Erratic printing on Mini

I hope this is the right place for some help with erratic printing on a Mini. My use is for designing and making components for converting manual microscopes to motorized. I have had the Mini for about a year, but don’t print every day and still consider myself a newbee.

The problem I am having is artifacts/errors that creep into an otherwise good print with PLA, using Cura and various profiles. There are several types of errors:

  1. random screwups,
  2. regular wavy patterns on the edges of parts,
  3. irregular lines where the infill zigzag shows on the outside.

I usually print at .25 thickness to keep the print time down. A uniform finish on the parts looks good, rather than going for higher resolution.

The picture shows an electronics enclosure top piece. It was printed upside down (top was bottom). Some layers are good, others are bad.

I can’t tell where the problem lies:

  1. filament quality
  2. settings - temp, speed
  3. cura limitations

General suggestions would be greatly appreciated.

I think it looks like over extrusion.

Measure the filament diameter in a few locations and put the average in the software

Turn the flow rate down to about 90%. Ideally, calibrate the e-steps for the extruder, but this is a good first start.

Set the first layer line width to 100% (default is 120% or so)

You may also need to set the Z-offset up a little. Start the print but cancel after the skirt. When the bed cools, remove the skirt and measure the height. Compare to the first layer height setting in Cura. Default is about 0.4mm, so if your skirt is 0.2mm, you’re too close. Go to machine settings and set Z-Offset to 0.2mm to compensate. Do this last as the extrusion rate can complicate the measurement.

Thanks ttbbal, will try these changes and see what happens. In the middle of a 3.5 hr run right now.

I measured the diameter and it is very good, right around 3mm.

Any wisdom on extrusion temp? I have tried from 180 to 205F on PLA but don’t seem to have a best place.

Is there a best/worst filament brand? I have been going for the cheapest one, since I can’t tell one from the other.

You need to be pretty accurate in measuring your filament diameter. Mine usually measures around 2.85 or 2.90 and changing slightly makes a slight difference.

It looks like you could benefit from a fine tuning like @ttabbal said. After checking and adjusting all the things he mentioned, I was printing with much better results. Every little thing helps.

Since you have access to microscopes presumably, look at the bottom of your first layer to see how it’s laying down. I have spent way too much time looking at my 5mm x 10mm x 15mm test brick I drew up in tinkercad. It only takes minutes to print and you can make changes and compare brick to brick.

Good luck, and please post back with your results and with any questions. If you aren’t sure how to do something, ask. Most here are very helpful and won’t just tell you to “duh, go do a search newbie”.

Keep trying. It gets better. Usually…

OK, put in the suggestions above (first layer 100%, flow rate 90%, first layer height default ~.4mm ). The filament diameter was 3.997, 3.045, 2.970, so I choose 3mm and began to print a part. I have been having sticking problems and the part is big, so I raised the bed temp to 70F. It printed the border OK, but began to peel as it printed the first few circles, so I aborted.

I decided to make the first layer thicker, so I set that at .5mm. That began to peel, so I decided that I should reverse course and make the layer thinner (.25), thinking that would lower the head. I printed a 5 x 10 x 15 mm block to check and it looked OK.

I printed another run and the results are shown in the pics below. The 2nd pic is the Cura first slice pattern to show what is was supposed to look like. The 1st is the beginning of the print. Note the circles didn’t even complete and the filament “slobbered” from jump to jump. This makes me think there is something wrong in the .stl file, or the sliced file, even though the slicer layers showed up OK in Cura.

As a final test, I went back to the CAD dwg, went to a second copy, hoping it might be clean, converted it to .stl, and ran it in the printer. Results were worse than the picture attached.

I think the problem is in the CAD file, so I printed the mating part with only 4 holes and it came out all right. Will go back and redraw the original part.
IMG_5701 2.JPG

OK, another try. Machine off overnight.

I completely redrew the part from scratch. Converted it to stl (high resolution) and tried to print. The results were identical to the other bad prints. First layer circles did not complete, slobbered between jumps, peeled the filament off the board. I tried 90% flow and 100% flow to see if that made any difference. It didn’t. I let it run to start the fill-in between circles and border and it did that OK. Since the part was unusable, I cancelled the print.

