High spots between towers.

[Gdot]

Swapped belts from one tower to the next on all 3 and even fed the belts in from the opposite direction to ride on different teeth to eliminate some belt issue. No luck, offsets were still as follows... .01,.01,0,-.01,.01,.02,.01,-.02,.16,.12,.09,.04,.03,.02,.01. Still a big jump in z offset on the first probe on the way back to the center of the bed.

[Toddwagner]

I'm not entirely sure if this is a logical solution to our problem but bear with me here:
Step 1: do the calibration with the steps per minute until you get a proper first layer print.
Step 2: print the 50mm box with the stock settings as far as extrusion and layer height are concerned.
Step 3: measure the box and adjust extrusion width (decrease or increase to make the part smaller or larger respectively) to get as close as possible to the 50mm.
Step 4: adjust the DR to make up the rest of the difference (should be a rather small change to rod length because hopefully the change in extrusion width got you moderately close to your goal).
Step 5: if these modifications change your first layer print as you showed in your photos, maybe change the first layer extrusion width to a little higher to hopefully squish those sections together more?

Do you think we should try that out?

[Gdot]

ToddWagner, my failure after adjusting Steps/mm to achieve a low deviation and eliminate my outer perimeter sine wave (high between the towers) was that I did not print a full bed test piece. I just took it for granted that my first layer problems were gone when I achieved a low deviation with the 6 factor calibration wizard.. I pretty much did the remaining steps that you outlined in your post. These test prints are very small of course, so I did not know that I still had an issue with larger prints out at the perimeter. I dont think the previous steps caused my large first layer issue, because after setting everything to default and running the calibration with standard DR length and steps/mm, I discovered what might be the true issue I am having. The fact that when I use z probe gcode to run from the center of the bed to the x tower, the offsets are very close to each other at each step. However, when it begins to travel back to bed center, testing all the same previous locations, the offset spikes from like 0 to .15 to .20 plus. I believe this is what is causing the gaps in the prints. NO matter how calibrated I have it, the nozzle is lifting up when it changes direction out by the perimeters. I am currently testing other towers and I have found that the Z tower does it as well. I will test the Y tower next. One thing though, when I only travel 50% of the way to the X tower and then return when z probing for offsets, there is no spike in the offsets when it travels back to bed center.

[Toddwagner]

Could you post that gcode you're running so I can try it on my rostock as well? Also just out of curiosity, did you build your machine yourself or did you purchase a kit?

[Gdot]

Yes!, great idea I am very curious what your output is. I built my machine myself.

Code: Select all

G28
G1 Z10
G1 X0 Y0 ;
G30
G4 S5
G1 X-15.53 Y-8.75 ;
G30
G4 S5
G1 X-31.06 Y-17.5 ;
G30
G4 S5
G1 X-46.59 Y-26.25 ;
G30
G4 S5
G1 X-62.12 Y-35 ;
G30
G4 S5
G1 X-77.65 Y-43.75 ;
G30
G4 S5
G1 X-93.18 Y-52.5 ;
G30
G4 S5
G1 X-121.24 Y-70 ;
G30
G4 S5
G1 X-93.18 Y-52.5 ;
G30
G4 S5
G1 X-77.65 Y-43.75 ;
G30
G4 S5
G1 X-62.12 Y-35 ;
G30
G4 S5
G1 X-46.59 Y-26.25 ;
G30
G4 S5
G1 X-31.06 Y-17.5 ;
G30
G4 S5
G1 X-15.53 Y-8.75 ;
G30
G4 S5
G1 X0 Y0 ;
G30
G4 S5
G28

[Toddwagner]

Thanks for the code! I'll try to run it in a bit when I finish my current print, although I do have to figure out how to send all that code information to my printer

As far as building it yourself, I think everyone who has had this problem has built their printer, and no one has purchased a kit and had it. So that's something to consider, maybe it's an error in the build instructions that a few people face...

[Gdot]

OK, just make a file with the extension .gcode. Then just paste the code into it. Load that up in your software, right now for me that's MatterControl, then run the program and watch what offsets it prints out to the terminal.

[Toddwagner]

OK, just make a file with the extension .gcode. Then just paste the code into it. Load that up in your software, right now for me that's MatterControl, then run the program and watch what offsets it prints out to the terminal.

