgot you are making a printer so contact excitation is out. Does not the mechanical principle of a "stepper" mean that there always will be an impulse? In regard to your router changing direction this is a motion controller issue. Very good motion controllers control accelerations and jerk. If you do not have jerk control then the decels or accels can produce poor tool paths at changes of direction or tight rads. Plus the global accelerations produce forces that move the structure which moves the tool. All very complex. Top end machines are exceptionally stiff. Cheap benchtop machines are in the 1N/um range, commercial VMC's are in the 150N/um and really stiff machines are in the +700N/um range at the tool. Rigidity and stiffness are my mantras and holy grail.

    All of these actions start with machine stiffness. If the machine or tool does not deflect (impossible it will deflect to some degree) then it does not recover and then elastically vibrate. Peter

    re cutting square corners. Controllers have settings for their paths. If your path (tolerance) is set wide then your controller will cut the corner. If its set narrow it will slow down more and keep within the path tolerance. If you use "exact" settings then it will be exact but slow. Looking at my controller (UCCNC) I also have corner tolerance settings. So depending on what you need you can look at your machine controller settings to improve your parts ie improve your toolpath.

    Some controllers plan forward and create the toolpath then run the toolpath and improve that in an adaptive method. Its rare to get anything right on the first pass!!

    Last edited by peteeng; 06-08-2020 at 05:36 PM.


  11. #11
    Member
    Join Date
    Jul 2011
    Location
    UK
    Posts
    73
    Downloads
    0
    Uploads
    0

    Default Re: Fast Cartesian 3D Printer

    Well that is interesting.. I use Mach 3 on my mill with a SmoothStepper motion controller. It hadn't really occurred to me that Jerk settings are missing! I wonder if this can be handled after the fact by the Smooth Stepper?

    I might have to get myself UCCNC.

    You also make me realise a mistake in my previous thinking. A linear moving mass with a constant acceleration / deceleration will produce an impulse response with a broad range of frequency components (thinking Fourier transform), but this is due to Jerk; the rate of change in acceleration (the first time derivative of acceleration) not the acceleration itself which may be constant. So the upper excitation frequency occurring from the linear axis' will be bounded by Jerk, not acceleration settings. Since I now realise Jerk is not part of Mach 3, it's no wonder my gantry makes the frame wobble a bit when cutting sharp corners.

    Tool path tolerance I have no experience with. I always assumed the tool path is what it is, and the physical properties of the machine just define how well it tracks. I do know my CAM program 'rolls' the cutter around sharp corners so that will help the machine physically keep up with the tool path.

    Regarding the stepper motors, the rotor aligns itself with the magnetic field created by 2 or more coils. At the poles the field on each coil will be 90 degrees out of phase, but the field can be anywhere between that, which is how micro-stepping works. However the torque will be highest when directly aligned with the poles so I expect some stepping action is always going to be present under load. If the phase between the coils is changed smoothly, I believe the rotor will move smoothly as well. However not all drivers make much effort to have a smooth continuous waveform. In the video I linked you can see one waveform with sudden discontinuities around the zero-crossing point. This is classic lack of bias on a pair of push-pull output transistors.



  12. #12
    Member peteeng's Avatar
    Join Date
    Jul 2018
    Location
    dum dum
    Posts
    6248
    Downloads
    0
    Uploads
    0

    Default Re: Fast Cartesian 3D Printer

    Hi Sash - UCCNC does not have jerk control. In the CAM world "S-curve" control is used a bit to describe jerk control. Buildbotics has jerk control. Anyone out there know of CAMs with jerk control? Smoothstepper probably does not control jerk either. Have a look at your motion tolerance settings loosening them may create smoother motion, there maybe corner tolerances as well. I'm not familiar with M3. Its difficult to "control" jerk by std path planning. I expect that to control jerk the path has to be calculated forward then reviewed then changed at spots of high jerk. The path planner like driving a car around a race track won't get it right the first lap. To create jerk you have to have a path then see what jerk you get then correct it, So an adaptive toolpath approach will be needed. Cheers Peter



  13. #13
    Member
    Join Date
    Jul 2011
    Location
    UK
    Posts
    73
    Downloads
    0
    Uploads
    0

    Default Re: Fast Cartesian 3D Printer

    Thinking about it I would expect Jerk to be handled by the motion controller rather than the G-Code software. Unless you are using software based motion control like a parallel port output.

