I bought my first 3D printer this week, a Bambu Lab X1 carbon. It printed this egg in just over 1hr, with no issues or stringing, no supports. Used generic filament too. So far I’m blown away with this printer.
Exactly , just dropped the file in, sliced then printed. NO changes to default and perfect print in an hour. I am on over 1500hrs printed so far. Only cleaning/greasing and belt tightening maintenance. So between your new out the box and my seasoned X1C fast accurate prints are straight forward on some printers.
I'm really proud of the progress that 3D printing has made. I'm still using my first 3D printer of an Ender 3 Pro, and hearing people like you getting into 3D printing with such significantly better printers makes me excited about what will be available years down the line when I decide to upgrade.
I think you need to ask yourself "why" does z-hop cause stringing when just horizontal movements don't? The solution comes out of that question. I think the difference in stringing between z-hop and no z-hop is two fold: 1) When you lift the nozzle directly off of the print surface, the action of doing this creates a suction effect which pulls some of the molten plastic back down the nozzle. You then move the head, stringing the molten plastic. 2) When you move horizontally (instead of lifting) you avoid the suction effect, and you create a wiping effect which helps clean the nozzle, avoiding stringing. With these two effects in mind, I believe you can see why your "diagonal Z hop" produces the effects that it is. I can see two possible solutions to this. First solution: Start the hop maneuver AFTER you have already horizontally moved off of the print surface (and you are in the air), thereby maintaining the wiping effect, and avoiding the suction effect. Second solution: Begin the hop by doing a quick filament retraction, and at the same time an in-place head orbit, providing the wiping effect, toward the end of the orbit start increasing your z height. In practice this looks like a corkscrew movement, as you begin your hop. The filament should be continuously retracted throughout the corkscrew movement, providing a reverse suction effect, to counter act the somewhat reduced suction effect (reduced because of the spiral lift) that will remain from lifting off the surface. I believe Bambu Lab is using this method in their slicer (this is what my X1C looks like it's doing), and it seems to work pretty well.
Solution 1 wouldn't solve the core problem, though, unless I'm missing something. The entire reason for doing the z-hop is to not move horizontally while the nozzle is still at risk of hitting the print. Moving horizontally to avoid stringing before hopping has most of the risk of collisions that not hopping at all does, I would think.
@@davydatwood3158 it depends entirely where the point of failure is. There are three potential failure modes here: 1) You are printing along, finish a run, and as you move to your next print location, you run into a protrusion close to where you were just printing. 2) You finish a run, and you run into a protrusion somewhere between where you just finished and your destination. 3) You finish a run, and you run into a protrusion very close to your destination print site. Number 1: doesn't cause failure all that often because the area is fresh and warm, and deforms easily, and so bumps tend to move out of the way. Number 2: happens some times (though most slicers try to avoid unnecessary travel over the model), but since it should be an island, you should jump over it with the first solution I proposed, as you are doing a z-lift as soon as you get into free air. Also, you could prioritize finding free air to make that z travel. Number 3: is the failure mode I observe most often, because the destination is usually cooled significantly (as its often gone the longest without being touched), thus it's already pretty rigid, and the print head has reached it's maximum travel speed, so it hits with the most force. The first solution completely addresses this failure mode.
@@bujin5455 Yes, I've seen that specially in supports. The print head can freely move away from the curling little horns (I hate those...), but when it's reaching the destination it will knock it out easily as it hits the curled up horns sharply. Happens a heck lot with top parts of a circle.
Love your way of explaining...Just sliced a model part in the z hop script that usually strings bad no matter how I tweak it. Probably 70% less which as i produce models to sell means less clean up time. Always something new to try....thanks for the effort....
This has been added to Orca Slicer (i switched from Cura, its been amazing). Under the Printer Settings --> Extruder --> Retraction --> Z hop type --> Slope Switching from Normal (vertical) to Slope solved my stringing problems. I was trying to modify temperature, retraction length, retraction speed, drying filament, etc, you know all of the traditional things that can help. The only thing that worked was disabling z hop. Of course you lose the benefits of z hop, and risk knocking off prints. Then I came across your video and discovered that its already in the slicer! I reset my settings and only changed to use Z hop Slope, and the stringing is virtually gone! Im at 220 C temp for PLA, 0.8mm retraction distance for a DD extruder, 45 mm/s retraction speed. Pretty normal I think. Anyways, this makes me love Orca slicer even more. Orca has all the best features from the other slicers and some unique ones.
