Print Quality Pointers

Hi,

I've been trying to improve my Tantillus by updating and reprinting parts as I had major issues with layer alignment. So far I have replaced the Z-axis with a Cable-Z, reprinted the X and Y ends and printed bearings, and reprinted the lower X/Y gears and extruder small gear. Here is my latest print, which was printed using medium model parameters from Sublime's calculator and a relatively low speed/high quality setting in Kisslicer. There is still some banding, but what I'd like to know is whether it is possible to improve on this further, and by what means. The print was almost spoilt by a big shift in the glass about 4mm into the print. (ridge in rear leg). Could the other bands be caused by minute movements in the glass bed/perspex interface? I'd be grateful for any clues.



Cheers

Greg

I would say yes the glass moving could be causing this. Yes I think you can get it better, but white can make the lines show up way more with oblique lighting angles.

Are the cables musically tight?
Is the hotend really rigid in the carriage? If it moves wrap some tape around the top of the hotend and reinstall it.
Did your Z steps per mm turn out to be the same as mine at 100.66? If not did you use the intermediate calculator and enter your steps per mm?
Do the Z smooth rods move? If yes then tape the ends so they are tighter in the Z brackets.

The written instructions do not mention them and they are not part of the BOM so you most likely do not have them but just in case you noticed the springs under the bed in the rendering and added them we don't actually want them because it causes these kinds of effects.

---

FFF Settings Calculator	Gcode post processors	Geometric Object Deposition Tool Blog
Tantillus.org	Mini Printable Lathe	How NOT to install a Pololu driver

Thanks for the suggestions. I went round the X/Y cables and tightened them all (tuned to middle C I think) I also took more care with the clamping of the glass plate, but I think that I still need to work on the clamping solution (stronger bulldog clips and larger glass area, particularly towards the front to give more bearing surface. I checked the hot-end and that is solid.

Yes, I just set Z steps to 100.66 and I get nice round holes and good accuracy on the 20x20x10 calibration block. The rods did move in the sockets the first time I built up the z-axis. I put a layer or two of tape around all four ends and now it there is absolutely no play.

No bed springs. One thing I do notice is that the left z-axis fork is about 1.5 mm higher than the right fork, which makes bed levelling a little challenging. I thought that the cable may have been pulling asymmetrically, so I made sure that the cable ends were directly above one another, but it still pulls the left side higher than the right. I don't think this is a cause of the wobble though.

I'm wondering if the z-axis bearings are a little rough. The rods are very smooth tool steel. Perhaps it may be worth investing in higher quality bearings.

Different model, but same settings as before

If the bearings are tight then yes you could be getting uneven movement of one side over the other. I would take a heat gun or hair dryer and warm up the Z axis parts (be careful and go slowly, you only want to just get it warm) and then let them cool in place. That would allow the bearings to align themselves with the rods as well as get the two arms the same height. Once it is all aligned the Z axis should be able to fall on its own under gravity (with the motor unplugged). If the bed can not fall from the force of gravity the bearing may be too tight. If it just takes a small nudge to get it falling you should be alright.

---

FFF Settings Calculator	Gcode post processors	Geometric Object Deposition Tool Blog
Tantillus.org	Mini Printable Lathe	How NOT to install a Pololu driver

Thanks again. I'll give that a go.

A lot of people with Mendel variant machines swear by using layer heights = to whole steps. It may be worth trying to see if that helps, or even = to half stepping. To try you would enter the steps per mm in the calculator as 6.29125 for whole steps or 12.5825 for half steps.

---

FFF Settings Calculator	Gcode post processors	Geometric Object Deposition Tool Blog
Tantillus.org	Mini Printable Lathe	How NOT to install a Pololu driver

I printed that elepant and I can see why the glass would shift. Because it has some quite steep overhangs it likes to curl and then the head smashes into the curl. Tips on getting less curl would be welcome. I printed another one with the fans running much slower and it I got less curl and less head collissions, is that expected? does high fan speed make more curl?

Its a cool print by the way. It got dibs straight away.

Your print looks pretty good. I think mine is roughly the same in quality.

