@garyd9 Sure! The final result is admittedly more "cobbled" than I usually prefer as a result of the RobotDigg carriages having a 40mm distance between the flats compared to the 35mm on the printed effector of the D300VS+, but here goes any way:

• Purchase 3x 700mm MGN12H rails and carriages from Iverntech on Amazon.
  • I started with the ones from RobotDigg, but they were really terrible quality, so I wound up using the rails from Iverntec.
  • I found that carefully pulling the carriages off of the rails, cleaning with isopropyl alcohol and re-lubricating with Slick 50 Supercharged One Lube is really worth the effort because they move completely smoothly now
• Purchase a set of RobotDigg carriages MGN12H-OP-BLACK
  • Carefully chase the threaded holes with an M3 tap as deep as you can go because the anodize inside the threaded holes really gums the threads up
  • Again carefully, file off 0.5mm from the faces that each ball stud will mount to for a total distance of 39mm between the faces (this will make more sense later)
• Thread the ball studs in as deep as they will go on each carriage by hand and note that one side has a shallower hole than the other. Remove that ball stud and shorten those threads by that amount and re-install the ball studs on to the carriages so that they are fully seated.
• Get ahold of some NO snap acting microswitches with a lever arm and replace the ones that came with the printer because without the lever the switch will not activate on the top of the MGN12H carriages (the screws for the wipers are taller than the switch plunger)
  • While you are doing this, re-route the wires for the switches to the sides of the towers because you will be using the inner-most face of the towers to mount the rails.
  • Go ahead and slide the switch mounts up as high as they will go and tighten the mount back down there
• Get some M3 x 8 screws and some 2020 T-slot "hammer-head" M3 nuts and install the rails on to each tower, facing inward.
  • I did this by putting bolts in the holes of the rail at even spacings (I think I started one hole from each end and left 4 holes between bolts on my rails) and loosely starting the nuts on the threads.
  • When you put the rail against the tower and start to tighten the bolts, the nuts should center the rail on the tower nicely.
  • Place the top of the rails just beneath the lever arm of the end stop switch so that when the carriage comes to the end of the rail, it will hit the endstop switch before balls start to fall out of the carriages.
• Assemble the RobotDigg carriages to the linear rail carriages with the tensioning screws pointed down, install the belts and tension them
  • If you watch the video on their website, they show wrapping the belt around the tensioning blocks and pressing them into the carriage. Save yourself the headache, it doesn't fit like that. Just run the belt down one side and get some slightly larger washers to pinch the belt in place when you tighten the tensioning blocks in place.
• Purchase some 1m thick shim washers from McMaster (PN 98055A099) and place two between each ball stud and the effector. Now both the carriages and the effector have 39mm between the ball studs.
  • I originally planned to shim the effector to the full 40mm of the carriages, but I couldn't get the nuts for the ball studs in the effector to catch. With the 39mm spacing, the nuts just catch on the ball studs in the effector and you can tighten them to the point that they start to compress the plastic in the effector.
• After all of this work you'll frustratingly notice that the springs between the carbon rods don't fit over the carriages anymore. For a moment I thought that was game over myself, but then I just decided to try getting some zip ties and strapping the springs slightly down the rods, away from the carriage so that there would be clearance. To my surprise, this actually works very well.
  • The zip ties do walk around on the rods a little, so I may add a little epoxy to hold them in place at some point. For now, however, periodically moving the zip ties back to where they should be seems to be working ok.

Now, after ALL of that work, you should have a rock solid, smooth moving delta carriage system that will eliminate what appears to have been the source of all of my problems. Run the auto delta calibraion a few times, upload the values to your config file and smile at how nice your prints come out all of the sudden!

Happy printing!

@phaedrux YES! That looks identical to the pattern I'm seeing, glad I'm not the only one.

Out of curiosity, can you tell me the length of one of the straight walls and count the number of bands you see where they interesect the top edge? I'll throw your numbers in to my calculator and see if we are getting the same volume of extrusion/band.

If it turns out that the volume/band is the same on your CoreXY as on my Delta, then we may really be on to something.

I ran a print last night that really shows this issue as clear as day. Normally, I would immediately say it is an under extrusion issue, but my single and double wall test prints come out at the exact right wall thickness, so I don't think that's the. See the link below to see what I'm talking about.