I would've implemented this at a constant diagonal angle, like 10-20 degrees from horizontal, up until it hits the desired z-hop height. That seems like the best way to avoid those curled-up bits consistently, and also makes small hops much less of an issue.
I think this has great potential. I think it might work better if the travel path is shaped more like this _/---| basically, move horizontal first as if no z hop at all, but then quickly go up at an angle to max z hop height to get as much benefit as possible. Then go straight down to the part to print in case the part you are about to print to has curled up. I think that would make it less likely to catch the nozzle than dropping diagonally to that point.
This is pretty much exactly what happens in Orcaslicer if you enable slope zhop. It works extremely well and I was able to print the torture egg on the first try with nearly zero stringing.
There's two factors contributing to the failure: (1) The warped protrusions that the head colides with. This is the real issue, and is the first thing to fix, if possible. (2) Bed adhesion... if the part is immoveably stuck to the bed, it will help avoid disaster when collisions do occur. Michael, along with all the 3D printing channel Gurus, have long since adopted their favourite systems. Some time back, I watched an excellent series, pre-dating 3D printers, on Rapid Prototyping by Dan Gelbart. He explained that if you clean a surface with Isopropyl Alchohol, or products like thinners, you actually leave a layer of hydrocarbons. Although it is typically only one atom thick, it acts as a separation layer. I think he said that a detergent which can be rinsed off completely is preferable. Ivan Miranda demonstrated his preferred method in a video: basically clean it with dishwash liquid, (no additives, colouring, perfumes), rinse with hot water, pat dry with a paper towel (I just shake my glass bed until only the tiniest of droplets remain; the heated bed gets rid of those.) From that point on *** DO NOT TOUCH IT... E V E R ! ! ! *** This has worked for me, but Michael's test model will be a challenge for any method. Watching with interest!
I swear all my devices are listening to me :) I'm having an issue with my very large printer hotend dragging, pulling and damaging the print. Miraculously this comes up in my you tube feed today. Life saver!!!
While I can understand the 'up' portion of the diagonal move, I'm not sure the 'down' portion of the diagonal move actually contributes much to the improved print quality. If the benefit of the diagonal move is to clear the nozzle from curled edges, you would theoretically want to reach the maximum height of the z-hop as soon as possible (at an angle that still provided the low-stringing benefit of course), sustain that z-position, then drop to the model at the intended destination (such as in your alternate diagonal move) Moving diagonally upwards the entire travel move seems to just increase the risk of clipping some element of the model over long distance travel. So perhaps you could include another parameter that controls the total distance traveled before the maximum z-hop height is achieved.
Do you have "wipe while retracting" set in PrusaSlicer (unsure what it is in cura) when enabling z-hop? I would expect that what your change does is similar to: retract while moving the nozzle over printed stuff to basically wipe it clean and the do z-movements to avoid hitting stuff. That's what I would expect "wipe while retracting" to do when z-hop is enabled: retract and wipe and then move up. So, did you have that option set?
I remember years ago an experimental slicing tech where, at the end of a perimeter move, it would partially re-trace the perimeter to wipe the tool and smoothly lift off the part following that direction and then when it was off the part would curve mid-air towards the next extrusion target.
It seems like the moving horizontally has the effect of wiping the nozzle clean. I wonder if a method of retract, then move horizontally off the part, then ramping upward would be effective. Then a simple lowering like you do in the alt method and restoring the filament back to the nozzle would work. Further, to prevent whatever limited filament might be dragged off the edge, a brief reverse in horizontal direction might keep more material away from the part edge. Whatever might remain would potentially have minimal impact on the next layer deposited. I"m still new to all this, so I don't have the experience to predict the possible results.