I can understand those Mendel machines have Z issues if they have two steppers in parallel. The back emf of the motors upset the microsteps I experimented with steppers in parallel. Each motor should have its own driver so the constant current can work properly.

Quote
Robonz
I printed that elepant and I can see why the glass would shift. Because it has some quite steep overhangs it likes to curl and then the head smashes into the curl. Tips on getting less curl would be welcome. I printed another one with the fans running much slower and it I got less curl and less head collissions, is that expected? does high fan speed make more curl?

Its a cool print by the way. It got dibs straight away.

Your print looks pretty good. I think mine is roughly the same in quality.

I can understand those Mendel machines have Z issues if they have two steppers in parallel. The back emf of the motors upset the microsteps I experimented with steppers in parallel. Each motor should have its own driver so the constant current can work properly.

If the material is prone to shrinkage like ABS and some PLA mixtures then yes the fans will make it curl more. If the material is like pure 4043D PLA then lots of cooling will harden it instantly and you get very little curling.

For me I prefer to run with lots of cooling and a low enough print speed and temperature that the entire layer is cool before the next layer goes on. This way you get very little curling at the price of speed. For some objects that are prone to curling and need very few retractions I would also reduce the travel speed to ensure if it collides it does not skip steps. And lastly is use support. I find Kisslicers support is easy enough to remove on most objects that I use it fairly often now.

---

FFF Settings Calculator	Gcode post processors	Geometric Object Deposition Tool Blog
Tantillus.org	Mini Printable Lathe	How NOT to install a Pololu driver

What speeds are you using?

I used m/s
loops=24
Solid=21
Sparse=31

I find that if I use higher speeds there is a risk of axis screws loosening. I guess I should check before each print, as something always goes wrong when I push the envelope speed wise.

Going to assume you mean mm/s :-)

Quote
beekeg
I used mm/s
loops=24
Solid=21
Sparse=31

I find that if I use higher speeds there is a risk of axis screws loosening. I guess I should check before each print, as something always goes wrong when I push the envelope speed wise.

You can use removable locktite on the screws to prevent them from coming loose.

One reason they may be coming loose when trying to push the machine faster is because the motors overheat and that makes the gear warm enough that they come loose. If the motors are too warm to touch then I would reduce the driver current a little.

You can also put a drop of crazy glue in the shaft holes on the gears so it glues the gears to the ends of the shafts so that the gears can not turn if the screws do come loose. If you ever need to remove them you just put a drop or two of acetone in the same place and it will soften the crazy glue.

Edited 1 time(s). Last edit at 04/07/2014 02:40PM by Sublime.

---

FFF Settings Calculator	Gcode post processors	Geometric Object Deposition Tool Blog
Tantillus.org	Mini Printable Lathe	How NOT to install a Pololu driver

Quote
wingnut
Going to assume you mean mm/s :-)

m/s would explain the problems I'm having with the axis screws coming loose......and embedding themselves in the wall

I'll try some locktite and check/adjust motor current.

Thanks!

Greg

p.s. tried the hairdryer tip to level/free up the z-axis and it worked a treat

Quote
beekeg
tried the hairdryer tip to level/free up the z-axis and it worked a treat

Do let us know if it made any changes to the print quality.

---

FFF Settings Calculator	Gcode post processors	Geometric Object Deposition Tool Blog
Tantillus.org	Mini Printable Lathe	How NOT to install a Pololu driver

I think that the side-to-side shift has all but gone. I am beginning to think that the banding is due either to variances in the z-axis step or is related to an as-yet undiagnosed extrusion issue. Here are some pictures of a new elephant. It started off printing well and then about 2 mm in it the extruded width reduced so that the 25% infill became stringy and the two skin loops became distinct from one another. You can see this in the first picure. I put some additional handraulic pressure on the filament to rescue the situation and raised the hotend temperature slightly. All was well until half way up the legs where another series of uneven layers occurred. These can just be seen in the on the front leg in the second photo.




I am a little perplexed at the random nature of the layer variability which make it hard for me to diagnose the issue. Some areas are really good such as the top of the elephants back, and I feel tantalisingly close to what I consider to be good output.

I am sure I have asked some of these before and you can just ignore those ones.

Have you lubricated the threaded rod? It should help reduce the friction and thus allow it to move in microsteps more easily.