Extrusion Pulse Photo

Anyone else have further thoughts?

Hi all,

I have been working on calibrating my E3D Titan Aero for extruder non-linearity, and while it does work and seems to make an improvement it is very filament and temperature dependent.

So, I was going to start making a library of correction factors for different filaments at different temperatures, but immediately stopped when I realized how long that was going to take and the hassle it was going to be to remember to change those factors for every print I run.

That got me thinking, there must be a better way...

Has anyone ever thought of / tried to implement an idler wheel with a hall effect sensor (or any other rotary sensor) on the filament before it gets to the extruder that actually measures the extrusion distance, compares it to the expected value, and tunes the extrusion rate in a closed feedback loop?

Just a thought.

Happy printing!

Have you seen this adapter from 713Maker?

713Maker - Smart Effector Ballcup Adapter

Hi all,

I have been working on calibrating my E3D Titan Aero for extruder non-linearity, and while it does work and seems to make an improvement it is very filament and temperature dependent.

So, I was going to start making a library of correction factors for different filaments at different temperatures, but immediately stopped when I realized how long that was going to take and the hassle it was going to be to remember to change those factors for every print I run.

That got me thinking, there must be a better way...

Has anyone ever thought of / tried to implement an idler wheel with a hall effect sensor (or any other rotary sensor) on the filament before it gets to the extruder that actually measures the extrusion distance, compares it to the expected value, and tunes the extrusion rate in a closed feedback loop?

Just a thought.

Happy printing!

@garyd9 Sure! The final result is admittedly more "cobbled" than I usually prefer as a result of the RobotDigg carriages having a 40mm distance between the flats compared to the 35mm on the printed effector of the D300VS+, but here goes any way:

• Purchase 3x 700mm MGN12H rails and carriages from Iverntech on Amazon.
  • I started with the ones from RobotDigg, but they were really terrible quality, so I wound up using the rails from Iverntec.
  • I found that carefully pulling the carriages off of the rails, cleaning with isopropyl alcohol and re-lubricating with Slick 50 Supercharged One Lube is really worth the effort because they move completely smoothly now
• Purchase a set of RobotDigg carriages MGN12H-OP-BLACK
  • Carefully chase the threaded holes with an M3 tap as deep as you can go because the anodize inside the threaded holes really gums the threads up
  • Again carefully, file off 0.5mm from the faces that each ball stud will mount to for a total distance of 39mm between the faces (this will make more sense later)
• Thread the ball studs in as deep as they will go on each carriage by hand and note that one side has a shallower hole than the other. Remove that ball stud and shorten those threads by that amount and re-install the ball studs on to the carriages so that they are fully seated.
• Get ahold of some NO snap acting microswitches with a lever arm and replace the ones that came with the printer because without the lever the switch will not activate on the top of the MGN12H carriages (the screws for the wipers are taller than the switch plunger)
  • While you are doing this, re-route the wires for the switches to the sides of the towers because you will be using the inner-most face of the towers to mount the rails.
  • Go ahead and slide the switch mounts up as high as they will go and tighten the mount back down there
• Get some M3 x 8 screws and some 2020 T-slot "hammer-head" M3 nuts and install the rails on to each tower, facing inward.
  • I did this by putting bolts in the holes of the rail at even spacings (I think I started one hole from each end and left 4 holes between bolts on my rails) and loosely starting the nuts on the threads.
  • When you put the rail against the tower and start to tighten the bolts, the nuts should center the rail on the tower nicely.
  • Place the top of the rails just beneath the lever arm of the end stop switch so that when the carriage comes to the end of the rail, it will hit the endstop switch before balls start to fall out of the carriages.
• Assemble the RobotDigg carriages to the linear rail carriages with the tensioning screws pointed down, install the belts and tension them
  • If you watch the video on their website, they show wrapping the belt around the tensioning blocks and pressing them into the carriage. Save yourself the headache, it doesn't fit like that. Just run the belt down one side and get some slightly larger washers to pinch the belt in place when you tighten the tensioning blocks in place.
• Purchase some 1m thick shim washers from McMaster (PN 98055A099) and place two between each ball stud and the effector. Now both the carriages and the effector have 39mm between the ball studs.
  • I originally planned to shim the effector to the full 40mm of the carriages, but I couldn't get the nuts for the ball studs in the effector to catch. With the 39mm spacing, the nuts just catch on the ball studs in the effector and you can tighten them to the point that they start to compress the plastic in the effector.
• After all of this work you'll frustratingly notice that the springs between the carbon rods don't fit over the carriages anymore. For a moment I thought that was game over myself, but then I just decided to try getting some zip ties and strapping the springs slightly down the rods, away from the carriage so that there would be clearance. To my surprise, this actually works very well.
  • The zip ties do walk around on the rods a little, so I may add a little epoxy to hold them in place at some point. For now, however, periodically moving the zip ties back to where they should be seems to be working ok.

