1. Got a question or need help troubleshooting? Post to the troubleshooting forum or Search the forums!

Z Axis - precision and accuracy

Discussion in 'Mods and Upgrades' started by Ziggy, Mar 25, 2014.

  1. Mike Kelly

    Mike Kelly Volunteer

    Joined:
    Mar 11, 2013
    Messages:
    6,967
    Likes Received:
    2,275
    Ace is close by, there is a bike shop but how much would they charge for them?

    I don't get waving, just curious if it would help on other quality issues
     
  2. Ziggy

    Ziggy Moderator
    Staff Member

    Joined:
    Feb 20, 2013
    Messages:
    707
    Likes Received:
    530
    They would have to be pretty mean to charge for two 3mm ball bearings. But you might have to buy a bag for a few $'s
     
  3. 1d1

    1d1 Active Member

    Joined:
    Feb 21, 2013
    Messages:
    418
    Likes Received:
    216
    Are you putting a bearing between rod and motor?
     
  4. Wilco

    Wilco New Member

    Joined:
    Oct 29, 2013
    Messages:
    7
    Likes Received:
    0

    That's what it's starting to sound like, isn't it?

    C'mon, Ziggy... make with the details. :)
     
  5. 1d1

    1d1 Active Member

    Joined:
    Feb 21, 2013
    Messages:
    418
    Likes Received:
    216
    And just to muddy the z-waters further... not really, but I'm curious as I noticed this the other day: These are two objects printed simultaneously. The right side finished and the left side continued on upward. It is at that exact point where the left side began printing smoother when there were no longer any big movements by the print head. No settings, temps, or speeds were changed. My question: Could something like this, and maybe some portion of z ribbing, be due to the movement of the x belt and some stretching that occurs in the transition? The ribbing here looks worse than it is - the direction of lighting makes a big difference in defining details. 20140606_171525.jpg
     
  6. 1d1

    1d1 Active Member

    Joined:
    Feb 21, 2013
    Messages:
    418
    Likes Received:
    216
    Oh - I meant X and Y belts..
     
  7. nickster

    nickster Member

    Joined:
    Oct 7, 2013
    Messages:
    82
    Likes Received:
    52
    I never fully appreciated how stupid precise the Z has to be to get consistent layer surface finish.

    I am still running with this set up: http://forums.robo3dprinter.com/index.php?threads/precision-z-coupler-upgrade.1471/#post-11304

    Springy Z couplers are in tension and hold the Z screw against the ball bearing and stepper shaft to form a flex uni-ball joint. So there is a ball joint at the stepper/screw interface at at the screw/carriage interface.

    Since the steel ball between the screw and stepper determine the screw location, the coupler could actually be printed out of thin wall ABS, PLA treated with carb cleaner or Nylon. Offered to design something but no one was interested.

    Results are excellent. Surface is boringly smooth. Running 240C ABS; parts are as strong as injection molded. Yes all your belts better be tight. There should be no wobble in your linear bearings. I bought a few dozen cheapy linear bearings on eBay and hand matched them to my rails. X carriage has NO jiggle. Also need high quality filament who's diameter does not vary. Robo filament is not bad, but not the best. I started with the Z ribbing muddy water blues, and it is not until you fix the last variable that you get fine prints.
     
  8. nickster

    nickster Member

    Joined:
    Oct 7, 2013
    Messages:
    82
    Likes Received:
    52
    Almost forgot. Check that your hobbed bolt teeth are uniform. My original one has serious issues as well as some replacements I got from eBay. The extruder springs that hold the filament against the hobbed in the stock robo are way too wimpy. I went over to ACE and got some that were 24mm long 0.67mm wire dia, od spring dia 6.25mm and about 10 turns. Had an overhead spool jam because the filament was looped under; the entire x carriage lifted upwards on the Z where it hit the overhead cover before the filament finally stripped. There should be no filament shards in the teeth of your hobbed bolt after running for 8 hours.
     
  9. 1d1

    1d1 Active Member

    Joined:
    Feb 21, 2013
    Messages:
    418
    Likes Received:
    216
    Thank you, Nickster! I would be happy to print out your design in nylon if you send it to me. I have no access to lathe for precision drilling but might give it a shot anyway. Boringly smooth sounds oddly fascinating...
     
  10. nickster

    nickster Member

    Joined:
    Oct 7, 2013
    Messages:
    82
    Likes Received:
    52
    I'll start thinking about the coupler sans lathe.

