Differences in PLA vs. layer evenness

I'm starting to become painfully aware of how different various PLA filaments can be, even from the same seller.
And no, it's not just the temperature, as most posts would lead you to believe.

I've been using Faberdashery's electric blue PLA yesterday and couldn't, for the life of me, get it to print nicely. The layers are horribly uneven. Okay, maybe not horribly, but very very noticeably, especially at oblique lighting angles. Faberdashery's black, on the other hand, prints awesome and very even, especially if you lower the temps a bit (counter intuitively - most tips i've heard suggest increasing the temperature to get more even flow).

Now, at first, i thought it's a simple matter of lighting and texture. But putting a few test prints side by side, it gets obvious that the more noticeable unevenness is not an artifact of lighting (sorry, don't have a photo right now). Then i thought, maybe it's temperature, and played around a bit. And nope - same deal, regardless of temperature (tried everything from 180C all the way up to 210C).

And then i've noticed something... Faberdashery's black, as well as Ultimachine silver and some other filaments that i've found print really nice and even, all have one common characteristic - they don't break easily. As in, if you make a free air extrusion and bend it, you can bend it nearly 180 degrees 5-6 times before it breaks. In fact, some, like Ultimachine silver, i have to cut with scissors when there's a strand of extrusion left hanging cold from the nozzle. I can bend them and twist them to no end and they simply don't snap or break.

Faberdashery electric blue, on the other hand, as well as Ultimachine white and black, break easily, with a loud SNAP sound. And all of these tend to print unevenly.

I've been thinking what the cause of that correlation might be and so far, my working theory is that the brittle filaments get mangled much more by the hobbed bolt, as they're less elastic (and less plastic). Instead of the teeth digging into the filament and deforming it, chunks of filament break off, clogging the bolt teeth and slipping of the filament during retraction and post-retract extrusion. Sure enough, after using the electric blue for a few hours, i've removed the idler block and the bolt really was pretty clogged up with it.

It might be a good idea to research this further and find a way to categorize the various PLA filaments that are on offer by its mechanical properties and "neatness" of print results...

Edited 3 time(s). Last edit at 09/30/2012 04:29PM by orcinus.

I've tried testing the hypothesis of filament "crushing" caused by it's brittleness being the cause of uneven layers (or, rather, causing shifts in flow from layer to layer) by reducing the retract and printing out an object with a minimum number of extruder speed changes / starts / stops (a very sparsely filled 20mm calibration cube).

As expected, the unevenness was minimal and there's a noticeable period to it:



(And before someone says it, no, that's not wobble.)

As a sidenote, the filament in question clogged the barrel of my hotend not long after printing this.
That's on an Arcol.hu with more than ample cooling. The last time this happened (months ago) was with Ultimachine white, which i've mentioned in the original post as being similar to this one in its mechanical properties.

Edited 1 time(s). Last edit at 10/01/2012 11:55AM by orcinus.

Looks like z-leadscrew wobble to me, looking at the corner that is closest to the camera. Print and install some z-isolators, like this one, and see if it improves it: [www.thingiverse.com]
Another thing to check for wobbly walls is extruder calibraton, especially fine calibration. If this is even slightly out, the inner perimeter pushes the outer perimeter out with random results. See [github.com]
Having said that, your print really doesn't look too bad!
I do agree with you about some PLA being more difficult to print - I find decreasing the temperature until stringing goes away (or increase until it starts) gives different temperatures for different PLAs. I have used the Faberdashery PLAs (I got a 'rainbow' pack a while ago) and while generally very consistent, I think I found the light blue the most different from the others, needing higher temperatures. This could be due to a number of reasons - perhaps it was stored differently, or absorbed more water (always worth drying it out if it's particularly troublesome), or was extruded too hot, or had some extra additive. Really hard to say. Some of the PLA from China at the moment has conditioners put into the PLA to lower the melt point and make it easier to print. Not sure about that myself.

It's not Z wobble, like i've said.

a) it's not present on other filaments (apart from the ones mentioned)
b) it's not present on thin-shell objects (continuous extrusions, very little starts and stops)
c) it changes depending on the shape of the object
d) its period is not a multiple of the lead screw period

B would point towards it being an issue with extruder calibration (as you've said), however, i can get it to show in single-wall objects as well, with no fill - as long as the perimeter is not simple, but consists of a lot of direction changes.