Have you tried to increase the Z axis current? This should help it overcome any friction and get it to move better. But if too high it can make microstepping inaccurate (too high is a warm/hot motor).

What speeds are you printing at? I find higher speeds give a better outer finish.

Is the extruder idler tight? If you remove the filament from the bowden tube (un-load) are the marks on the filament from the hobbed bolt fairly regular or are there notches or only teeth marks every few mm's.

Does the temperature stay some what stable? Less than 7c swing in either direction or better.

Is the filament consistent in diameter? If it varies much it could cause the extrusion width to vary.

Are the cables musically tight?

Are the X and Y gears secure on the motor shafts?

Is the hotend tight in the carriage? No wiggle.

---

FFF Settings Calculator	Gcode post processors	Geometric Object Deposition Tool Blog
Tantillus.org	Mini Printable Lathe	How NOT to install a Pololu driver

Your last picture looks to me like you've got a layer shift in one dimension (x or y). On one side of the model, there's a cut inward, and exactly opposite is a layer bulge. I'd study that axis closely.

If you are using a threaded rod for your Z lift you can try my z wobble hack. its a really quick easy upgrade. I would be interested to see your feedback.

As robster said, you are definately getting an x or a y shift that can be seen clearly above the elephants eye. I would make sure the x and y are really smooth. Turn it by hand with the power off. If there are any bumps or stiff points you need to resolve them. Also with the motor current, I just turn them up as high as I can get away with. If the stepper drivers are not too hot to hold your finger on you prolly need to turn them up. If they are too high you will get thermal cut out., e,g. the motors will shut off every couple of seconds. I set them to where I can just hold my hand on the heatsink and then turn them up a little more.

As far as the extruder goes. Have you got the big die springs as WWFP designed in? His mod works great. Its in the metric editiion hacks on github.

Cheers
Keith

Quote
Sublime
What speeds are you printing at? I find higher speeds give a better outer finish.

Did you mean *lower* here?

Quote
wingnut

Quote
Sublime
What speeds are you printing at? I find higher speeds give a better outer finish.

Did you mean *lower* here?

Nope I meant higher. It was actually a complaint I had with the wording of Kisslicer with its speed slider. I am sure on some printers a lower speed is more accurate but not with all.

Now I do not mean faster is simply better as there are other things that effect the quality. If you go too fast you will have too much pressure in the bowden tube which will cause blobs at the retraction points and not give the last layer enough time to cool. But at the same time too slow and you will have too much heat travel up the hotend and cause high back pressure and too much heat being transferred to the printed part from the hotend being above it too long. When I want smooth parts like this wing [www.youtube.com] I run the machine at 120mm/s at 0.149mm layers. That wing is large enough that cooling is not an issue at those speeds. My calculator has a size setting because we want to print fast but the smaller the object the more time we need to let it cool so it slows down for smaller objects.

---

FFF Settings Calculator	Gcode post processors	Geometric Object Deposition Tool Blog
Tantillus.org	Mini Printable Lathe	How NOT to install a Pololu driver

Quote
Robonz
If you are using a threaded rod for your Z lift you can try my z wobble hack. its a really quick easy upgrade. I would be interested to see your feedback.

As robster said, you are definately getting an x or a y shift that can be seen clearly above the elephants eye. I would make sure the x and y are really smooth. Turn it by hand with the power off. If there are any bumps or stiff points you need to resolve them. Also with the motor current, I just turn them up as high as I can get away with. If the stepper drivers are not too hot to hold your finger on you prolly need to turn them up. If they are too high you will get thermal cut out., e,g. the motors will shut off every couple of seconds. I set them to where I can just hold my hand on the heatsink and then turn them up a little more.

As far as the extruder goes. Have you got the big die springs as WWFP designed in? His mod works great. Its in the metric edition hacks on github.

Cheers
Keith

I'm using a cable-z lift. The shift above the eye is interesting as it only appear on the head and not on the main body. There is slight delamination above the left eye, which is not clear from the photo. Could this be due to cooling differential? Having looked at the model in more detail I think that Robster and you are right that some of the variation is due to shift across the bed L-R. I will review the drive. it is strange that the shift occurs randomly and is then consistent and then shifts back to where it was. The other interesting area is the top of the leg sockets. This discontinuity appears to have caused the deep depression/offset on the side of the body, or of course it may be a coincidence.