Now, after ALL of that work, you should have a rock solid, smooth moving delta carriage system that will eliminate what appears to have been the source of all of my problems. Run the auto delta calibraion a few times, upload the values to your config file and smile at how nice your prints come out all of the sudden!

Happy printing!

Hello all,

Sorry for the long silence here, but I thought I would share my findings and final satisfaction with this printer in case anyone runs into the same problem.

I wound up building a new effector to use a standard E3D V6 hot end and built a 9:1 belt reduction drive bowden extruder since I was so certain the extruder was the source of the issue, but low and behold, the exact same surface ripple persisted...

At this point, I was at the end of my rope and almost gave up but just started really trying to shake stuff around. It wasn't until then that I noticed that the carriages on the printer were flexible enough that I could pretty easily get one of the three wheels to come off of the aluminum extrusion if I pushed in the correct directions. So, I ordered some 12mm knockoff MGN12H rails and carriages and the delta carriages from RobotDigg. With a little bit of cursing and filing, everything fits up pretty great and the first test print shows that the artifact is GONE!!

So, I put the E3D Titan Aero back on and the result repeated, NO MORE SURFACE RIPPLES! After all of that, it was just an issue with the carriages being a little too flexible and creating what must have been some kind of stick/slip where the wheels are supposed to roll up the towers...

In the links below, you'll see a picture of the installed rails and the before and after test print comparison. The black print is with the new rails and the blue print is with the old carriages. The horizontal sections you see are my experimenting with different speeds and extrusion rates.

New Rails Installed

Printed Results Comparison

With that change, this is now hands down the best "consumer" grade printer I have ever been around and would gladly recommend it to anyone that can get their hands on one.

Any way, happy New Year and happy printing!

@phaedrux YES! That looks identical to the pattern I'm seeing, glad I'm not the only one.

Out of curiosity, can you tell me the length of one of the straight walls and count the number of bands you see where they interesect the top edge? I'll throw your numbers in to my calculator and see if we are getting the same volume of extrusion/band.

If it turns out that the volume/band is the same on your CoreXY as on my Delta, then we may really be on to something.

I ran a print last night that really shows this issue as clear as day. Normally, I would immediately say it is an under extrusion issue, but my single and double wall test prints come out at the exact right wall thickness, so I don't think that's the. See the link below to see what I'm talking about.

Extrusion Pulse Photo

Anyone else have further thoughts?

@dc42 The period of the banding corresponds with 0.17mm of linear filament movement into the extruder, or roughly 0.405mm^3 of volumetric extrusion. As far as I can tell, that coincides with the following movements in the extruder (roughly):

• 9 full steps of the stepper motor/pinion gear
• 8.2° degrees of rotation of the stepper motor
• 2.7° of rotation of the large drive gear
• 1/2 of the pinion gear pitch

The only thing that I can relate any of those numbers to (and this is a stretch, I'll admit) is that there appear to be only 2 teeth meshing between the pinion gear and the drive gear, so it is possible that this phenomenon is related to every time a tooth in the gear drive makes or loses contact with its mating gear. I wonder if switching to a helical cut gear set of the same size would eliminate the issue?

@Phaedrux FYI: I replaced the bearings in the extruder last night with brand new replacements from E3D with no impact to the banding. Nice to have that eliminated as a possible cause though.

@JoergS5 Thanks, I am interested to see what the community thinks as well. While that may be a great explanation for other controllers, the Duet Wifi hasn't used segment approximation for delta movements since 2015, one of the major reasons that it has taken the delta printer market by storm. There may be some funky back torquing going on for other reasons, but I don' think it is as a result of segmentation.

