Ultimaker 2 default nozzle size is 0.46 in Cura

[neotko 1,417]

I suppose. But also, like on filament there's a warning. There's no warning on e3d, etc nozzles.

Also if E3D nozzles are 0.35, 0.40 etc. I assume a error on the third decimal point, not on the 0.05 range, that error range it's a lot.

Also, if we pay for good stuff, we want the good stuff. Lack of precision it's something I left for my china shop adventures.

Edited March 21, 2016 by Guest

[neotko 1,417]

Anyhow, back to the topic. I though the main issue of extruding 0.40 on a 0.40 it's that the preassure changes. And we need to think 3dmensinally. We are extruding through a 0.40 circumference that will be squished on another area. Slicer makes the math to make that extrusion look like 0.40 making all the math. So, it's easier to extrude 0.46 on a 0.40 with more precision or it's just 'cleaner', or it's just that the software now assumes we use bad quality nozzles.

To understand more about nozzles I always recomend to see the video

they explain quite nicely the extrusion size output. But they never say something about a gigantic +-/0.05 error

[nilrog 103]

There is quality control, but I imagine that some tolerances are simply not reachable for a certain price. Yes you can have a nozzle with a 0.4mm size and a <0.01 mm tolerance, but what if you'd have to pay 2-5 times as much for that? Would that be worth it? That's pretty much the dilemma that you always face when making hardware.

 

If the advertised tolerance was publicly stated as +-0.05 mm it would perhaps be kind of okayish (but Ultimaker never state any tolerances). But still, you are saying that >10% tolerance is acceptable for a 0.4 nozzle that Ultimaker are selling...for a printer that prides itself on giving high quality prints? If that's true then I don't know what to say.

Ultimaker is not "a chinese seller" so I would expect better accuracy for the price we pay. If I go to ebay/aliexpress then, sure, I could see that the tolerances could be crap. Usually they are...but sometimes you are positively surprised.

[Nicolinux 288]

I'd say to not get hung up on the nozzle tolerance. The UM2 prints very well. The 0.35 nozzle setting is more like a safety setting in case some nozzles are not perfect. I guess it optimizes the last 5% print quality so that's not a big issue right?

And since you don't know how perfect E3D nozzles are (or where they are manufactured), maybe 0.05mm tolerance is standard...

[neotko 1,417]

I think it's quite something and a half... 0.05 tolerance. I sent an email to e3d so they put their input. I could understand a 0.00X tolerance, not a 0.05

[DidierKlein 729]

So to be clear i don't have information what the real tolerance is, i don't expect it to be 0.05 on our machines (or on e3d/3dsolex). It's just the software trick that gave the best results during testing, when you think about it, it makes a lot of sense.

It's much easier to have a cleaner print if you ask for a smaller line (even if the nozzle has a real 0.40mm nozzle). Less pressure in the nozzle, is less oozing, i meaning cleaner prints.

And you can print up to 150% of the nozzle size if you, so with a 0.40mm you can print 0.60 shells (it needs tweaking but it's possible).

[cloakfiend 995]

Regarding price, i could in theory pay ten times the price seeing as i never replace my nozzles. Ive been using the same one in one printer for over a year and a half and plan on never changing the other in my personal one unless i really need to.

Id happily pay 10 - 20 dollars or pounds for something that yielded better accuracy that something else, but seeing as its the prints and not machine part measurments that matter then im just going to assume im going to just have to figure it out myself.

[neotko 1,417]

I disagree and agree. I use a lot of 0.32 on my 0.40 and there's a limit to it, specially at some temps. I don't suffer this because 1.75 but. If you print too slow and have ooze issues (dripping) and you print an object with many retracts, there's a point where you can get false underextrusion because there's filament lost between print/retract/print. To combat this I use 1.75 but also for heavy retract prints (39665 in 3h) I can't go bellow 0.32 without small issues on text areas because even when my system has almost zero dripping I can't fight gravity. So when extruding that small, there's a factor of drips that depends on the real nozzle size. I fight this when I need faster/hotter prints, I use extra restar distance (s3d) but that's another history. So. Long history short. You need real numbers and tons of tests to balance a fake nozzle number.

Edited March 21, 2016 by Guest

[neotko 1,417]

Oh and an extra note. If my 3 e3d nozzle where different in 0.05 (bought them in different time periods) I would not be able to print with the same quality on three different printers. So.. Who makes 0.05 error nozzles? And why Ultimaker cares if isn't ultimaker?