    Two other Z issues-
    1) Default Marlin is to turn on the Z stepper drivers as a z move command enters the Arduino buffer, so that 8 or so commands later depending on the buffer depth, the Z stepper is all powered up and happy and ready to move. I was having problems with my Z not slewing fast enough and it would drop steps, bad. Might have been a function of ambient temp or lack of airflow in the base by the ramps. (BTW be sure to add a fan for your RAMPS. Mine is just held on with double sticky tape to the case) (My machine has so many fans now it sound like a hovercraft). Anyway I instrumented the Z current and increased the current by about 30% which fixed the slew problem. I then noticed that I would get random Z stepper jumps of many (not micro) steps right as the stepper driver gets enabled. Could be ground bounce. Never found root cause but I ended up rebuilding my Marlin to leave the Z steppers on all the time. You are not going to burn out your Z steppers; they still see less current than the X or Y. The stepper driver is thermal protected, but the stock heatsinks are microscopic. (I machined up a larger custom HS for mine). Will every one this problem? duh no, but this could be a lurking analog signal integrity issue. Also could be a "weak" stepper driver.... So -

    Recommend you get a cheapy dial indicator on your bed and measure the precision and repeatability in Z of your X carriage height over the bed. Also need to add some flags on each Z screw. Home Z, set the flags to a known position, do a print, and check to see if the flags go to the same position with Z=0 after the print.

    2) The Z nuts are kind of wiggly in plastic extrusion. The have to be so they can eject if the z stop fails. If you look at the nut, there is a bevel on the outer edges. Problem is the nut's mating surface on the interior of the plastic X/Z extrusion is not precision flat. In fact, of the Z holes on the original X Idler extrusion are over size so the bevel of the nut sits on edge of a hole. This could be good because the nut might tend to center itself if there is not too much opposing lateral force from the Z screw. But. Any lateral displacement of the Z (wobble if the Z screw) results in a vertical displacement. There are a number of thingiverse projects that attempt to deal with this issue by letting the nut float against a flat ptfe washer. Just can't let the Z nut dance around uncontrolled.
     
    2 people like this.
  11. Ziggy

    Ziggy Moderator
    Staff Member

    Joined:
    Feb 20, 2013
    Messages:
    707
    Likes Received:
    530
    I have used a similar but simpler Z Ball Coupler arrangement as part of the overall solution to Z ribbing. It was an obvious thing to do to overcome the limitations of the helical coupler in this kind of Z axis design. As far as I know the technique has been around for a long time.

    Unfortunately I think when your mod was originally posted the causes and significance of tiny variations (literally 100ths of a millimeter) in coupler length were not generally well understood.
     
  12. Ziggy

    Ziggy Moderator
    Staff Member

    Joined:
    Feb 20, 2013
    Messages:
    707
    Likes Received:
    530
  13. SteveC

    SteveC Well-Known Member

    Joined:
    Jun 12, 2013
    Messages:
    894
    Likes Received:
    316
    I totally missed that thread, nice design and discussion. I just added a simple 3mm ball bearing and and kept Z steppers active as suggested by Ziggy and am evaluating the results.

    Do you know if that Slic3r bug ever got addressed in the new versions? I think I have seen that in the past.

    Nice, I think that's the first mention of ground bounce on the forum.
     
  14. nickster

    nickster Member

    Joined:
    Oct 7, 2013
    Messages:
    82
    Likes Received:
    52
    I had seen the disable_z quite a while ago and should have posted sooner. Sorry.

    Don't particularly like my Z coupler set up. Would like a generic solution without machining. Current issues are that as the aluminum flex Z coupler is put in tension, it bends to one side, which throws the free position of the Z screw off axis. I had to add the three adjusting screws in the outer coupler ring to recenter the screw. In practice, I install the Z steppers with couplers and adjust the Z screws so they are less than 5mm out of center. Then drop the X carriage on. Reason why centering is important is that even though the acorn nut that supports the X carriage centers in a drilled out bolt under the carriage, the capture angle is not steep enough so you can force the carriage to ride up on the round acorn nut if there is enough horizontal displacement force from the screw. More carriage weight would help as well as making the contact ring for the acorn nut larger. Tweak.

    First layer ripples? On first layer if flow rate it too high for the actual Z space between head and glass, you get an exceptionally wide extrusion width. Edges of the overflow don't attach well. When the hot end comes along on the next pass it acts as a snow plow and creates a wave ahead of it, lifting up any high spots from adjoining layer. This blurp causes the current overflow edge to not attach as well. Net result is attach distortions propagate at about 90 degrees to the direction of the hot end path. I was able to manually decrease the flow rate in Repetier to make the patterns go away.

