First layer not sticking to bed

[dmpalmer]

New H1.1, I have sometimes managed to get a Yoda head, but I am having a lot of trouble getting the first layer to stick.

I am using PLA (nozzle at 200 C) on an unheated bed with 3 layers of blue tape (3M #2090). Repetier Firmware for Rambo using Repetier Host Mac with Slic3r. First layer is 200 microns and I have tried speeds as slow as 5 mm/s and as fast as 60 mm/s. I have tried with the head scraping across the bed and obviously too high as well as the right height. I have tried doubling the extrusion flow rate.

Nothing sticks.

Any suggestions? Should I be running my nozzle hotter/colder? Moving faster/slower? Something other than 3 layers of blue tape? Laying down a raft (how do you do that in Repetier?)?

I bought the heated bed, but haven't installed it yet. Should I do that instead of trying to get it to work on the unheated bed? (What is the installation? Do I put a layer of cork down on the plastic bed, then the heater, then tempered glass and clamp it all together with bulldog clips? Or do I need an air gap, mounting bolts, springs, backing board?)

[Fiero2M6]

PLA should stick to the blue tape. After laying down the tape (sticky side down!), try cleaning it with rubbing alcohol. This helped my bed adhesion a lot.

[dmpalmer]

Thank you. The rubbing alcohol was a huge improvement. Most of my test object (the nautilus gear) stuck to the bed this time. Full fresh tape, then wiping with alcohol, will probably make it even better.

Semi-successful nautilus gear print

Apart from the partial non-sticking, the biggest problem is a lack of solidity ( i wanted 100% infill but the strands inside the gear teeth have gaps between them, and the weight of the piece is less than 3 grams, whereas based on slic3r it should have 3.1 cm^3 -> 3.9 grams of plastic.) There is blobbing and threading, and the extruder has a tendency to stop feeding partway through a run. But now that I have adhesion, I can work on those other problems.
(Any suggestions?)

Extruder jams at Yoda's eyes and Willendorf's breasts

[Fiero2M6]

Those nautilus gears are tricky. I had to slow everything down on first layer to get it to stick.

It sounds like Slic3r may not be extruding enough filament. This could be because the steps/mm is not calibrated correctly or because Slic3r is not calibrated correctly.

If the extruder stops feeding, it could be jamming, slipping, or the PLA filament could have absorbed water.

Than green looks awesome. I'm going to have to get some.

[dmpalmer]

The specific filament is
1kg Green 1.75mm PLA Filament for 3D Printers, Makerbot RepRap UP sold by onlinefilament
but you may want to wait until I figure out whether the filament is the problem.

When my extruder is working (and heated up), I can easily push the filament through the pinch rollers turning the gears manually with very little force, and out squirts a stream of plastic from the nozzle.

When the extruder is not working (but still heated up) I can't force the filament into the nozzle. I can pull it back a few centimeters easily, then push it in easily until it gets to its original position and bottoms out.

There are two states, so it's not as if sometimes it is slightly easier and the roller just barely work, other times just barely failing.

This suggests that something is blocking the nozzle: maybe debris of some sort, or maybe unmelted plastic.

I've read about contaminants in cheap filament. However, I also don't know how well-calibrated my nozzle's temperature readout is, or how close the system is to freeze-up. (The thermistor is near the resistors. The nozzle is sticking out into the air. If it looses enough heat, might the plastic in the nozzle freeze?)

When things are working, I have found that response to the feed is more sluggish than I would expect. I hit the extrude button e.g. for 5 mm, the gears turn a few second, and a second or two after the start the string comes out the nozzle and continues for a while after the gears have stopped. Is that normal?

[Fiero2M6]

Maybe your extruder's teflon tube is intermittently jamming. Do you have a fan blowing on the hot end? I cannot consistently print PLA without a fan.

I've never heard of plastic freezing in the nozzle. I don't think that would be an issue.

My nozzle will ooze plastic when it is not extruding. Sometimes the extruder has to run for a little bit before I get a steady stream. This is why Slic3r can create skirts before a print.

If you're concerned that it is stopped up with some contaminent, then you can take the nozzle off, burn the PLA out with a blowtorch, and rinse out the nozzle.

[dmpalmer]

I added a a fan to the hot end, and I still keep getting jams. When it jams, I can just pull out the fiber, snip off the end, feed the end back in, and there's no more jam.

I got some ABS and it worked the first time. (Not very good surface texture and I was fiddling with the temperature along the way, but my neolithic porn now has the sexy sexy head of a hot brainy temptress who can build fire and chip stone.)

[dmpalmer]

Printing at 250 C, 200 microns, 5 shells (vs. 220 C, 400 microns, 3 shells; both cases had 30 mm/s perimeter speed) made a much nicer surface texture. However the ABS filament jammed.