Perfect. I'll keep you posted

[Toddwagner]

Ok so I ran the code. Going towards the X-tower starting at center: -.01, -.01, .00, .02, .04, .04, .03, -.01. Then going away from the X towards center: .12, .11, .09, .06, .03, .01, .01.

[Gdot]

Interesting, but I think I found the solution to my problem, at least for the x tower issue. after trying everything, I noticed the hotend whip was leaning against the delta rods for the x tower. I adjusted them and re-routed the bowden tube away from inside the sleeve. That helped tremendously, my z offsets are now 0.01,0.01,-0.02,-0.02,-0.01,-0.01,-0.01,-0.01,0.02,0.01,0.01,-0.01,-0.01,0 going to and from the x tower. I also noticed that the delta rods had tight spots in the ball joints, so I lightly sanded both ends of all 6 till they had a smooth motion. So, even after all this I still have the high spots between the towers. Z to Y = .40 z offset, Y to X = .20 and X to Z = .20. Maybe there is still some tight restrictive motion at those extreme locations, but when I rotated the table one notch, the Z to Y offset dropped to .35. So that maybe remedied when my new bed insulator arrives.

[Toddwagner]

Excellent! I have one tip for you that worked for me. When you're calibrating your printer and writing down the values, you're plugging those into the deviation calculator. What I did instead, even though it does take considerably more time, is bump up my steps/mm and run the probing wizard, take note of the offsets and then plug them into the delta cal each time, then copying the EEPROM info to my machine. That way I'm actually moving up in increments of .1 steps/mm, whereas with your process all of your EEPROM settings (z-max height, horizontal radius etc.) are not being calibrated as you raise the steps/mm. This really helped me and now I have a virtually flat first layer out to about 80 percent of the build radius.

[Gdot]

OK, too scared to call this solved just yet.... I had my printer almost disassembled looking for any sign or clue of an issue that could be causing poor calibration of the outer perimeter. Nothing I looked at or adjusted made the problem go away. So I gave up and decided to put everything back together, tighten every screw and faster on the thing. I decided at that moment I was going to calibrate this machine to the smallest diameter settings that would give me at least a ,02 deviation. SO I started at 145 mm diameter by running G29 3 times followed by the 6 factor delta wizard and decremented by 10 mm each iteration. Each iteration I would load the proper x/ Y coordinates into the firmware and would run G29 with those probe points. After several iterations I finally arrived at 85 mm diameter giving me the .02 deviation I was looking for. After that I decided to run the default Rostockwizardprobe program to test the outer perimeter. As usual, it gave me very crappy looking Z offset errors all around the perimeter. So, I decided to run my full bed circular test print anyway....WOW!!!...the print actually came out almost perfect......!!!!!!! I have ran multiple test prints and they all look damn near perfect. I guess the automatic calibration does not work very well on the outer perimeter. Something makes it more accurate by probing further in instead of making the arms almost extend horizontal when probing. I am in disbelief right now....best July 4'th ever!!! Running the 50mm test cube right now to test dimensions. Also, just for clarification, I am using the default DR and steps/mm values ( 291.06 and 80 steps/mm).

[Toddwagner]

Each iteration I would load the proper x/ Y coordinates into the firmware and would run G29 with those probe points.

First off, congrats on the success!
Secondly, I followed most of what you were saying, but what X/Y coordinates are you loading, I don't recall ever having to edit X/Y coordinates in my firmware. I also am slightly confused about the 6 factor calibration you're running, is that the Escher 3D program? Do the X/Y coordinates I mentioned earlier have anything to do with running this program? I apologize for the silly questions, I'm a bit slow if you haven't noticed .

EDIT: How did you reset your EEPROM to stock settings?

