Print postmortem

[reeper 0]

Greetings to all,

I finished assembling my Ultimaker a few days ago and after some reading I have made a few attempts at printing, with mixed results. In spite of my initial research there are still many things I’m uncertain of and would appreciate your assistance if possible in diagnosing some print issues.

For my first print I went with the tried and tested calibration cube. I used repg26 alpha2 on osx to slice(skeinforge35 quality profile) and print, along with the stock 5D firmware.

The result:

http://dl.dropbox.com/u/16630702/cubeSIDE.jpg

http://dl.dropbox.com/u/16630702/cubeTOP.jpg

Seems alright, the lines are visible but fused, some blobs and strings.

I then decided to move to the beta 1 version of marlin(version found here

) and skeinforge 45 following the setup given here

[ddurant 0]

I'd stick to 0.25mm or 0.20mm layer heights until you get a few really good prints done. Thinner is certainly possible but it's easier to work through issues if you start with the easier settings..

Good move, going up to SF40+ and Sprinter/Marlin!

Pretty much all of these pictures look to me like they need more plastic. You should only be able to see the outer-most layer and I can see the layer(s) underneath on most of these pictures.

What's your start.gcode look like? If you're not sure, just post the first ~20 lines of something you sliced recently.

Are you sure you've measured your filament diameter accurately and told SF45 what it is? It's in the Dimension tab. When you measure the filament, make sure to get a variety of readings from different angles and use the average. Also, the Filament Packing Density setting should also (for now) probably be set to 1.0.

FillSolid Surface Layers tells SF how many totally-filled layers to do near outside surfaces. People usually forget about changing this when they change layer height.. Usually, for a layer height over 0.25mm or so, this parameter gets set to 3 which means 0.75mm (= 0.25mm * 3) of solidity on outside surfaces. If you keep this at 3 but change layer height to 0.10mm you're only now getting 0.30mm (= 0.10mm * 3) of solidity, which probably isn't going to give you a clean outside surface, if you're trying to cover up a less-than-solid infill. There's no hard rule for how to set this but I'd move up to 5 if you're at or under 0.10mm layer heights. This is one of those things that just takes practice: keep an eye on it when you see it starting to cover up sparce infill - the first layer or so looking crappy is normal but the outside surface should be nice.

You didn't mention CarvePerimeter Width Over Thickness and FillInfill Width Over Thickness.. The w/t values control how skeinforge slices things on the horizontal. This value times the layer height is the width of the threads in the print. With a 0.40mm nozzle, you probably want threads in the 0.40mm to 0.6mm range, though you can probably get away with as low as (despite common sense) 0.35mm or as high as 0.75mm or so. As you change layer height, make sure to also set these values. At 0.10mm layers, w/t values of 5 will get you 0.50mm thread widths, which is a good place to start. For now, just divide 0.50 by the layer height and set the two w/t values to the result..

 

...and what the cause of my seemingly “shifted” infill layers might be(belt tension issues ?)

Could be too much belt tension but probably not too little, I think.. The machine tried to push the print head in one direction something pushed back so much that the stepper motor lost some steps. The electronics don't know when this happens and just assume that every move happened as-expected.

With the power off (or steppers disabled), make sure the print head moves smoothly in each direction. If it doesn't, take a close look to make sure stuff's lubed and orthoganal. If you've got blobs on a print, that can also cause skips if the print head smacks into one at high speed..

[reeper 0]

Firstly thanks for the reply, it’s much appreciated

 

I'd stick to 0.25mm or 0.20mm layer heights until you get a few really good prints done. Thinner is certainly possible but it's easier to work through issues if you start with the easier settings..

Haven’t tried to print again, but will switch to 0.2 when I do

 

Pretty much all of these pictures look to me like they need more plastic. You should only be able to see the outer-most layer and I can see the layer(s) underneath on most of these pictures.

What's your start.gcode look like? If you're not sure, just post the first ~20 lines of something you sliced recently.

 

At first I used your start gcode here

http://dl.dropbox.com/u/16630702/start.gcode

).

 

Are you sure you've measured your filament diameter accurately and told SF45 what it is? It's in the Dimension tab. When you measure the filament, make sure to get a variety of readings from different angles and use the average. Also, the Filament Packing Density setting should also (for now) probably be set to 1.0.