I printed the big die spring holders, but to be honest I didn't know how to use them

Thanks for all your comments/suggestions so far. This is a great way of learning what influences the quality/reliability if the printer.

Cheers

Greg

Quoth Sublime: "Are the X and Y gears secure on the motor shafts?" And the gears on the rods.

In my very limited experience, your print artifacts are consistent with a gear that's able to wiggle slightly on a shaft infrequently.

Hi Greg

In the metric github there is a really nice drawing showing where all the parts go. Check it out.

You can see the die spring holders on my crappy video Look around 12 seconds to 30 seconds. The die spring plate is white, look for that too as you need to print a total of 5 parts.

Another good upgrade is to change your gears out for belts. It makes for a much quieter printer and doesnt wear so quick like the gears. No lubrication needed either.

Cheers
Keith

Hi, I checked the rod/gear tightness and all is good. I decreased the mesh slightly on one of the gears and now it runs a bit smoother. I'll also turn up the current a bit and see if it improves the accuracy in case there is a bit of cogging going on.

Thanks

Edited 1 time(s). Last edit at 04/23/2014 06:07PM by beekeg.

Hi Keith,

I see now that I need to use with the revised idler plate. What is the benefit of using die springs? Greater pressure or more ability to adapt to variations in filament diameter? Thanks for the pointers.

Hi Greg

The bigger springs give you more "fine" control on the tensioning as you are not cranking them all the way down. Too loose and you slip a notch or two then your hob will give bad results until you clean it. The main downside of it being too tight is your bowden rips out when you get a blockage. I guess you want it just tight enough where it will slip if you get a blockage.

WWFP swears that one of the most important things is to clean your nozzle regularly and I have to agree. That's why I put thumb nut on my head.My favourite upgrade is the heated bed though.

I got a laser cutter today! yay. Looks like about 60 hours ahead to install it though. Anyone need laser cutting in Auckland?

Cheers
Keith

Thanks Keith I'll see if I can find some suitable springs. Yes I cleaned out my nozzle today using the 80 degree and yank method. It wasn't blocked but it pulled out a load of sooty deposits on the surface of the 'plug'.

Those nuts look good. I may make something similar. I have some knurled inserts that I use at present but they don't look as good as yours. Managing without a heated bed at present, but I do quite fancy the idea.

Wow a laser cutter - respect! What are you going to make with that?

I think I may have found the cause of the layer unevenness. Just after I cleaned the nozzle I checked the hotend for wobble and there was an alarming amount of play. Previously I had checked this when cold and all was well. That may explain why the first few layers are normally OK and then things start going wrong. Anyway I added more tape to make it super-rigid and here is the result.

I'm pleased with this., though I can see that I need to do some tuning to sort out the corner problem on the mouth, but I think its getting there. Big thanks to all you guys and especially Sublime for reminding me to recheck to hotend. I made this for my son, who's 18 and back off to University tomorrow and he thinks its cool, so it must be

Greg

That print looks great Greg, The dome looks really smooth. You must have spent some time working things out. If I had to guess the mouth is caused by retractions not being spot on. Being a bowden system you never get them perfect..

I make all kinds of things with the laser cutter. The cool thing about them over 3d printing is the speed at which they complete a job. So having a laser and a 3d printer means I can chop and change where needed.

Cheers
Keith

Looks good. I am glad you figured it out.

I would say the problem with the mouth is actually a problem with Kisslicers logic. It likes to place the perimeter seam on corners if possible and that is what it has done. You can get around that by turning the model 180 degrees in a modelling program before opening in it in Kisslicer (will not work if you rotate it in Kisslicer). This will force the seem to the other side where there are no corners for it to find and place the seam on. You could also use the Jitter function which will distribute the seam around the object and not use the corner logic.

---

FFF Settings Calculator	Gcode post processors	Geometric Object Deposition Tool Blog
Tantillus.org	Mini Printable Lathe	How NOT to install a Pololu driver

Thanks. I'll watch out for the seam issue in future.

Greg