[Gdot]

Yes, I'm speaking of the escher 3D program. As far as the X/Y coordinates, in the firmware are the programmed points for the probing locations when using the G29 command. You can change the points to custom locations for better probing. I forget the exact label for them but its something like X1 Y1, X2 Y2 and X3 Y3. What I did while in the Escher 3D program was input my max printable radius and click suggest probe points. Then I took the probe points for the X,Y and Z tower and plugged those into the proper firmware locations to make G29 probe those points. What I found was that by decreasing the diameter locations of the probe points (bringing closer to the center of the bed) my calibration became much more accurate. Once I got to the locations that were from plugging in a max printable diameter in escher3D, I was down to a 85mm diameter. This gave me a .02 deviation after plugging all 10 probe points into escher3D. I just modified the x/y coordinates in the original rostockwizardprobe program to probe these 10 points which is what escher3D suggested as probe points after plugging in 85mm printable diameter.

[Toddwagner]

Ughhhh!!! So I took your advice and did the same calibration technique, and it worked great as far as flat prints are concerned. Now that I'm printing flat I'm having problems with the individual strings not fusing together on the plate, giving me ~.3mm gaps between the lines... Time to try to fix this problem next

[Gdot]

Sounds like maybe you need to tweak your z max height. I do it by manually adjusting that setting in the eeprom as opposed to using the lcd screen to adjust it. It's the same difference from what I can see. That is if the .3mm gap is consistent across the entire print. Increasing z max height will lower the nozzle giving you more squish.

[Toddwagner]

How did the dimensioning go for you by the way?

[Gdot]

I came out at around 49.9 +/-. I plan on re-calibrating with a shorter DR length to get it even closer. The firmware comes default with a 291.06mm setting but I read a post where geneB states the length is 290.8. Calculating with 626Pilot math, 49.9/50 = .998 x 291.06 = 290.477. I will try that once my new bed insulator gets here since I will be throwing off my calibration with what I'm hoping is a flatter bed. Hopefully I can get deviation down to .01!!!! Before that though, I'm running prints while I dial in my retraction settings.

[geneb]

If the firmware default is 291.06, I'd run with that. It's entirely possible that a chance was made and my 290.8 figure is no longer accurate.

g.

[Toddwagner]

Argh!!! I finally got my printer to print an excellent first layer with PLA, so I switched over to ABS to try some prints with that on the whole bed... It didn't turn out well... I cleaned and reapplied glue to the plate and I'm heating up to try it one more time. I'm not sure if it is actually adhesion issues or if there is something bigger at play here, but this is the worst failed first layer yet

[geneb]

ABS is kind of like playing the 3D printer game on "hard mode". If you can successfully print that big in PLA, your calibration issue with the printer is solved.

g.

[Gdot]

A quick update:
So I have been getting pretty good prints since I modified the way I calibrated the machine. I calibrate with the machine at printing temperatures and the probe points moved in from the outer perimeters. With the calibration wizard I get a .02 deviation. I put on my new bed insulator and re-calibrated, but this made no difference. As far as my first layer, I still get a small area just off of the x tower where the lines dont fuse. Also, there was a sightly bigger area straight across from it (between Z and Y towers) where the lines don't fuse. I had a hunch though, the slicer was setup to do 45deg starting angles for infill, I changed this to 0 and the problem areas were gone, except a new small area of unfused lines now on the other side of the x tower(between x ans z towers) where there was none before. I think my problem from the get go was with the arms, bar bells, and the end effector platform. So, since this machine is worth upgrading, I ordered all new tricklaser parts for it, including the carbon fiber arms and aluminum platform. That should do the trick!!!

[Toddwagner]

How big of an issue is it really if your first layer doesn't completely fuse?

[Gdot]

For me, big issue, because the bottom layer is the top to some of my prints, like a phone case for instance.

[Toddwagner]

Ok so I just figured I would add an update on my quest for a level first layer for anyone who is reading this looking for a solution. I did get a level print on my entire bed, but the way I did it was incorrect in itself. I adjusted my steps/mm and calibrated using the delta wizard every .1 step change (extremely time consuming ) Ended up at 81.85 steps per mm when I finally got the full bed print I desired. I thought all was well at this point so I printed a couple of 50mm test cubes and measured the z-axis height to be 51.17mm, which is 2.3% taller than it should've been. So I did the math and lo and behold, the increase in steps from 80/mm to 81.85/mm is also 2.3%! So even though I did end up getting the level bed I wanted, I'm printing parts a huge amount taller than desired.
My next step in this journey happens this weekend when I get the TrickLaser CF arms and aluminum effector plate in. I'll keep y'all updated on my end of this journey!!