[neotko 1,417]

Totally agree @cloakfiend that's why I can't wait to get the new veryhard anders-olsson nozzles. Also they print with an amazing precision (for better and worse if you don't control all the printing aspects). - Check this beta test for his super hard nozzles)

Edited March 21, 2016 by Guest

[cloakfiend 995]

i love ultra close pics that dont hide anything, and you sure have them. Its the anti vibration properties that realoy impresses me. I have no idea about esteps or what im meant to do with them but i do know i want a ruby nozzle! Lol. Thanks for the link!

[neotko 1,417]

Lol indeed. In the end the important part it's how the print looks. The less vibration on the ruby accoding to anders it's because on the tip of the ruby it cold down a bit and makes a better layer deposition. Ofc it's a freaking ruby so ofc it prints better than brass (according to some guy anders talked and say something much more scientific that my 'ofc it's freaking ruby') XD

[foehnsturm 969]

Well, in this case you shouldn't expect relative tolerances like +/- 5%. Typical radial runout errors of chucks are in the range of 20 - 50 µm, if they aren't super expensive. No matter what the bore diameter is - and this is mass production. I had thousands of nozzles manufactured for my wok burners with  0.58, 0.60, 0.65 and 0.70 mm. To my experience, burr free edges and minimum surface roughness are more important than accuracy better than +- 0.05 or 0.03 mm.

Edited March 21, 2016 by Guest

[gr5 2,167]

3dsolex indeed had a bad batch of nozzles where several hundred nozzles where .35 instead of .4mm. It was kind of obvious. .35mm is pretty damn far off.

I have "nozzle cleaners" that are .36mm and they fit loose in the .4mm and not at all in the .35mm nozzles. You could see it also. They were quite a bit bigger than the .25mm nozzles.

So I ended up selling them as ".35mm nozzles limited edition".

But the people who had those nozzles noticed. It was a pretty big deal. 3dsolex gave out quite a few free replacements of course.

In reality the tolerance is probably closer to .01mm on most nozzles. I mean for this kind of thing to happen someone has to use the wrong drill bit.

but drill bits can wear down and so holes I would expect might be smaller than planned but .35 is silly small. It's visible. I inspect every nozzle by eye and I can easily tell them all apart.

[Labern 774]

I eould say that the tollerances are close enough not to worry about. Also twist drills don't create round holes. So you want to get really picky they would need to be laser drilled or eb drilled or something to be more accurate.

But in the end it looks like UM have done a lot of tests and found that setting the line width in the new cura lower then the actual size improves the quality of the prints. After my tests I would have to agree. I have a bunch of E3d, 3Dsolex and um ones so will comparing 10 of each prove whether nozzle hole accuracy will have any effect or not in the line width theory be enough? I don't have S3D so I can't compare it to the new cura. The title of this topic is for old cura so it can be tested with that.

[yellowshark 153]

Ah manufacturing tolerance, now it all makes sense to me; following the thread from the start, nearly everything said just did not make any sense at all. If my printer has a measured 0.4 nozzle and can lay down a 0.4 line why would I want to overextrude by telling the software it was bigger, or vice versa and underextrude - and yes I may have those around the wrong way but that is irrelevant.

Ultimately there are just too many variables; has your filament got a tolerance of .005, no. Has you printer mechanics got a tolerance of 0.005, no. Has your bed levelling got a tolerance of .005, no. Etc etc.

[neotko 1,417]

Well I was surprised to see 0.05 as the tolerance. I buy faber because even when they advertise 0.05 it's more close to 0.01, and I suppose there are others that also like better stuff. If you say 0.01 I don't care, but if you say 0.05 it's normal, then I do care, because there are too many 0.05 already. 0.005, that's way beyond what I care for my prints.

[neotko 1,417]

I eould say that the tollerances are close enough not to worry about. Also twist drills don't create round holes. So you want to get really picky they would need to be laser drilled or eb drilled or something to be more accurate.

But in the end it looks like UM have done a lot of tests and found that setting the line width in the new cura lower then the actual size improves the quality of the prints. After my tests I would have to agree. I have a bunch of E3d, 3Dsolex and um ones so will comparing 10 of each prove whether nozzle hole accuracy will have any effect or not in the line width theory be enough? I don't have S3D so I can't compare it to the new cura. The title of this topic is for old cura so it can be tested with that.