I think the alt z-hop that would work best is the one that starts at the same level at the print, retracts (and in it's movement away acts as a sort of wipe), but then change height to max z-hop height during travel, to then stop and lower into place over the next starting point. It accomplishes the "wipe" when leaving, and the z-hop to avoid little lips (as you pointed out) when moving back to a section.
while watching the first part of the video i was thinking that it would work better if the z hop only goes up vertically and then drops straight down to avoid hitting curled up edges. im glad i watched to whole video before commenting because apparently you had the same idea ;)
I messed with this via cura post processing script a while back (I"m older so a while back was probbaly almost 2 years lol). Not for stringing issues but for a smoother surface when using z hop. Lifting straight up was leaving a stippled texture. Never thought about it for stringing nor tested it for that. My end print macro does pull away up and diagonal now so makes sense that could help. I may have to dig the code back out and play with it some more. Maybe I'll release a cura post processing script for it.
I think this will work amazing. Industrial machines already use this idea. My jobs laser cutter let's you choose been curved, diagonal, out rectangular movement when moving from part to part.
Within orca slicer which is a fork of Bambu slice you can use the spiral z hop that Bambu uses on their machine. I use orca for my ender 3 pro. Printing one of these now with spiral z hop.
A simpler and potentially safer (in regards to avoiding collisions) solution would be a short horizontal wipe while retracting and then a normal z hop before completing the full travel move.
I'm curious about Patrick's egg. His looks perfect and even zooming in you can see almost none of the overhang jaggies yours has. Did he expound on his settings to get that and his printer? I'm still not sure what causes stringing with z-hop and why you don't get any if you don't do it and go in a straight line.
This is a thought i had for a gcode modification that i dont have the skills for but could reduce travel times in prints. What i have noticed is if you align the zseam to one spot like a corner the slicer will be starting and ending each wall in relatively the same spot. So i thought why have it stop, retract, move, prime, print when it could just continue from the end of one wall to the adjacent start of the next wall out. Effectively connecting all the walls into one cohesive snaking wall for each layer. To me this seems like it could reduce travel times pretty well.
You should do triangular Z hop. Move diagonally and land vertically (as a rectangular triangle). Edit: Ahaha ... end of video :D The best would be to use curve instead of direct lines. Or at least first move is horizontal for 5-10%, then move to diagonal until vertical.
Watching the animation showing the difference between standard travel z hop and diagonal z hop really clarified this for me. I'm new to 3d printing so I am taking in all the info I can and this is great stuff. But one thing hit me: Why do the hop from 2 edges that at close together instead of from the "outer" edges? Using the tower you showed in this video would it help to do the hops at the outside edge, travel over the printed surface doing the diagonal hop and then "land" on the outer edge on the other tower followed by moving inwards as those layers are applied? I suspect I'm not describing it well but I would think that the stringing could be reduced if the strings are connected inside the print surface instead of at the edges.
I think you have to avoid any nozzle movement but the horizontal one just after the retraction - as far as I get it that’s the main cause of reduced stringing when no Z-hop is in use. The Z-hop has to be done just before the nozzle reaches the next extrusion point to avoid the collision. Of course the transition between movements can be fluent like S-shape to reduce resonances, however it’s a matter of fine tuning the whole idea.
Clever work around and perfect timing because it may not be needed soon, I just saw a solid state cooling fan with no moving parts at CES, super cool and would be perfect for 3d printing. And what do you know the issue of the print not sticking is directly related to cooling. I wonder how much speed we're actually loosing because of non optimal cooling. I see fan upgrades in the 3d print community all the time but nothing much more said after that. The company claims it can improve laptops performance x2 is how bad cooling is these days , giving them another 5-10w of headroom, which is alot if your processor is only 15w to begin with so a pretty major breakthough in coolin technology i cant wait to see it applied to 3d printing. Maybe you could mod an ender 3 to print at 500m/s with that kind of fan? I saw the video on PCworld's channel that shows it off at CES if anyones interested and Frore Systems is the company. It looks like a flat lithium battery that apparenty is filled with these small vibrating membranes that suck air in at incredible speeds and blow it out the end.
When looking at the alternate diagonal z hop, it seams that the stringing is more at short travel moves. I guess thats the result of a bigger angle between horizontal and the travel move. Maybe you get better results by defining the diagonal angle to be constant instead of the travel height. You could still have a maximum height and after reaching that a horizontal move.