    In most 3D printers (Marlin firmware) there is Jerk control. This is handled by the hardware motion controller that interprets the G-Code and generates the step pulses for the motor drivers. These are usually not very powerful controllers, using something like an ARM C4 processor, certainly not as powerful as the FPGA based SmoothStepper. So in my router system I'd expect Jerk to be part of the hardware motion controller setup too, if it's possible.

    Perhaps by tool path you do mean the kinematics derived from the G-Code and turned in to motor step pulses? When I read tool path I'm thinking CAM and generating G-Code.

    Maybe it's time to run our DIY routers and mills from 3D printer controller boards!



  14. #14
    Member peteeng's Avatar
    Join Date
    Jul 2018
    Location
    dum dum
    Posts
    6248
    Downloads
    0
    Uploads
    0

    Default Re: Fast Cartesian 3D Printer

    Hi Sash - Yes accel and jerk are controlled by tyhe motion controller. The CAM produces the string of co-ords called Gcode, which is the toolpath. The Gcode is not necessarily the actual toolpath. The actual toolpath is decided by the motion controller and is dtermined by the velocity, tolerance and accel limits set in the config. Its a "road" that the tool can travel down so it will cut corners, go a bit left or a bit right depending on curvature, velocity and accel.

    So in the Marlin system you can enter a jerk limit? Peter



  15. #15
    Member peteeng's Avatar
    Join Date
    Jul 2018
    Location
    dum dum
    Posts
    6248
    Downloads
    0
    Uploads
    0

    Default Re: Fast Cartesian 3D Printer

    Hi Sash - 3D Printers are entirely new to me. But I want to make one of my desktop machines into a printer and have been reading up on this and its new universe for me. Could you write down a short list of things I need to do this? Read the marlin stuff and it looks good. My future machines will have lifting gantries so will be much better for printers and mills. Peter

    Attached Thumbnails Attached Thumbnails Fast Cartesian 3D Printer-brevis-2-jpg  


  16. #16
    Member peteeng's Avatar
    Join Date
    Jul 2018
    Location
    dum dum
    Posts
    6248
    Downloads
    0
    Uploads
    0

    Default Re: Fast Cartesian 3D Printer

    Hi Sash - Did a bit of reading on "jerk" in 3D printers and it seems its not jerk as engineers would use. Its a starting velocity which prevents the machine from a "jerky" start trying to take off too fast? Cheers Peter



  17. #17
    Member
    Join Date
    Jul 2011
    Location
    UK
    Posts
    73
    Downloads
    0
    Uploads
    0

    Default Re: Fast Cartesian 3D Printer

    First data is in!

    It's already not what I expected.

    I've tested two 40x40 hollow steel tubes, 2mm wall thickness and 3mm wall thickness. Each 90cm long (just didn't have scraps 1 meter).

    I immediately encountered a problem, becasue the magnetic motor of the 'exciter' attracted itself to the steel tube and bottomed out the suspension (excursion). I used a small 20mm ali block to couple the exciter to the steel tube and offer some physical separation.

    I looked at all types of ways to analyse the data but much of it didn't clearly show differences. The most clear way to see the data is actually just a frequency response graph.

    Thin wall tube, hollow (no filling):
    Fast Cartesian 3D Printer-thin-tube-hollow-fr-png

    Thick(er) wall tube, hollow (no filling):
    Fast Cartesian 3D Printer-thick-tube-hollow-fr-png

    We can see the lowest frequency resonance in each tube is actually pretty much the same frequency 138Hz (thin) vs. 128Hz (thicker). However we can see a significant difference in amplitude. The thin walled tube has higher amplitude resonance and the width 'Q' of the resonance is also higher. The thicker walled tube resonance has -5dB lower amplitude. It also has a broader width which indicates there is higher damping factor, but I'm not sure this is actually the case.

    If we look at the decay of the resonance using a waterfall we can see that the thin wall tube decays (slightly) faster.

    Thin wall tube, hollow (no filling):
    Fast Cartesian 3D Printer-thin-tube-hollow-csd-png

    Thick(er) wall tube, hollow (no filling):
    Fast Cartesian 3D Printer-thick-tube-hollow-csd-png

    I think this points simply to the thick wall tube having higher mass, not higher damping (not really surprising, since both are rolled steel). The initial amplitude of the thicker walled tube is lower because it has higher mass that moves less for the same energy input. On the other hand we see that once in motion, the higher mass takes longer to decay. Basically we see the same energy input to the beam, but expressed differently through the time axis.