That said, my rig does have some wobble (ORDBot Hadron w/ spiral cut clamp type couplers), but it only shows on very tall objects, because it's constrained to the top of the threaded rods. That does show up and shows up much more noticeably (note that the 20mm box shown was printed at 0.2mm layer height) with the period corresponding to the period of the threaded rods.

Re: temperature, i've tried everything from 180 to 210 deg C with this filament. It was even worse at higher temperatures (which is counter-intuitive - i'd expect the flow to be more even at higher temps). The sweet spot seems to be around 185, at least as measured by my hotend (which was within 3-4 degrees off from the temperature inside the nozzle, measured with a K thermocouple).

The bottom line is, i think i've mostly eliminated everything but the mechanical properties of the filament itself as the source of issues.
I'm glad to hear i'm not the only one finding the electric blue Faberdashery PLA finicky.

FWIW, i've had only good experiences with their black. I've printed it from 175-215 deg C without any issues. In fact, i've been experimenting with surface finish by varying the temperature. There's quite a noticeable difference in gloss when printed at 180ish vs. 210 deg C (curiously, there's little to no difference in texture w/ temperature when using the electric blue).

Forgot to mention...

This is what a relatively neat print looks like with that particular PLA filament.
What i haven't posted is the prints i've rejected (herringbone gears), because they were too uneven - and those had variations in Z that were completely random. No periodicity *at all* to them.

It's actually kinda surprising to me that the reprap community hasn't gotten more bent out of shape on the issue of filament variability, since

a) the single largest cost over time will be filament
b) filament quality dictates in large the quality of the printed object(s)

I think the only thing that will change their behavior is pressure from the reprap community and demands for proper characterizations of the product from both manufacturers AND the import merchants who distribute those wares. Word of mouth is only as good as "who you know". Real industry would never get away with the shoddy quality control these folks are profiting from while hiding behind the "open source" Moniker, as if open-source is an excuse for selling crap..

I wouldn't call Faberdashery PLA crap. Most of the filaments i've tried were awesome.
Ultimachine's silver was awesome too.

It's more of a "relative" thing than something "absolute" - i.e. i'm not complaining about it being crap, i'm complaining about some PLA being crappiER than other PLA from the same vendor. I know there are variations in glass transition and melting temperature from color to color, due to additives and pigments used, but i never paid attention to mechanical differences, which are - at least so it seems - a pretty big factor in print quality.

Edited 1 time(s). Last edit at 10/01/2012 06:14PM by orcinus.

Some PLA needs to be extruded at a much lower temperature than you have tried. Some need as low as 160c.

For quality I was always thinking the same as xiando in we need it perfect but then it occurred to me that it makes no difference if the filament changes diameter often as long as the average size is used in the slicer. This is because we are using the filament as a push rod and the diameter gets evened out in the melt zone. This not to say really crappy stuff works fine but rather that its the quality of the resin that matters most and then the extrusion quality comes second.

Is the banding possibly a result of the filament not unrolling from the spool correctly and the tension is causing the problem.

Also is your 0.2mm layer height a multiple of your Z_steps_per_mm

---

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

Orcinus, I didn't accuse any one company. I stated that I am surprised at the lack of concern by the 3D printing community at the inconsistent quality (crap) that import houses and domestic distributors and the manufacturers they ultimately represent are passing off as quality filament. And it seems to apply to almost every source.

Especially considering that

a) Plastic filament represents the single larges long term expense for someone in or getting into this project.
b) Filament quality dictates in large the quality of the printed object(s) (beyond initial tuning)

Edited 1 time(s). Last edit at 10/01/2012 06:52PM by xiando.

Sublime Wrote:
-------------------------------------------------------

> Also is your 0.2mm layer height a multiple of your
> Z_steps_per_mm

I was thinking along the same lines. It would explain the regular pattern in the "wobble".
I haven't figured out how it would show on some filaments and not on others though...
But then again I only have experience with clear PLA.

evil B

Once again, the photo i've posted illustrates the pressure variations i've been talking about mostly eliminated.
The regular pattern shown above isn't the original issue i've been talking about.

The original issue, caused by the filament, is a *random* banding pattern.
Here's an example, exacerbated by a smaller nozzle (0.35mm instead of 0.5mm) and retract turned on again:




To answer your question - no, it wasn't a multiple, although it was close.
The "skip" caused by a mismatch in layer height vs. Z steps is completely obfuscated by the banding seen in the photos i've posted (it gets noticeable with other filaments).