    Here's some typ quality I am getting. Printed 0.2mm LH ABS 240C directly on glass w/Emers Extra Strength. Little pock marks are from a bad roll of Robo ABS. Little explosions as it comes out of the hot end. Could be water, but some of the other rolls see fine. Also printed external perimeter first for better precision, but doesn't yield best surface quality. BTW latest experimental version of Slic3r has an option which shrinks the perimeter; allows size compensation when printing inner perimeter first. Measured Z error to be 0.005mm or better over 5mm of screw motion. Debunks need for metric screws. Save my ToyBuilder filament for when I need best prints. More expensive, but tolerances are really tight.

    [​IMG]
    Also be sure to tighten all stepper, idler and base plate screws on a monthly basis. My Robo 3D black base plate *had* serious cold flow issues where things would loosen up. Finally gave up and did a CNC aluminum base plate:
    [​IMG]
     

    Attached Files:

    2 people like this.
  15. Bob64

    Bob64 Member

    Joined:
    Mar 18, 2013
    Messages:
    104
    Likes Received:
    23
    Wow. I really like your smooth rod holders.
     
  16. SteveC

    SteveC Well-Known Member

    Joined:
    Jun 12, 2013
    Messages:
    894
    Likes Received:
    316
    I was looking at your comment here in that thread:
    "I stuck some old test cubes in the first pic as a spacer to make the acorn nut visible. Very interesting -if you look closely, you see bumps on the edges of the test cubes. Turns out this is a Slic3r artifact caused by movement from inner to outer perimeters. Too much filament gets deposited on these transitions because flow does not turn off instantaneously on the moves. In this case, the pattern repeats every 4 0.3mm layers where the start point precesses one face one each layer. I have modified gcode where the start point is the same, and the faces are absolutely smooth. Vases are also smooth."​

    I don't think I see this in Slic3r 1.1.4, I'm pretty sure I saw it an old version.
     
  17. nickster

    nickster Member

    Joined:
    Oct 7, 2013
    Messages:
    82
    Likes Received:
    52
    Pic below is 10mm test cube Slic3r 1.1.4 with new feature aligned start point enabled. You can see the starting groove on the right. Previously the without aligned feature, starting groove advanced one corner from layer to layer creating the corner artifact.

    0.2mm LH 240C RoboABS Elmers, 3 vert shell, ext perimeter first, steam explosions on surface.[​IMG]
    @Bob64 tkx. Went thru a bunch of rod holder alternatives. Had some really cool looking ones that were all 3D contoured and shaped. In the end, went back to basics. Rods fit into a triangle shaped hole and rest against two of the sides 1/4-28 set screws form the third opposing contact point. Didn't want to deal with any traming issues so I put the rods with attached blocks in a lathe collet and did a facing cut. Would have rathered 3D printed blocks, but printer was out of commission.
     

    Attached Files:

  18. Bob64

    Bob64 Member

    Joined:
    Mar 18, 2013
    Messages:
    104
    Likes Received:
    23
    Woah. I also noticed the cable chain under the bed. What other upgrades have you done? Do you have a show-n-tell thread with all the stuff you've done to it?
     
  19. Mike Kelly

    Mike Kelly Volunteer

    Joined:
    Mar 11, 2013
    Messages:
    6,967
    Likes Received:
    2,275
    Nick do you have any more pictures of your Y axis cable manager?
     
  20. nickster

    nickster Member

    Joined:
    Oct 7, 2013
    Messages:
    82
    Likes Received:
    52
    Probably should start a separate thread, but here are some pics of the cable manager. Original part dimensions were for old Robo with drawer slides. Hacked the pieces up to fit the Aluminum base plate version with linears. Design was done in Freecad and OSCAD. Could post what I have on thingiverse or where ever, but you might need to do some tweaking. Idea was to route the thermistor and heater wires to the edge of the glass and down a guide. The guide snaps onto the glass (there is a recessed clip) and is captured by one of the posts standoffs supporting the glass bed. Cable chain laid flat on the old black plastic. It is kind of hanging in the pic on the left side; height is diff for the linears. Also had to sand off the straight bottom edges of the chain links. Didn't matter on the old drawer slide design, but they were snagging as they went past the linear slide bearing blocks. Printed in ToyBuilder Orange ABS 225C Kapton w/slurry Stock Robo hotend. Chain is snap together. Chain ends force fit into attach points. Need to put the nuts back on the support post that goes into the Robo base; just hanging there.
    DSC02706.JPG DSC02704.JPG DSC02703.JPG DSC02698.JPG
     
    8 people like this.

Share This Page