It also turns out that I was moving too fast when doing those calculations and put the gear ratio of the extruder in the wrong place (how embarrassing), and the volume/step doesn't line up with the ripples at all... Sorry for the red flag folks... See my updated calculations with the gear ratio in the CORRECT place below:

CORRECTED Extrustion Ripple Calculations w/ Extruder mm/Step

The extruder is still roughly 4X less precise than the movement of the print head, but not nearly as bad as I was thinking. I guess I'm back to digging for a mechanical issue and I'll come back if I find any solutions.

EDIT: I had the extruder gear ratio in the wrong place in the calculations, please see my comments in the next post

Well, I am not 100% sure, but I think this may all stem from a full step positioning error of the extruder stepper, at least in my case.

I was updating to the 2.02 firmware last night when I noticed that the StealthChop mode was now available, so I thought I would give it a try. First thing, it was really impressive how much quieter the printer was with that chop mode enabled. Second, there was a noticeable difference in the print quality changing the chop mode, which I did not expect. In the picture below, the part on the left was printed with StealthChop and the part on the right was printed with an old firmware version using constant off time chopping. The difference isn't dramatic, but it is noticeable:

StealthChop vs Constant T-Off

I wondered why that might be and I did a little research about that chopping method and saw that they claim it helps minimize zero-crossing positioning error. That got me thinking and I calculated the volumetric flow of the Titan Aero per full step of the extruder and compared that to the length of extrusion required to consume that same volume of filament (accounting for the fact that the edges of the extrusion are actually radiused). Lo and behold, the spacing between the ripples in my prints match the volume of extrusion per full step of the extruder almost exactly for all of the layer heights I have tested. See a screengrab of my calculations below (note: I'm tessting with a 0.6mm nozzle):

Extruder Ripple Calculator

I figured this all out pretty late last night, so I haven't had time to experiment more, but my next step is to switch the extruder stepper to the other driver that I am currently not using to see if that helps. However, after a little more math, I'm not wildly optimistic that even the most precise stepper driver would fix this.

I have a 400 step/rev stepper with a 16 tooth pulley on my towers, giving me 12.5 full steps/mm of travel for the motion components, which seems reasonable that the human eye would not likely detect the stepping error at that close proximity.

The extruder, on the other hand, with a 400 step/rev stepper, a 3:1 gear ratio and a 7.3mm hob printing with a 0.6mm nozzle at a 0.2mm layer height has a 2.86 MM/STEP ratio when you compare the volume of extrusion per full step to the motion required to consume that volume. In other words, even with a geared extruder and a 0.9° stepper, the extrusion is 35.75 TIMES less accurate than the motion of the carriage...

It seems to me that the only way to remedy that is to run an extremely high gear ratio on the extruder which would cause significant speed issues, or run a filament diameter smaller than the nozzle diameter to alleviate the natural "inverse gearing" that happens when you squeeze a 1.75mm filament through a 0.4mm hole (inverse volumetric ratio of roughly 19:1 by my math).

Someone, please tell me my math is wrong because otherwise, this is a pretty sad realization...

@dc42 , any thoughts on this?

@JoergS5 : that's definitely a possibility, but I have tried two different brand new E3D steppers (compact and pancake) and had the exact same defect so I doubt that's my issue in this case.

@kuhnikuehnast : what kind of extruder are you running?

@garyd9 , sure thing. See the link below:

https://forum.e3d-online.com/threads/titan-aero-repeating-pulse-pattern-on-prints.3138/#post-32237

@Phaedrux , thanks for the tips! I definitely agree that this thing appears to be pretty picky. I have already clocked the body and adjusted the large gear on the hob to get the hob and filament path to align. I have replacement bearings on their way as I type this, so I can eliminate that as the issue soon.

The only other thing that I can think of is where the stepper shaft gets centered into the heat sink. Since there is not a bearing in there and I'm sure that the idler is moving the shaft around a little, I wonder if the shaft is actually making contact with the aluminum housing and causing a repeating stick-slip condition? Hmmm...

If I can't get this resolved shortly, I'm probably going to nuke the Titan Aero and go get an effector from 713 Maker and put an E3D V6 and a Zesty Nimble on it since I had such good luck with that basic setup in the past. I'll report back on whatever I find that actually fixes this!