Oh, and i went down as low as 170 with this particular filament. Anything below 180 was unusable (spotty or would stop flowing completely). And there was no spool, i've unrolled a length of filament for testing purposes and had it strung from a desk lamp.

Edited 1 time(s). Last edit at 10/02/2012 02:15PM by orcinus.

xiando Wrote:
-------------------------------------------------------
...
> Especially considering that
>
> a) Plastic filament represents the single larges
> long term expense for someone in or getting into
> this project.
> b) Filament quality dictates in large the quality
> of the printed object(s) (beyond initial tuning)

I hear you there.
The amount of posts, wiki entries and blog posts about printers far far outweigh the number of those devoted to printing stock.

OK one last thing. It could be your E_steps are too high (or extrusion multiplier). This results in good extrusion until the extra steps mean it is trying to extrude too much (wider bands) and once the pressure goes up high enough the hobbed bolt starts to slip and you get less flow (narrower bands) and once the pressure has dropped off from slipping you are back at good flow and wider bands. This would repeat over and over and leave a result like you see.

---

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

Did you try this calibration I suggested in the 3rd post? [github.com]

Once again, this does not happen with all the filaments. Just a few of them (a minority).
And yes, i did go through the calibration.

Also, it does not, for example, happen with hollow single wall objects (because they pretty much get printed in one continuous extrusion, especially if you turn the retract off), regardless of the filament used.

Look, it's not like i'm trying to debug the basics. I've been printing just fine for a while now. I'm just trying to figure out why *some* particular filaments result in uneven layers. And so far, my working theory is as stated - slippage due to mechanical properties.

For the regular banding from the first shot, on the other hand, i'm not sure what the cause might be, but it's nearly unnoticeable or completely unnoticeable with most filaments (and only somewhat noticeable with the "misbehaving" ones). So i'm not too bothered with it to think about it yet

NB: my large extruder gear has a slight eccentricity, so that might very well be part of the cause for the regular banding - and it might be getting more visible with filaments more prone to slippage i.e. more sensitive to the bolt grip.

Quote

Also, it does not, for example, happen with hollow single wall objects (because they pretty much get printed in one continuous extrusion, especially if you turn the retract off), regardless of the filament used.

This to me says it's probably a calibration issue.
It demonstrates that the plastic can be put down in layers consistently, i.e. it's not something like the filament diameter varying.
I assume you've measured the filament, and I assume you've measured the single wall extrusion width, and verified it matches the expected value, specifically with this filament, because it can vary with the filament diameter and how hard the filament is, because the latter effectively changes the radius of the hobbed bolt (albeit marginally) because the bite will be a different depth.

Practically there are few things that could cause this,
Inconsistent extrusion, the fact a single walled piece doesn't show it tends to imply this is not the case.
Too much plastic volume, the inner layers randomly displace the outer ones.
Temperature too high, upper layers deposited before the lower ones cool, but the effect would be worse with a single wall piece
Some sort of backlash or missed steps in the retract, leading to inconsistent volume deposited after the retract, this is possible, again the filament being harder or softer could result in the force requires to turn the hobbed bolt during retract being higher, but this would require the torque to be pretty marginal as it is.

There isn't much else to it as far as I can see, I don't know what else in the filament could cause what you're seeing.

Quote
droftarts
Did you try this calibration I suggested in the 3rd post? [github.com]

Why does that instruction use the ratio of the square of the theoretical over the actual extrusion width? I just tried it and it didn't work right. I had to use the simple ratio to set the extrusion multiplier. If the extrusion multiplier is changing the actual feed length by that value, then taking the square of the ratio is wrong. The extrusion width should be linear with the feed length since the extruden filament layer thickness and length are not changed.

Edited 1 time(s). Last edit at 10/02/2012 07:22PM by brnrd.

Perhaps the density of the filament is changing along it's length. This would give these strange results, assuming accurate calibration. And explain why it is more difficult to extrude at some points than others. But virtually impossible to detect!

Polygonhell Wrote:
-------------------------------------------------------
> This to me says it's probably a calibration
> issue.
> It demonstrates that the plastic can be put down
> in layers consistently, i.e. it's not something
> like the filament diameter varying.

Ah, but here's the thing.
If i print a single wall object with an extremely sparse fill (so that it cannot influence the width of a layer - say 10%), the problem is still there.
And it does not vary with fill.

