infill not touching edges / surfaces not smooth

[alaris2 1]

I've been playing with settings in Cura and can now print at 70mm/s 195C. however I've noticed for a while that the infill (20%) does not always touch the outer walls of my print - sometimes to the extent where the infill and outer wall are separate (almost 1mm gap)

I thought this might be a problem with the XY alignment or tension so printed some circles and measured them. the diameter varies by less than 0.5mm all round the circle, so presumably this is good.

the other (perhaps related?) problem is that top surfaces of objects never look very good when printed. they don't look solid - sometimes missing small bits or you can see the random zigzag infill through them. this isn't helped by the outer edges not joinging with the infill of course.

what sort of quality of top surface should I be expecting - it should be fairly smooth solid and flat shouldn't it?

I wondered if this was extruder related - but if I extrude 100mm of filament with the Cura setup, I get 100mm exactly - does that exclude extruder problems, or could I have missed something?

this is Cura RC1, marlin 1.0.0.0

(I'll try and get a photo posted if this isn't clear enough?)

[alaris2 1]

same problem on Cura RC2 (so I don't think the problem is software related).

perhaps there's a way I can improve the quality of the top surface by extruding more material? fiddling with the filament diameter, E or the packing density perhaps?

[destroyer2012 0]

Sounds like a problem with loose belts/backlash. Have you isolated this problem to cura? Does the same thing happen with slic3r?

If you print slower does the same thing happen?

Ideally if your extruder is perfectly calibrated your top surface should be perfectly flat with very few holes (only when the space to be filled is smaller than the width of an extrusion).

Are the holes you are seeing also present in the gcode? For example using PrintRun's gcode visualization? (YOu can set the width of lines on the display to be the same as your actual measured extrusion width)

[alaris2 1]

what's the best way to determine if it's loose belts/backlash? I don't think cura is to blame, although as an experiment I tried some different settings to see if I could improve matters.

(I'll try printing slower tomorrow, right now I'm having a different problem which is that the edges of prints are curling off the bed during printing which makes it hard to complete prints without them getting kicked off the bed)

http://i177.photobucket.com/albums/w228 ... r/keyg.jpg

shows what printrun reckons the top surface looks like. pretty solid really, no problems there.

but

http://i177.photobucket.com/albums/w228 ... 100577.jpg

shows what happens.

the top print is with the settings I mentioned in the first post - the walls separate from the infill on the right bottom edge of that photo and you can see almost 1mm gap present between them. however the circles are perfectly circular. oddly the wall separating from the infill seems to happen predominantly on the X axis, but only ever on the 'falling edge' so to speak. ie. as you scan from left to right, it's wall, infill, wall. and the first wall touches the infill, but the last wall doesn't.

the left keyhole is printed with similar settings - layer height 0.2mm, walls 0.8mm, top thickness 0.8mm, fill 20%, speed 70mm/s, temp 190C, packing density 1.0

the right keyhole is a significant improvement - but note the 'hatching' effect of the top layer is still very visible and there are gaps towards the bottom right of the print.

settings for this were height 0.1mm, walls 0.4mm, thickness 0.8mm, fill 20%, speed 70mm/s, temp 190, packing density 1.2

I'm printing with PLA, but turning the packing density up and wall thickness down seems to help considerably. (layer height makes the edges look better but shouldn't affect the top surface)

I've also tried temperatures as low as 180C but the results were visually identical.

[Daid 306]

You're definitely under-extruding. You can see that the 2 perimeter lines also don't touch, and those should overlap slightly. Most likely the filament is slightly slipping at 190C 70mm/s. You could try upping the temperature a bit, check your steps-per-e and the filament diameter.

[alaris2 1]

thanks all,

filament diameter - check.

temperature - switched back to 200C (altho white PLA seems to burn slightly at this temperature)

can I check the steps-per-E procedure - am I doing this right?