I have printed several versions of this part before with good results. I changed it a little and tried to reprint. That is when the trouble began.

I have sent a note with the stl to Lulzbot support to see what they say.

The first layer in the pic looks quite high. I find 0.4mm first layers work well for me, I use that regardless of the actual layer height. It also helps get that first layer locked down to the PEI.

Speaking of that, you have sticking problems with PLA?!?!? That’s weird. Maybe try to clean the PEI with alcohol? Light sanding with very fine sandpaper, 600 grit or higher? You might try a different brand or color as well. I’ve only used a little PLA, eSun. It seemed to work well for me.

Get a blue scotchbrite pad, heat the nozzle to print temps, then use the pad to clean the nozzle really well. Watch the auto-leveling process to make sure it doesn’t push the bed down. I would expect this to result in the first layer being too close, but it doesn’t hurt to be sure.

Try 0.4mm first layer, 90% flow, and 4 skirt lines (the … button next to platform adhesion type). Cancel after the skirt prints. Let the bed cool and pull the skirt and measure the height with a micrometer in a few places. I suspect the first layer issue is that the z-offset needs to be adjusted.

OK, I have implemented many of the suggestions above: diameter, flow %, first layer thickness, etc. Thanks for all the suggestions.

I finally determined that the starting problem was the PLA not sticking to the platform. It seemed to change suddenly and I didn’t recognize what was happening. Everything it the photo above can be explained by the filament not sticking. I cleaned the platform with alcohol and sticking improved dramatically.

Implementing the filament and feed rate suggestions, etc. greatly reduced the poor registration of the layers, shown in another picture.

I still have lots of questions. Probably too many for a single post. Regardless, I will list some of them below.

I really don’t know what to expect as the best possible print quality. I don’t mean the highest resolution, I mean the most uniform print.

  1. I get pits where the print head changes direction/jumping to a different track. There seem to be all kinds of irregularities in the print process.
  2. Sometimes the filament will retract when jumping between places, and sometimes it will drag a filament across the surface. This looks bad on the top or bottom of a piece.
  3. When the print finishes, the head retracts slowly, leaving a vertical strand of filament that has to be cut and smoothed. It could do a retraction and jump that works well in other places. Why is this?
  4. Warping at the edges of a flat surface. I have tried 195-205F for my PLA, but can’t seem to find a best place. I have gone from 50-70C on the platform, again, with no silver bullet. Some items warp, some don’t.
  5. The head cleaning process doesn’t work most of the time. I usually clean the head before starting and the 3 minutes or so of robo cleaning is a waste of time. I have an idea for eliminating the need for head cleaning. If vertical blocks were placed on the corners, then the head could touch on the square shoulder and not the nozzle tip. PUtting a 1mm gap between the nozzle tip and the disk would virtually eliminate the need to have a clean tip at the start.
  6. Retracting and extruding the filament at the start doesn’t seem to work correctly. I have to manually push about 5-10mm into the nozzle to get it to start correctly. I presume there are mods to the start code that might help.

My intention is to produce a medical product for sale this year, using 3D printing for the electrical and controls enclosures and for the structural components that mount on a microscope.

From one noob to another…i had remarkably similar problems on my previous DIY/kit printer. Sledehammer fixed most of the warping with Elmers Glue and more heat. Not totally though. Ended up filleting corners and cutting back on infill. Enough to mitigate the warping enough for my application.

Still, though…i would get failures in the same areas. I too, falsely accused my slice/gcode.

Ultimate resolution came from drying out my PLA and ABS. I know a lot of people leave theirs sitting around in the open and never have problems. I didn’t either until I kept failing on these 2 parts. Big difference between a Pikachu and what I was trying to make.

Don’t discount the moisture. While I’m a noob at printing, I am an injection molder by trade. ABS HAS to be dried before it is molded. I gather it’s not as picky when it comes to printing. But, I gather PLA is also worse with the moisture thing.

I got new filament and printed no problem with no changes to the printer. I keep it all in a tub with those microwavable silica thingies now. Just today, i went and used some of the old stuff that’s been in there for 2 weeks and on the old printer. The “part from hell” came right out.

Bottom line…if you’re not keeping your filament dry…its one more variable that can be in play. And i would assume it can impact material flow…and just about everything else. I found that out the hard way.