 

I measured in a few places, got anything between 2.95 and 3.1mm so I decided to go for the lowest value, and see if I got too much plastic from time to time. One could argue that the blobs and strings might be examples of that but as you pointed out, overall, it seems to be lacking plastic. The Filament Packing Density is set to 1.0

 

Fill\Solid Surface Layers tells SF how many totally-filled layers to do near outside surfaces. People usually forget about changing this when they change layer height.. Usually, for a layer height over 0.25mm or so, this parameter gets set to 3 which means 0.75mm (= 0.25mm * 3) of solidity on outside surfaces. If you keep this at 3 but change layer height to 0.10mm you're only now getting 0.30mm (= 0.10mm * 3) of solidity, which probably isn't going to give you a clean outside surface, if you're trying to cover up a less-than-solid infill. There's no hard rule for how to set this but I'd move up to 5 if you're at or under 0.10mm layer heights. This is one of those things that just takes practice: keep an eye on it when you see it starting to cover up sparce infill - the first layer or so looking crappy is normal but the outside surface should be nice.

 

Thanks for the explanation, I hadn’t altered that but I will certainly do so for the next print.

 

You didn't mention Carve\Perimeter Width Over Thickness and Fill\Infill Width Over Thickness.. The w/t values control how skeinforge slices things on the horizontal. This value times the layer height is the width of the threads in the print. With a 0.40mm nozzle, you probably want threads in the 0.40mm to 0.6mm range, though you can probably get away with as low as (despite common sense) 0.35mm or as high as 0.75mm or so. As you change layer height, make sure to also set these values. At 0.10mm layers, w/t values of 5 will get you 0.50mm thread widths, which is a good place to start. For now, just divide 0.50 by the layer height and set the two w/t values to the result..

 

Yeah sorry about that, I had them both set to 4 for all prints(besides the calibration cube which the repg26 SF35 high quality profile handled), so a thread width of 0.40mm for all. I’ll up it to 5 as you suggest for the next print.

 

Could be too much belt tension but probably not too little, I think.. The machine tried to push the print head in one direction something pushed back so much that the stepper motor lost some steps. The electronics don't know when this happens and just assume that every move happened as-expected.

With the power off (or steppers disabled), make sure the print head moves smoothly in each direction. If it doesn't, take a close look to make sure stuff's lubed and orthoganal. If you've got blobs on a print, that can also cause skips if the print head smacks into one at high speed..

Sorry, I was a little bit vague in my comment, I wasn’t talking about the layer shift which I also thought was due to the nozzle hitting a blob.

It’s not easy to see in the photos, but earlier when I was talking about infill intersection with the perimeter, it seemed to me that the infill itself was shifted a bit to the left relative to the perimeter of the whistle, but looking at it again it’s not really a shift, it just misses or intersects too much in places. For an example of “excessive intersection”, in this photo

http://dl.dropbox.com/u/16630702/second ... ECTION.jpg

you can see that on the opposite side of this ridge, the infill doesn’t touch the inner perimeter of the whistle indicating that in that area the infill has come too far down.

Do you have any idea why that might be ?

Thanks for the help.

[reeper 0]

I had a go at another print, not much better but at least the sides appear to be slightly more solid.

To summarise, I used:

Marlin beta 1

skeinforge 45

repg 26 alpha2(osx) with the control panel closed and temperature monitoring off

temperature: 210

flow and feed rates: 150

travel rate: 450

layer height: 0.2

infill perimeter overlap: 0.2

infill solidity: 0.4

w/t for both perimeter and infill: 2.5

solid surface layers: 4

filament packing density: 1.0

filament diameter: 2.95

same start.gcode as above(first layer didn’t start well after I removed the plastic collecting on the nozzle just before it started printing)

The result:

http://dl.dropbox.com/u/16630702/2012-0 ... .54.23.jpg

http://dl.dropbox.com/u/16630702/2012-0 ... .54.54.jpg

http://dl.dropbox.com/u/16630702/2012-0 ... .55.26.jpg

http://dl.dropbox.com/u/16630702/2012-0 ... .56.00.jpg

Lots of stringing and blobbing, there are places where the head seems to stop for a longer time that usual(at certain layers near the mouth of the whistle for example) resulting in the blobbing effect. The threads that form the loop of the whistle never seem to settle properly so that remains a mess. Still a lack of plastic being extruded it seems, not sure what the issue is, I haven’t seen any evidence of the bolt slipping on the filament.