1) release catch on the extruder,

2) heat up hot-end, pull filament back out until it reaches the extruder

3) pop off the bowden tube at the extruder end

4) line the filament up with the top of the quick release fitting and apply the extruder catch again

5) use Cura to extrude 100mm

6) measure with steel rule, enter actual extrusion and lo and behold the E is calculated.

is that right? whilst it checks the extruder is working properly it says nothing about how much filament can be pushed through the hot end.. I agree with your under-extruding statement which was why I tried changing the packing density to see if I could fool it into pushing a bit more filament out.

[destroyer2012 0]

That method is way too complicated. Here is how I measure steps per e:

1) Draw a line on the filament before it goes into the hobbed bolt (some two inches or so from the bottom of the plywood extruder body).

2) Measure the distance from the bottom of the extruder body to the line you've drawn using vernier calipers, and record the result.

3) Extrude 10 mm of filament

4) Measure the distance to the line and record.

5) Repeat steps 3 and 4, noting down the new distance each time

6) Average the distance the filament moved each time. Depending on what E steps/mm was set to before, calculate the number of steps the extruder moved each time. From these values calculate steps/mm

Use your vernier calipers to measure the diameter of the filament to a thousandth of an inch and report this to your slicer.

These are the only calibration steps you should ever need to do. Leave the packing density or width/thickness as default. THe values that are there ensure good layer adhesion and nice flat top surfaces.

[alaris2 1]

so are you saying you've heated the hot end to 190C or whatever first and then executed the gcode to extrude 10mm?

I'll give that a try, thanks!

[destroyer2012 0]

Yes I do it with hot end hot. For me that's 230 C with silver PLA. I use pronterface which has an option to extrude x mm in the GUI, so I don't need to write any gcode.

[alaris2 1]

hot diggity, that made a big difference.

OK, using the E-calibration procedure from Cura gave me E=844 or so. I thought this was right on the grounds lots of other people mention numbers similar.

but using the method described by destroyer only extruded 5mm or so, so I changed E to 1087 and got a much better print straight off.

I'll try and fine tune it a bit more now, but 1087 seems a big number - what kind of 'E' values do other people have?

[kaz911 0]

hot diggity, that made a big difference.

OK, using the E-calibration procedure from Cura gave me E=844 or so. I thought this was right on the grounds lots of other people mention numbers similar.

but using the method described by destroyer only extruded 5mm or so, so I changed E to 1087 and got a much better print straight off.

I'll try and fine tune it a bit more now, but 1087 seems a big number - what kind of 'E' values do other people have?

You will usually get to high E numbers if your extruder tension is to high. Back of a bit on the tension and measure again. (The stepper needs to make 1087 steps to vs. normally 865) - do the initial calibration without the bowden tube connected to the material extruder.

if extrusion temperature is too low - you might also compensate with more E steps - but that can lead to blocked print heads. If the head does not melt quick enough - then you will have increased tension on the material and that leads to extruder skipping - and then needs more e steps needed.

[alaris2 1]

you mean the thumbwheel on the extruder is too tight? I thought I had that set right..

with the bowden tube popped off the back the calibration (see my earlier post) gives E=844, but once I try to push filament through the hot end (as per destroyers post) I need (at least) E=1087

I've been watching the motor to see if it skips but haven't seen anything yet - wouldn't it make a horrible noise and mess of the cog?

[destroyer2012 0]

It's probably because the hobbed bolt is SCRAPING the filament instead of digging in and pushing. This is where the "slip" is coming from. Depending on how tight your tension screw (thumb wheel? Whatever) is, adjust it out (if it's super tight) or in (if it's sort of loose) and then measure again. Take a picture of your filament coming out of the filament feeder, seeing how deep the indentations are would help us judge how tight that screw is (whether its too tight, whatever).

[alaris2 1]

good call kaz & destroyer - I managed to get a decent photo at

http://i177.photobucket.com/albums/w228 ... 100580.jpg

I don't think I can tighten it any further, not by hand anyway so I'll see what happens if I loosen it slightly.