In general the diagonal Z hop is similar to the wiping option in the slicer, providing a little more surface and time to capture oozing filament. A short retraction as part of the Z hop may also help the process for objects with small surfaces. And you may get even more improved results by accurately tuning Linear Advance, and of course doing all you can to avoid ooze and wet filament that generates uneven pressure. Oiling the filament (e.g., with a paper towel dipped in vegetable oil) is another old-fashioned way to reduce ooze because it produces a slightly more regular extrusion. The object I sometimes use for this kind of stress testing is the classic Voronoi Skull at a very small size. Even before we had Linear Advance I was able to tune my old crufty Prusa i3 to the point where the output was almost completely string free. Anyway, diagonal Z hop is a fine variant of wiping and it would make a useful OctoPrint plugin while we await the slicer option.
I think it's worthwile and should be introduced into cura.... I will test it out (and if it doesn't work you can bet you'll hear about it :D - politely of course, I make betas too)
Will test soon. But: I think the results will be even better with a reversed start-section. What I mean is the following: Z-hop to the opposite moving first for a short bit and then do the real motion, a bit overshoot and move back. I'm pretty sure this can eliminate the stringing even further.
Basically a wipe in the reverse direction? I was thinking something similar. It might be a bit tricky though because it might not be obvious which way to move back if there are many short segments.
Really interesting your research and testing. Very thanks for your resources in your calibration site and videos. Did you already talk about a way to recover a failed print? I mean a way to launch a new print in top of existing failed print by offseting Z and chopping the gcode at the right "height" or maybe another technique to get it. This would be super useful for those big prints that fail at the very end.
My bed is only 140mm but the higher the object gets the worse the effect seems to be... as the head moves between the left and right extremes it seems to dig harder into the surfaces at the sides... I was expecting it to dig more in the middle due to rail droop, but it is like the bed istelf is drooped... unlikely as it is a sheet of glass. Calibration of first layer is critical to adhesion. My bed is calibrated well in general and after I put some hair spray on the plate I can get several prints in without releasing, while still not needing the giymnastics needed to remove from glue stick... Sort out the 1st layer, maybe increase temp to 215C for layer one to make it more smearable, and it will stick
Can you have like 10% of the first movement purely horizontal, then use the alternate.. So that oozing of the PLA is not going up but over the part without lifting up a single mm.. (Hope you can understand my idea) Thank you for this great content!!
Are you using prusa slicer? Isn't there a wipe settings in there? I'm pretty sure that's what I turned on in order to remove a lot stringing. I'm not going to claim that I'm trying to print anything nearly as torturous as this. But anyway
Just a thought: Cura reduced stringing by ending each layer on the inside and the nozzle essentially scrapes off would-be stringing. I seem to remember you doing a video on why it had less stringing than Slic3r. What if you set the path to only raise z after some percent of the path of some distance, ex 1mm? Z hop: |---------| Your first z hop: /\ Your second z hop: /| Scraping z hop: _/| Hard to show in ASCII, but hope that makes sense!
I tried the Alternate Z Hop and it defiantly made an improvement to the stringing. (have pics if you want) is there any way to wipe the nozzle and then have it perform the Z Hop?
To be honest - I use special 3d printing glue, rather than stick glue. It helps the detail stay in place. Why I am mentioning that is 'cause when the detail sits in one place the heat from the nozzle wouldn't just knock off the print, but it will actually melt the tiny bit of curl on the edges. So far, for 5 years it proved me that it's completely perfect solution. To be fair - I am printing Voronoi and web type of prints, never-ever had to use Zhop. May be you (all of you) should try using glue too. I am suggesting Dimafix Pen 3d. :)
new idea, what if you print 2.. 3.. 4 layers of one side, then start the zhop? of course that could use some way to detect the nozzle cleareance/ height? :D
Interesting idea but I think it's introducing more issues than resolving, especially seeing that printed egg. Shouldn't better retraction while z-hop tunning solve that issue? Especially seeing that filament leftover while doing initial z hop is making that stringing while dragging it through part.