Differences in the finish at layer changes and such is pretty normal. Slowing the speed down can help, but it’s tricky to completely eliminate them.

Retract is determined by the slicer. You might try changing settings, or even slicer software. Some programs work better for different parts. As this is a custom design, there may be issues there as well. If nothing else, you are introducing another variable.

I’m not sure why the default end script does that. It’s never bothered me particularly, but it wouldn’t be hard to add a couple lines to the end script in the advanced settings to change it to do a quick retract, then move the head away before going up.

Warping at corners and such can be an issue. It’s not usually as big an issue with PLA. Try cleaning and sanding the PEI. Rounded corners and such seem to help. Lower the infill some as well. You can also try printing a brim, but you have to remove it later which can be a pain.

You can remove the cleaning from the start script. If you do, make certain you clean manually. I find it generally works alright for me. The bulk of the time is spent heating/cooling which you need for the autolevel to prevent filament from oozing from the nozzle and messing that up. Something like your suggestion to not use the nozzle for the leveling probe could help there, but would be less accurate and need more user calibration to work well. You also couldn’t have it hang below the nozzle, or it would hit the bed/parts while printing…

There are some start script changes that help with the initial priming. I print an “anchor” off to the side in the start script. The downside is you need to make sure your parts don’t print there or you print in it. :slight_smile: Adding another line or two to the skirt can help here as well.

It might not be practical for your needs, but there are reasons injection molding is used for commercial parts. Per part it’s cheaper, and more repeatable. Printing is great for prototypes, but results are often not as clean. You might also consider HIPS for enclosures. It’s easy to print with almost no warping. Little things like keeping filament dry will make a difference in final finish as well, so do try that as well.

Thanks for all the good suggestions. I have put many of them into practice and they have improved my print quality greatly. I have changed the flow %, first layer thickness, speeds, bed temp, and the combing (off). Turning the combing off has been the biggest help. The stringing is almost eliminated (see pic). I added a brim/skirt to one part, but getting it off was no fun. Will see if I can solve the problem another way.

I understand the question of the best way to manufacture these parts. Initially, the volume will be low, and we may run into end user variations that require mods to the design. As the volume picks up, we can adjust the manufacturing method to suit. We intend to be upfront about 3D printing to the customer, who will be “friendlies” initially. Their response will help with the decisions about fit and finish.

Molding of any kind requires a volume we have to reach before it is feasable. I have designed some blind corners into the structure that might be a problem for molding. We will see. Some parts could be metal, others need to be plastic. I can get them all to work OK with 3D plastic (PLA). Haven’t looked into 3D printing hubs yet.

I have been through maybe 100 iterations on the various parts of this device, which I never could have done using conventional prototype machining - just too costly.

As to the head calibration using posts to touch the square body of the hot end: the posts would be on the platform, not the head. They would only touch at two places on opposite sides of the nozzle (not 4 sides as you would imagine). The error would be whatever error there was in the square block the nozzle screws into. With parts made in CNC mills, that should be minimum. The calibration could be done while the head is heating from ambient to extrusion. This should eliminate all the retraction/extrusion mismatches - maybe do 5mm extrusion to purge the nozzle (similar to your anchor). The expansion of the square block during heating might be the main error source, which I don’t think would be significant, since the calibration doesn’t take as long as the heating.

As to the programming. I know enough to be dangerous, but not enough to be proficient. I can add and delete commands, but I don’t know gcode well enough to stay out of trouble. Any suggestions would be appreciated.

BTW, I mounted a small 8in fluorescent tube up the right hand vertical post, which helps visibility a lot - and makes the filament look thicker than it is.

I think you may have moved on from your original grey box.
I have been struggling with a similar look when printing on my TAZ6.
It happens when printing an object with overhangs of about 45 degrees or so.
My solution to the problem happened in 2 steps.
1 - In CURA, Fill Density(%) setting, Next to that is a “…” box.
Clicking that box will tell you that default infill overlap is 15%
I reduced the infill overlap to 1% and the very bad looking anomaly is gone.
2- Also, I ended up slowing down the Outer Shell Speed (mm/s) to 20.

I’m running these parts using ABS material with 100% infill because I need very maximum strength on
these parts, but I see it on all of my projects with overhangs.