Worst of all, the whistle doesn’t work :(

[ddurant 0]

I measured in a few places, got anything between 2.95 and 3.1mm so I decided to go for the lowest value, and see if I got too much plastic from time to time. One could argue that the blobs and strings might be examples of that but as you pointed out, overall, it seems to be lacking plastic. The Filament Packing Density is set to 1.0

Is that filament from Ultimaker? That sounds bigger than what most people have - on my old MBI machine, 3.1mm would definitely get stuck and jam things up. It's also got a lot of variation in it - mine consistantly measures 2.82mm to 2.86mm. 2.85mm or so seems to be the 'standard' when people talk about '3mm' filament.

Do you have different filament you can try, just to see if it behaves differently?

For the "excessive intersection" I'm not exactly sure what you mean..

If the infill is consistantly getting too close to the perimeter, lower Fill\Infill Perimeter Overlap. Lower values cause less overlap, higher values cause more overlap. I think most people use about 0.15 on this but feel free to set it to 0.0 for now then start working it up as your prints improve.

If the infill seems off with just one side of X and/or Y, redo the tension on the short belts on the motors. On earlier machines, the bolts were a little too long and didn't grip the wood very well - a couple washers on each bolt solved that for me (and have more surface area so might be a good idea anyway).

 

flow and feed rates: 150

travel rate: 450

Starting at 50/50 with 100 travel might be a better way to start.. Start with slow & easy settings until things look good then put the spurs to 'er.

Don't worry about strings for now - just make sure Comb is enabled and, for now, ignore them. Blobs should be first priority.

[reeper 0]

Yep, the filament is from Ultimaker, I took some more measurements and admittedly I didn’t find anything much over 3mm this time, so I guess the range is more 2.9 - 3.0 mm.

To give you a better idea of what I meant with regards to infill intersection with the perimeter, I’ve pointed out the ridges that appeared in the whistle perimeter surface here

http://dl.dropbox.com/u/16630702/1.jpg

and the possible cause by showing the gap between the two(inner perimeter), and where the ridges might be being caused due to the infill having shifted downwards and intersecting more with the perimeter below

http://dl.dropbox.com/u/16630702/2.jpg

 

[ddurant 0]

I wouldn't worry about that until you get that infill right.. On layers that should be solid you can see the layer underneath, which is something you shouldn't see.

Some people here are running different filament drives and electronics and stuff.. Is this a whole kit from Ultimaker or do you have other parts in there?

Which bolt from

http://wiki.ultimaker.com/Ultimaker_rev ... eparations

do you have installed?

[reeper 0]

Standard kit, received on the 22 december with a new bolt, and so my start gcode has been set accordingly. Perhaps the version of marlin I got from the tags section of the github site should be replaced with the current state of the beta branch

https://github.com/ErikZalm/Marlin/tree ... /Marlin_v1

?(although I guess it isn’t a firmware issue)

[ddurant 0]

Yeah, I wouldn't suspect firmware yet..

Does it look like your start.gcode is really showing up correctly at the top of generated gcode files?

[reeper 0]

The start code is definitely there :/

I do have a copy of netfabb, just haven’t set it up yet. I could give that a try if it would be of any use in diagnosing the issue.

[ddurant 0]

The start code is definitely there :/

Hm.. The only thing I can come up with is that your filament size isn't set correctly..

If you're really sure it is correct, changing the Filament Packing Density is probably the thing to do. Lower values get you more plastic and higher values get you less plastic.. Normally, I'd suggest changing it at 0.025 intervals but yours seems pretty far off so give 0.90 a try and see how that works.

edit: are you sure this is PLA plastic? Does it smell like maple syrup or like hot plastic? If it smells like hot plastic, it's ABS..

[reeper 0]

It’s labelled as PLA, smells somewhat sweet. I suppose I’ll have to play with the density then. Thanks again for your help

[reeper 0]

Ok, to save a bit of time I went back to the calibration box instead of the whistle.

Everything is the same as before except I dropped the temperature to 200, the flow/feed rates to 50, travel rate to 100 and the filament packing density to 0.9.

I forgot to mention that I dropped extra shells to 0 as suggested by the maker of the whistle on thingiverse, but only before attempting to print the solid whistle. It didn’t appear to make much of a difference(not sure what purpose shells serve).