What if you started the Z-hop by having the nozzle first move diagonally up and away from the final direction of travel? That way the strings might be "wiped off" in that reverse movement and stay within the walls of the print? If that makes any sense...
Can you share a downloadable script? Exporting from slicer, postprocessing on site and uploading gcode by hand is not most convinient way when you are on Klipper and not supposed to run with flash drive😄. I'd better insert it in postprocessors in superslicer
Hey there Michael, could you add a Z speed to the gcode generators on your site? I have some modded printers that use a Tr8x2 (vs the TR8x8 stock) lead screws but the Z speeds on the tutorials is much higher than the Z motors can handle (basically any z movement just grinds then the height is not correct) love your content. I was just bummed when i realized my big brain leadscrew upgrade mean in needed to slow Z speeds.
Egg falls off bed. Me: Add brim, set bed to 70(actually 65 when I measure it with my main printer) and try again. Depending on where the extra buildup is, I'd probably do a combination of coasting, neg prime, increased retraction, lower hotend temp, full cooling, max travel and raise jerk speed. ..and as a separate note, not try it in PETG ever
I just tried a flipo that I failed 4 times to print (previously) on my ender 3v2 with a traditionaly diagonal z hop @ 5mm/2mm - printed with similar though beter results, now trying the alternate alternate provided worst results, simply a glump of pla around the nozzle object was thingiverse # 4591728 though I layed it "flat" and the slicer didn't think it needed supports at 50x50x15mm and I sliced it using creality's slicer (cura based) with super quality defaults and 100% (geometric) infill
Very cool idea. I’ve always thought it would make more sense to have an extrusion path finish then retract followed by a wipe move away then a vertical lift. Followed by a travel move. All the benefits of a hop without stringing.
Wipe while retracting is an option in Superslicer. Has about a dozen options for adjusting it's behavior too. Prusaslicer has it too, possibly other slicers.
@@LightCarver these options exist in Cura too(wipe distance within wall/path after retraction etc). Also there is an option for coasting within wall/path before/after retraction, before movement to other parts of the print or layers.
When printing parts with little first layer contact area, I add a 5mm brim and use glue stick. You can also add a little paint on supports even though you may not need them. The support base adds a little sticking action to small first layer parts. I also slice the model using Prusa's cut function. This adds the extra step of gluing the parts together and adding a little filler to sand. I greatly admire your skills at programming G-code.. very impressive work on developing the diagonal Z hop. Nice work!
My idea for this was to removed the original z raise and add the Z height increase to the destination gcode before the gcode for lowing back down on the Z. This is diagonal on the take off, but normal Z lower as you go back down to the part. I have created post script I run from excel VBA code to achieve this. (I see you got to this point as well. Didn't watch the video first.. lol)
I think you're right with this. The was the diagonal hop is done in the video would still lead to a collision or still have a higher chance of one. Your method actually gets to clearance height where it matters, not in the middle where the hop height doesn't make a difference.
I have a related idea which I've been wanting to try, but I'm not quite good enough at G-Code yet. Let's call it something like "Non-Planar Bridging". I'm wondering if we extruded bridges in just the right arc/parabola/catenary shape, whether we could eliminate sagging.
I don't believe that this is possible. But I can give it a try and it's an interesting idea. You don't expect the extruded bead to fully cure while you're bridging. Indeed thus helps you keep it somewhat clean as surface tension helps you fight gravity and maintain low sag. So when you draw your bridge faster, it actually succeeds more. You can also underextrude the bridge, since area will drop a little with extrusion width but volume and weight will drop much faster. But very slow bridges cure more and maybe it is a thing that can be done.
It really seems like the more blended the axis movements become the better the prints turn out. Eventually I'd like to see a slicer that moves more organically mixing all 3 axis smoothly.
The reason I use z-hop is to avoid hitting printed parts, as you demonstrated with the curling edges of the egg. Your modification appears to ignore this necessity by making the upwards travel move so slight over larger distances that there is probably more likelihood of it hitting the printed part on larger prints. I would suggest you need a slight modification to your 'alternate' approach which is to make three movements; first a short diagonal move upwards which introduces the required height for the z-hop, but also keeps the nozzle moving horizontally, then move the rest of the way horizontally and finally descend as appropriate to the required location. Imagine the path like the flight of a plane; short and quick ascent, linear travel, and then descend. That would appear to satisfy everything. As for how well it prints... let's see?