If the photo imports correctly, the one on the left with green tape was my 1st attempt at printing in ABS material.
With the one on the Right without the green tape, the only change I made to the print profile was reducing the
Infill overlap(%) to 1%

When Watching that bad anomaly on enough parts, I finally started considering the layer before and the layer after
the layer currently printing.

Just an observation
I hope it helps you in some manner.


Thanks for the interesting info. I will try and see what happens to my prints. I am trying to keep my total print time down and I don’t know if your setups will do to that. Right now, I am looking at about 10 hrs of printing for the controls and probably the same for the motor mounts. So every hour counts.

After I got my prints looking better, (temps, speeds, combing…), I still had some long stringing - say 2 inches from place to place. The interesting thing is these strings were tiny and they didn’t sag. I now the rule of thumb for unsupported horizontal bridging is 5mm, but if anyone could control this tiny diameter stringing, distances over an inch (25mm) could be achieved. I don’t know if a greater retraction for the combing feature would help prevent this (don’t see any control).

The second goofy thing I noticed was some screw holes got much smaller. I presume that was due to slower printing, giving the filament time to spread out. Now I have to recalibrate my holes.
also, when two parts fit together, I am experimenting with duplicating the interior part, expanding it by 1%, then subtracting it from the other at the joint area.

Finally, I noticed that my .100 walls (2.5mm) and 1mm shell thickness don’t seem to get along very well. There is a 0.5mm fill in, which eats a lot of time and a lot of jumping back and forth. I wonder if the “pros” pay attention to the wall thickness vs shell thickness to minimize the stitching. I could use a 2mm wall and 1mm shell, or a 2mm wall. Wonder if anyone has experience with this.

Your latest grey box photo looks like you achieved a very good final result.

I believe your stringing problem might have to do with the slots in your case.
In the Expert Config area, In CURA the expert config area is not available until you are in “Full Settings”
I think that’s where you adjusted the combing, there is a minimal extrusion before retracting thing.
You might want to play with that setting, but too many retractions over a very short amount of extruded filament will result in
stripping your filament.
The grey box is a bit of a tricky design.

As for your wall thickness
The only way I can ever get clean smooth easy and fastest walls is to use whole number wall thicknesses in my can drawings.
1mm is good
2mm is good
3mm is good

1.5mm leaves a gap that is filled in with infill material - time consuming and printer does not run in a smooth manner.
2.5mm is the same as the 1.5 in that the gap is filled in with infill material.

It appears to work perfectly fine to set shell thickness at 2mm when your walls are less than 2mm thick.
Curves are sometimes tricky if there are not enough segments in the curve of your STL’s.

I made a sample of wall thickness behavior last year that confirms what I’m saying here.
I tried to snap a photo, but my camera is not able to catch the small details of that sample.

the long horizontal bridging works best with large z axis steps like .38 mm or larger.
The smaller your z axis steps are, the worse any filament will bridge.

I hope my observations will help you in some manner.

I am approaching the time when I need to make some betas for the field testing. These will go in medical/research labs and I want the best results I can get appearance wise. I have twiddled with all the settings as mentioned above and can get a pretty decent part.

Moisture seems to makes a big difference. I found some plastic containers with tight fitting lids that I use for storage with the dessicant inside. I tried putting a roll in a small oven for several hours, which seemed to work, but came close to creating a mess.

I would like some help with three subjects.

  1. Head cleaning. The Mini’s head cleaning doesn’t work. I have to manually clean the head 80% of the time. So I would like to comment out the code for cleaning.

  2. Filament feed. The filament never extrudes enough at the start of a print. A Lulzbot support guy told me to manually feed the filament right before printing. I would like to add code in the start script to prime an extra 10-15mm before printing. I have some parts too big to have a border around before printing.

  3. Quick retract at end of printing. Several of my parts have the top side showing and the final spider web at the end of printing doesn’t finish up well on these parts. I would like to add something to the end script to fix this.

Here is my problem. Every time I comment out some of the code, something wierd happens and the unit wants to run wild, or head for a crash. I tried adding extra filament feed before printing in the start script, but something is happening in the Cure run time that defeats my efforts. Obviously, I am not a good enough programmer to get it right.

I would appreciate it if anyone could give me the code to accomplish the tasks above.