Here’s the box:

http://dl.dropbox.com/u/16630702/box1.jpg

http://dl.dropbox.com/u/16630702/box2.jpg

http://dl.dropbox.com/u/16630702/box3.jpg

The threads meet but are still distinguishable. The result is somewhat better than the first box I printed but not ideal :/

[ddurant 0]

That looks far better to me - you can't see the layers underneath any more!

What settings changed between the two top-view pictures?

 

I forgot to mention that I dropped extra shells to 0 as suggested by the maker of the whistle on thingiverse, but only before attempting to print the solid whistle. It didn’t appear to make much of a difference(not sure what purpose shells serve).

Extra Shells are extra laps around the perimeter.. If you look at your top-view pictures, you can see one lap around the outside of the object then zig-zag infill on the interior. If you turned extra shells back up, there would be multiple laps around the outside.. You might find

http://davedurant.wordpress.com/2010/10 ... sic-terms/

useful, though it's a bit dated..

Having a look through

http://techwall.net/analyzing-your-first-print

is also worth doing - Florian has written up some great articles, though they're mostly netfabb-based.

[reeper 0]

Thanks for the clarification, I had seen both links but I guess the definition of shell hadn’t stuck with me. I’ll have another look at florian’s link to see if I’ve missed anything and at yours for further tips on skeinforge configuration(I haven’t read all the articles in the index yet)

Those pictures are all of the same cube, top, bottom and one side. Is it not possible to obtain a smooth surface at this w/t thickness(0.5mm) ?

[ddurant 0]

Is it not possible to obtain a smooth surface at this w/t thickness(0.5mm) ?

It really depends on your definition of smooth.. Once you get it printing well at easy settings, you can decrease your w/t so that it prints thinner threads. I think down to about 0.35mm should work.

Pictures at a 45 degree angle would be better since ones taken head-on don't show depth. That said, things still look a little thin to me - maybe try dropping down the filament packing density a bit more to 0.85 and see how that looks. If it's better but still thin, drop it a hair more. If it starts to get blobby, turn it up a hair.

[rooiejoris 5]

just my thoughts: If it extrudes 'half' of the filament, maybe the jumper settings on the extruder stepper driver are not set correctly....

Or is the filament skipping/slipping, you can feel that by hand.

[bradley 1]

I think you're trying to accomplish too much too fast.

You need to use a set of fail-safe settings and see what that gets you on a 20mm box print. Until you can obtain a good box, you shouldn't try anything more ambitious.

Fail safe settings (as well as solutions to everything you're experiencing) are documented on the wiki (do the research yourself). You should probably be on standard firmware and not Marlin just yet. Downgrading is easy.

Once you have a good 20mm box print on standard firmware, using RepG's standard, failsafe settings, then you can experiment with other things -- newer SF, netfabb, marlin, etc.

Otherwise, you'll end up with lots of frustration, failing prints and a sour attitude towards UM.

Best,

*B

[ddurant 0]

... You should probably be on standard firmware and not Marlin just yet. ...

I disagree - everybody should move off the standard stuff and up to SF40+!!

[reeper 0]

I haven’t held the filament to gauge the movement, but it did appear to be moving without slipping and the “teeth” marks on the filament as it travels up are always regular when I have a look.

I may have found part of the problem(not entirely sure). At first I noticed some plastic leaking between the nozzle and the aluminum block, but I pegged it down to printing at higher temperatures and considered it normal based on the photos I have seen.

I have only just recently noticed that It’s not as apparent from the front, but looking from the right(in a cooled state) there is a substantial amount of leaked plastic(my last print was at 200 degrees celsius). Not sure how I didn’t notice this sooner, but it could account for the lack of plastic and random blobbing.

I suppose I’ll have to heat the nozzle, remove it, clean it up and try to make sure it’s fastened more tightly.

Front:

http://dl.dropbox.com/u/16630702/nozzle1.jpg

Right:

http://dl.dropbox.com/u/16630702/nozzle2.jpg

 

[ddurant 0]

Not sure how I didn’t notice this sooner..

If it's any consolation, I had to stare at the picture for a minute to spot it.

If you take it all apart, you should get some teflon tape in there when you re-assemble, like in

http://wiki.ultimaker.com/Ultimaker_rev ... usion_head

.

Also, be VERY gentle when you remove the nozzle - several people have reported breaking to tube that it threads onto..