    What I do find also interesting is the CSD plot below the main resonance. We see some reasonably linear energy decay from 20Hz-200Hz in the thin wall tube, but much less so in the thicker walled tube. I guess this is simply the effect of greater stiffness in the thicker tube. We do not see resonances, but there is reasonably linear deflection. I think this is what we would see expressed in a static load simulation.

    So what do we make of it?

    I think it depends on the machine, but it's probably preferable that IF we can't avoid hitting a resonance, that the initial amplitude is lower and the energy is distributed more gradually over time (lots of smaller chatter marks, than one or two big chatter marks on a cut?).

    This will get more interesting when I fill with concrete and see how much impact the thickness has then.

    (P.S. the dB scale is not absolute, that just depends on recorder gain etc.. I will maintain the same gain levels throughout tests.)

    Last edited by ssashton; 06-11-2020 at 10:54 AM.


  18. #18
    Member
    Join Date
    Jul 2011
    Location
    UK
    Posts
    73
    Downloads
    0
    Uploads
    0

    Default Re: Fast Cartesian 3D Printer

    Quote Originally Posted by peteeng View Post
    Hi Sash - Did a bit of reading on "jerk" in 3D printers and it seems its not jerk as engineers would use. Its a starting velocity which prevents the machine from a "jerky" start trying to take off too fast? Cheers Peter
    Yes I think you're right about that. I guess it's to do with the theory of trying to instantaneously accelerate from 0mm/s. You can't have discontinuities in velocity, no matter how small. So if we instantaneously apply acceleration from 0mm/s we will have a discontinuity in velocity from nothing, to something. If we assume an initial velocity of something, we can solve the equation for how fast the motor needs to turn by the next step pulse.

    I think that is what it's about, anyway.

    A lower jerk value means the acceleration curve will come in to effect sooner during the change of velocity.

    Kind of confusing!



  19. #19
    Member
    Join Date
    May 2011
    Location
    Canada
    Posts
    790
    Downloads
    0
    Uploads
    0

    Default Re: Fast Cartesian 3D Printer

    Quote Originally Posted by ssashton View Post
    Thanks for the gear calculation. I think I will use a drive ratio of 1:1 after thinking it through with your calc.
    With the design that you have (looks nice BTW), if you make the stepper position adjustable (slotted holes, and / or additional tapped holes?), is there anything preventing you from buying a few different pulleys of different diameters and experimenting?

    In general, I like your design. One thing that I like from my research into the Duet boards is the ability to do independent homing on different axis driven by multiple motors and how easy it is to set this up in the firmware. For example, your axis that is driven by two motors should really have each motor homed individually to it's own limit switch. If not, your gantry will get out of skew. I'm guessing you have something similar planned with the electronics you're using?

    Also, I am wondering about sensorless homing using stallguard. I don't really know much about it. One concern I have is that when you break away from the low torque Nema 17's, if there is any chance repeated use might actually cause any physical damage to any of your machine components using this technique. You'd be right at zero rpm where you have the most available torque from the stepper.

    Of interest, if you look at the laser 2.x design, it uses a single stepper motor with dual shafts, eliminating the need for independent axis homing. But that is a belt driven design.

    I like that you've gone with using the proper free and fixed bearing supports for your ballscrews. I wonder if you should add a second Z axis motor, or change that up in some way? It seems to me like you've got a Cadillac top end, and a Yugo Z-axis. Yeah, they are both cars and will get you from A to B, but....

    Quote Originally Posted by ssashton View Post
    FYI, my motor and ballscrew are parallel to each other and connected by a short belt. Hence I asked what the difference between HTD and GT2 belts is?
    From what I've read on the reprap forums, GT2 is a further revision of HTD, designed to give less backlash, although people are saying there's not much difference. I don't have any personal experience to offer here.

    I assume you are going with 10mm wide belts? There are many closed timing belt options on EBay for this.

    Quote Originally Posted by ssashton View Post
    I absolutely agree that light weight at all costs is not the answer for fast printing. I took inspiration from PCB pick and place machines which only need to move tiny electronic components around. Despite the tiny parts smaller than your finger nail, good ones have heavy steel or cast iron frames and gantry, large pitch ball-screws and servo motors.
    I'm going to use surplus linear stages for my printer, the exact kind of stuff you'd find in industrial pick and place machines. No cast iron though.