The only variable i've found that seems to have any meaningful correlation (thus far) seems to be the number of starts and stops, or number of retracts.
In fact, if i print a single wall object that has a non-trivial shape (like the polyhedral cup), the problem pops up again.


droftarts Wrote:
-------------------------------------------------------
> Perhaps the density of the filament is changing
> along it's length. This would give these strange
> results, assuming accurate calibration. And
> explain why it is more difficult to extrude at
> some points than others. But virtually impossible
> to detect!

It could be. It might be that the radius is changing too, because - i admit - i haven't measured it except for the first half a meter or so.
I should probably measure it more thoroughly to exclude that. Also, i haven't tried printing at more than 210 degrees.
The prints got progressively worse up to 210, so i never went above that. Maybe there's a sweet spot between 210 and 220.
I'll redo the calibration again, just in case too.

How accurate of a scale do you have? Cut peices to the exact length using a caliper, say... 3 inches of each, then weight them, what do you get? A "Grain" scale would be best (Reloading scale)

I've got a scale i used to use for measuring out the fixer and developer powder back in my darkroom days. I'll have to dig it out of storage, i've got no idea where it ended up...

In the meantime, here's something bizarre:



Ever seen diagonal banding before?

In other news, i found out what has the greatest effect on "random" banding depicted here.
Cooling! Or, rather, fans! And no, i'm not talking about not having fans or about a lack of cooling.
It's the cooling/fans that causes the banding.



Left = fans off (which is why the top few layers on the side farthest from the camera sagged)
Right = fans on

No idea what's going on here. I've got one stupid hypothesis - which is that the airflow is too strong and is actually blowing the weaker-adhering layers away a fraction. And i've got a less stupid, but pretty fuzzy hypothesis - which is that some filaments react weirdly swell/contract-wise when heated and cooled quickly.

Edited 2 time(s). Last edit at 10/04/2012 03:57PM by orcinus.

Could the diagonal banding be coming from the infill?

As far as the fans, you don't want the fan to blow air on the heating block/nozzle. How many fans do you have? Sounds like too much air.

Two fans, one 40mm and one 50mm (blowing from opposite sides). I'll measure the nozzle temp with a thermocouple with both on vs. off.
The thing that throws me off is, if it was as simple as that (too much heat being drained away from the nozzle), it would've been a "stable" effect, not something as random as shown.

Re: diagonal banding - there is no fill.
It's a single-perimeter print.

Edited 2 time(s). Last edit at 10/04/2012 06:06PM by orcinus.

It just hit me... Since i have both fans blowing in opposing directions, i'd imagine the area around the nozzle is pretty vortexy.
It might be that the randomness is coming from the chaos of air currents...

If your fans are mounted on the x-carriage, then it can also be from vibration.

They are, but it's a pretty heavy, rigid aluminum carriage and they're attached with PLA clips.
But good idea, it's something to look into.

That bizarre diagonal banding might've been caused by some weird resonance.

I think i'm ready to give up on trying to get more out of this filament.
This is about as good a print (at 0.25-ish mm) i can get out of it:



The only way i can get decently even layers is with absolutely NO cooling at all (fans off) and a relatively straightforward single-shell geometries.
Which isn't very useful.

Dunno, maybe i'm being too picky and critical.
I mean, it's not like the large gear above is ready for the scrap drawer, it's useable and doesn't really look *that* bad...

Edit: Yes, i know there are gaps in the surface. I had the extrusion multiplier turned down while printing it, because i wanted to see if that will have any effect on the random banding. It didn't.

Edited 2 time(s). Last edit at 10/07/2012 09:58PM by orcinus.

Oh, yeah, one more thing... I've found another variable that helps alleviate the uneven layers somewhat - acceleration and jerk.
Bizzarely, if i increase them, the banding gets reduced. That's not a good solution, though, as it starts impairing the overall quality and reduces available cooling time.

Plus, i don't feel comfortable with my printer "walking" and "skipping" across the table

Probably retract settings and how much forward retraction on restart needs calibration.

my guess is also the retraction federate is too high and feedstock that is stripped slightly is being fed back in and it does not have as good a grip. try lowering feed rate for retraction to 10 or less , and retraction to something like 3, then start at 0.1 for forward retraction and work up from there in 0.1 increments until that diagonal banding disappears.