I know the X1C does a Z hop while printing but I just that ran model on my printer and it printed fine, no stringing, on default settings in 1 hr. Uploaded for proof.
Most of my stringing and overhang issues have been remedied by setting layer height to 0.16, 100% fan, retract & then wipe towards infill before moving to another area. No Z-Hop necessary.
Important thing is when retraction occurs. For long time retracting in slicer was done WHILE moving up Zhop. This was perfect receipt for string to pull. You can fine tune it with "wipe". Enforcing 50-90% of retraction before wipe will break string before mowing up. I checked 2.5 PS and this behavior is corrected. Still by using last few mm of path as place to wipe /not much settings in PS/ you could better break string. A blob will start to form, but a "reverse wipe" on landing could hide it. And after a month of tuning... :D
I think this is about 80% right, and I'm really intrigued by your thoughts here - I think there's an additional constraint/limit here... you need to leave at the start of travel, and arrive at the end of travel, at a certain movement angle, say 30 degrees, or so... if you do a long travel, you are risking hitting curled edges somewhere, since you're moving mostly horizontally at one or both ends. This means a move is in three phases... leave at a degree of uptake until (and if) you reach your z-hop height, then a horizontal travel section if needed, then at the right point, start your descent into the end of the move at the arrival angle. This means more complicated gcode and possibly accel/decel delays, but the angle you leave and arrive at, needs to be constrained to keep from these quasi-horizontal start/end of the move.
That was exactly my thought too. His issue with short travels was that the angle was too steep. Rather than limit the minimum distance, he needs to control the angle. It ends up looking a lot like an acceleration curve: a fixed angle up to the desired height, a flat travel at that height, then a fixed angle down to the end of the travel.
Here's an idea to tweak your alternate Z hop. If you can travel off the part horizontally and then once clear start the alternate hop motion. This way you still get the wiping action as the nozzle clears the part.
Outer perimeters need to be printed last, and the nozzle needs to move to the point where the nozzle has the shortest path to exit the printed area, even if it means moving away from the next printing point, otherwise you may hit your already printed section (curled borders). Once the nozzle is outside, lift and move it towards the next printing point, going down vertically. Also, retractions may need to be tweaked.
Cura already does such thing; parameter is called "avoid printed parts" and it can also avoid support material. It does choose between a z hop or an avoidant travel move though, but never both simultaneously.
Why don't you just use "wipe while retracting"? It will do a horizontal movement first then and the Z-Hop later. With Z-Hop + Wipe enabled I print dry PLA without any stringing.
Michael, I attempted to print some POM in my X1C today, and at one point I slowed the print speed down to 10mm/sec, and I noticed the X1C does something my Prusas do not-the print head does a little curly-q motion on retractions, so in effect it’s kind of doing what you’re suggesting here. Try it on your X1C and see what I mean. Very interesting.
What if, in the slicer, travel moves avoided areas near overhangs? I can still see your method hitting the curl in some cases on certain models. If there was a setting that said the travel move must begin away from any overhangs (including a margin), and would also need to avoid printed areas, I think it might be the best way forward.
i had thought about adding a slight raise during bridging to keep sagging down as it has potentially time to harden more of a "stretched" state so it's essentially bowing up to compensate for the sag
In manufacturing with beams or other structural steel, such as the main rails of trailers, etc this concept is called “pre-cambering” they camber the material so that it’s bowed upward when there is no load, but after the structure is loaded,it supports it flat
I think stringing is reduced with no z-hop because the bit of oozing filament is wiped off as the head moves sideways away from the just printed part. You could experiment with „micro wipes“. Basically retract, then do a little wiggle before hopping.
I am curious how someone who uses FW retraction and sets Z hop there could benefit from this? I found this Spiral Zhop setting in Orca which led me here. Orca has me wondering too since I usually use FW retraction. Wondering how this will react using Orca FW retraction and Zhop Alternatives?