[bradley 1]

... You should probably be on standard firmware and not Marlin just yet. ...

I disagree - everybody should move off the standard stuff and up to SF40+!!

I disagree back. This machine needs a fair amount of tuning mechanically before you begin to get the benefits of improved software. If you don't have it mechanically right, there is no reason to change the software. It only complicates things.

I'm still re-tuning my machine with every modification I make to it (the latest, a new style fan shroud, which changed virtually every parameter). Very frustrating but I'm back to great prints again.

If you can't get Makerbot-quality prints with the unit's standard settings, you'll never achieve Ultimaker excellence prints.

[ddurant 0]

I'm more disagreable than you are, Bradley!!

With SF-pre40, you've got to screw around with matching feed/flow rates every time you change layer height or thread width.. If your print's too thin, there's really no easy way to tell what the problem is - too much feed? too little flow? something mechanical? who knows! It's just too easy to go off into the weeds and get stuck wandering around in them..

With SF40+, you at least get some assurance that if you set feed and flow to the same value and measure the filament correctly, it's going to print something close to correct. In this particular thread, I'm a bit surprised (and a bit suspicious) that a leaky printhead could cause things to be off as much as he's seeing - it will be interesting to see what happens when he gets the leak sorted and starts printing again.

Volumetric 5D was the biggest advance in home printing for 2011, IMO. Backing away from it is not the way forward!

[bradley 1]

I'm more disagreable than you are, Bradley!!

Volumetric 5D was the biggest advance in home printing for 2011, IMO. Backing away from it is not the way forward!

I agree, but you have to have a sound machine first. It doesn't sound to me like this guy does. That's why I'm suggesting simplifying things.

Going from standard firmware to the latest marlin with all its jumping around is bound to shake all sorts of complexities into a new build. The stodgy old firmware and thick filament threads will allow someone to tune all the basic issues. Level layers, square corners, uniform threads, a solid first layer, etc. Alignment, belt tension, lubrication, stepper driver voltage, extruder temperature, filament tension, extrusion cooling -- all come into play with basic "fail-safe" software and settings.

I'm sure this guy hasn't yet got those perfect. It's not going to help stepping up the speed and lowering the layer height. That'll only lead to what we've seen from other users-- bad-mouthing this system.

Everytime something gets out of whack on my machine, I always return to basic prints. Once those are working right, I step back up to more challenging things. I'm still dealing off and on with difficult backpressure issues.

[tlalexander 0]

Hey Guys,

Jumping in here after seeing the e-mail about this.

A couple things: for what its worth, I told him to jump ahead to Marlin and SF40+. Thats what you guys all told me! :-p

For me, my machine seemed to work "well enough" and I was excited to just get printing without all the weirdness from 5D hesitating at every line. I've never actually tuned anything - I pretty much printed two cubes and a bottle opener on the old stuff then immediately jumped to all the new beta stuff. However aside from being very aware that my belts are loose, I've had plenty of prints that are useful to me.

Anyway, I might not be as picky on print quality. Which leaves me wondering - Richard, some of your prints don't even look bad to me. But then, its a little hard to tell - it looks like most of the pictures are dead straight on, which makes it hard to really see whats going on. Can you take some pictures with more of an isometric view? Also, the drop box links are kind of a pain if I want to go back and forth for between a few pictures, to get the "whole story" so to speak. Personally if you took some isometric pictures and put all your pictures on an online gallery like picasa or flickr, it would be easier for me to really understand what is actually wrong with the prints.

Also, I never found a picture that showed what you described (didn't look at them all though), but you said the infill wasn't hitting the perimeter at times. Have you seen this page?

http://wiki.ultimaker.com/Troubleshooting#When_a_layer_is_printed.2C_the_perimiters_and_infill_of_the_object_do_not_or_barely_touch._How_can_I_fix_this.3F

I don't like this forum thing so I may forget to come back, but chat me if I miss something.

Oh, I was going to say that trying Netfabb might be a good idea - use the low quality profile and just see if the problems are still there, or which problems are gone and which persist. If any of the problems go away, you know they are from Skeinforge and/or the Skeinforge settings. Changing one part of the system and seeing how it behaves is always a good way to troubleshoot. Although depending on how Netfabb does the topskins, it might not even be able to make a good whistle. But thats another issue and it will be clear if thats the problem. (I've never tried printing a whistle so I'm just guessing).

-Taylor