 

I mentioned s3d not just for my post about nozzle size play. (i remember that IRobertI told me on that post how he already mentioned that on his 3dprint guide tells to play with it to solve walls not touching. I sayd s3d because it's curious that they also been doing something like that for years, but 0.48 as 'auto' extrusion width when you use a 0.4 nozzle.

Edited March 22, 2016 by Guest

[Labern 774]

Just to kill my own curiosity

All 0.4mm, Top E3D, Middle Ultimaker, bottom 3D Solex.

Edited March 22, 2016 by Guest

[fxtentacle 2]

I'm very sure that the new nozzle size is a bug. As you can see here, there's a bugfix in the pipeline because the Ultimaker 2+ profiles were erroneously applied for Ultimaker 2 prints:

https://github.com/daid/Cura/commit/0b229f4663eb8b2d401f000f2c73fa82a2c03b16

A nozzle size of 0.4 with the 2+ adjustments would return 0.456 which is rounded to 0.46.

So the "new" nozzle size is probably a bug.

[neotko 1,417]

Ok now we can shoot the messenger @didierklein

Btw this even if in the end it was just a bug it's been a very interesting post!

[gr5 2,167]

I suspect it's a double bug. The first bug (now fixed in git) was to accidentally apply it to non-plus machines. The second bug (not yet fixed) I think is to make the nozzle size 8/7 as big instead of 7/8 (8/7 is basically 1.14).

But this is just a guess.

Quite a long time ago there was a video with Erik (a founder) saying something like "there were many problems with print quality that we fixed using software instead of hardware" which was not expected (fixing quality issues with software). And he implied that this happened over and over and that each time they think they have a hardware problem they manage to overcome it with software. Well not *every* problem but an impressive amount.

I think it's interesting that this still seems to be true - that you can still make improvements simply in Cura and Marlin. And that all printer companies that use Cura and Marlin benefit - not just Ultimaker.

[SandervG 1,521]

Ok, allow me to clear up some of the confusion.

Even though there can be some tolerance in the nozzle size, I don't believe this was the original reason why this change in nozzle size surfaced. And I don’t think it is that much either.

The Olsson Block uses different nozzles than the nozzles that were included in the Ultimaker 2 originally. Mainly the internal chamber is different, and it had an affect on the way it cooped with oozing. So we changed the printing profile for the Ultimaker 2+ (which always has these OB-nozzles) to achieve the results we grew accustomed to, which unfortunately lead to a bug for the regular Ultimaker 2 profile. So there was a change for the Ultimaker 2+, and it also affected the regular Ultimaker 2 print profile.

But with the latest Cura download, all is good and it is resolved

For the new version of Cura which Didier had a presentation about, it indeed has some interesting approaches about line width (among others!). More about that, when the new Cura is being launched

So if you download Cura 15.04.5 all should be good again!

Edited March 23, 2016 by Guest

[yellowshark 153]

I did a test today, with hardware set with a 0.4 nozzle. I did three prints using the same software settings, including setting initial layer to 0.3, except the Cura nozzle size was set to 0.35, 0.4 and 0.45; plus for the run with the 0.35 setting Cura warned me that the initial layer should not exceed 0.26 so I changed that setting to 0.26.

On each run I allowed the bed to cool down so I could just lift the piece of the bed.

Placing the three pieces side by side on the table, with layer 1 facing upwards, and giving them a reasonably cursory look they all looked good. I picked up the 0.4 and 0.45 pieces and, with one in each hand, re-angled them against the light source; it then became clear that the 0.4 piece was better than the 0.45 piece. 75%- 80% of the 0.45 piece was comparable with the 0.4 piece.

I then picked up the 0.4 piece and the 0.35piece. Was the 0.4 piece better, hmnn; was the 0.35 piece better, hmnn - I keep changing my mind!!

In the pic the piece on the right is darker only because of its position relative to the light source.

Right now, personally, I am not convinced of the reality of this subject. Within a range of 100 microns, whatever my physical nozzle size actually is, I am not seeing any significant difference.

I am going to test again tomorrow to see if I throw up anything different and also I now realise that the 0.4 and 0.35 comparison are not the same as the initial layer thicknesses were not the same, so I will do the 0.4 setting at 0.26 layer thickness next time.

[Labern 774]

Its not just flat surfaces. Its also less stringing, Better corners and less z scaring. The amount of difference is small but every bit counts.