    The thing I'm struggling with right now, is if I want to use the industrial Yamaha belt driven linear stage I have for one of the axis. This thing is really good quality, has four size 15 NSK bearing blocks on two rails, has gear reduction from the motor to the belt drive giving a lead of around 25mm, and the belt is more than one inch wide, heavy duty, reinforced.

    I'm considering if down the road I may wish to use this machine for more than 3d printing, and if it would be worth it therefore to go with all ballscrews. Even though the Yamaha shouldn't have any problems cutting wood.

    Quote Originally Posted by ssashton View Post

    Frame is welded 40x40mm steel with concrete and SBR (laytex) admixture core for damping.
    1mm Steel panelling (not pictured) welded on all sides for bracing except front and top.
    How good are you at welding and fabricating? On a forum like this, you really have no idea who you're talking with....you could be a master craftsman for all I know....so no offence implied in my comments. I don't know what your skill level is.

    Hollow steel tubing typically isn't flat. When you weld it, it will distort, making matters worse. The thinner stuff distorts more than the thicker stuff does. The more heat, the more distortion as well. Can you tig weld? It could end up being a fair amount of work to make your rail mounting surfaces flat and true to each other. Or it might end up close enough to what you need and not that big of a deal. There are ways to fix this if you can't get the top surface machined after welding but they are labour intensive, and can be a pain.

    T-slot aluminum is a popular go to material because it typically comes fairly flat. It you order it pre-cut, a good supplier like 80-20 will even machine the ends square for you.

    Have you considered using something like 2" x 2" solid aluminum bar? Solid aluminum extrusions typically come very flat. Do you have a mill or something to square the ends with and drill / countersink some holes? You could bolt it together.

    How would solid aluminum compare in resonance / damping? Is it possible to bolt something onto a solid piece of aluminum that would help with damping?

    I haven't seen any 3d printers that take damping seriously, even industrial ones, but I would definitely like to find out how much of an advantage it can give you. Interesting work.

    Do you know of any 3d printers that take damping seriously? It would be interesting to look at them.

    The 1mm steel plate....have you considered using something with better insulating properties? And perhaps putting some doors on the front and a top on this machine? Lexan sheet, attached with screws through tapped holes into the frame would be pretty easy to do.

    Quote Originally Posted by ssashton View Post
    Bed is 10mm or maybe 16mm cast ali tool plate, face machined for flatness.
    Looking at your design, I just assumed that the bed you have depicted was not the actual print surface. For example, you would have leveling knobs in the four corners to adjust the actual print bed.

    If this isn't the case, then you would have to have something like a BL Touch to map the bed and the bed would have to move up and down to match the X and Y movements. So your T8 lead screw on a Nema 17 would have to keep up with your X and Y ballscrews and Nema 23's. I don't think this is a good idea for the kind of performance you want to achieve, if that is what you are planning to do.



    I found the above video on youtube (I am not affiliated in any way with it's creator)

    Comments from the video

    "I was debugging this seemingly bad movement, and it turned out that it was caused by my z-jerk being too low, and this would cause deceleration and acceleration at every knot of my heightmap. Changing my Z-jerk to 1000 resolved the problem."

    Interestingly, his machine uses the same single stepper with dual shafts concept as the laser 2.x

    I hope you continue to post here, your knowledge about damping could be useful to many people on these forums, including myself, not just about 3d printers, but for all kinds of machines. And I'd like to hear about how this project goes for you.



  20. #20
    Member peteeng's Avatar
    Join Date
    Jul 2018
    Location
    dum dum
    Posts
    6248
    Downloads
    0
    Uploads
    0

    Default Re: Fast Cartesian 3D Printer

    Hi Sash - Here's a std for calculating damping may be useful. What is the length and wall thickness of your tubes and I'll do a modal analysis on them. length, wall thickness outside dims and corner rad? Peter

    Attached Files Attached Files


Page 1 of 2 12 LastLast

Tags for this Thread

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •  


About CNCzone.com

    We are the largest and most active discussion forum for manufacturing industry. The site is 100% free to join and use, so join today!

Follow us on


Our Brands

Fast Cartesian 3D Printer

Fast Cartesian 3D Printer