In this case, the jam was due to uneven filament. Just above the jam, the filament was 1.82 microns, which was too thick for the pinch rollers as I had them set. 30 cm further on, the thickness is 1.62 microns. Is this an appaling incosistency, or it is about what you would expect?

(Edit: The filament was white 1.75 mm (nominal) ABS from 3dprinterstuff.com via ebay.)

Also, Slic3r/Repetier toolchain predicted 10 minutes for the build. It was running about 10x slower than that, and the truncated Venus was about 90 minutes in. Am I doing something wrong (e.g. my config.h in repetier rambo) or is the toolchain just that inaccurate when it comes to estimating times?

[Polygonhell]

Cheap filament is often inconsistent in width, the repraper.com stuff IME is usually at least inconsistent and smaller than the specified diameter.
The filament I have from Ultimachine by comparison is very consistent.
Can you print a calibration piece or two, it looks to me that there is much more going on there than just bad filament. The two most important calibration pieces are the single walled test piece and the solid cube, you really need to be able to print those with fairly good dimensions before anything else is really worth the effort.

[dmpalmer]

Thin-wall, 20% infill 1 shell, 100% infill 1 shell

Here are the thin wall, 20% infill and 100% infill calibration pieces, all at 200 micron spacing. Dropping the 100% piece in water, it is about 90% dense. (It floats about 10% above the water.)

For the thin-wall, the Slic3r did sometign strange, and sometimes it zig-zagged when it drew the wall. I could see it slow down at that point in the wall, hear it change tone, and it shows up in the Repetier g-code visuallizer, but not in the STL visualizer.

zig-zag thinwall

[Polygonhell]

Which version of Slic3r are you using (many versions have significant bugs) and what speed settings I would suggest finding a copy of 0.72 it was the most stable and still has the best surface finish.
How do the parts measure?
Let's concentrate on the single wall test piece first, if that's correct, then the rest should be easy.
It's hard to tell from the pictures, but it looks like you are not getting consistent extrusion, I'd try increasing the tension on the pinch rollers, slow down the print, you can usually feel the filament stop feeding if you rest your finger on the point it enters the extruder.

On the 100% object, the easiest thing to do to see what's happening is abort the print half way through, and look at the infill.

[dmpalmer]

With slic3r 0.9.5 (the latest: I couldn't get 0.72b to work due to the difference in available options in my config files) I tried the 100% fill again and stopped it in the middle. Here's what it looks like using a loupe in front of my iPad camera:

I couldn't get Slic3r to not make wiggles in the walls of the 0.5 mm wall calibration object. However, when I have it do the 6543 cube (Thingiverse #1543 by Da3v) the walls are straight and smooth in the g-code rendering. Printing it slowed down to 10 mm/s (vs 30 before) at 250 C in PLA, I get nice smooth walls except at the origin corner (where it starts each layer height). However, there is some layer separation, probably due to the curvature of the base layer I see from my unheated bed.

Slic3r predicted 233 mm, so I measured that off and marked it, and the run pulled down to about 3 mm before the mark (i.e. 230 mm), so that feed rate is better than 98% accurate so the rollers don't seem to be slipping for this run. (When it hit the thick part of the filament previously the feed stopped turning and the motor rattled, but backing off the spacing between the rollers seems to work.)



Exterior dimensions are good (19.5 x 19.4 mm). Wall thicknesses are 0.47 0.56, 0.60, 0.64 mm (with +/- 0.03 variation depending on how I measure it at different places on each wall.) Nominal wall thicknesses are supposed to be 0.3, 0.4, 0.5, 0.6 mm. I don't know how good the thickness tolerance is supposed to be, and of course the variation in the filament diameter will make it hard to get good absolute thicknesses.

[Polygonhell]

at 250 C in PLA

That's WAY to hot for PLA if you're getting a reasonable temperature reading. Are you sure it's PLA, it's relatively uncommon for PLA to delaminate like that, it's relatively common for ABS.
Depending on what sort of PLA it is, it should be in the 180-200 range or possibly 200-220 range.
The single wall test looks like a good start, make sure you are printing on blue tape if you don't have a heated bed, you'll probably also have to print the first layer a little lower than ideal to get good adhesion.
Exterior dimensions being within 0.5mm is about what I'd expect.
With Slic3r the walls should all be the same thickness, not .6 .5 .4 .3, the number it should be is the single wall extrusion width noted at the top of the GCode file, not the thickness of the walls in the model.
Rather than using the 0.5mm calibration piece use the solid cube, but set infill and solid layers to 0 to get a single wall piece. The extrusion width should match the value noted at the top of the generated GCode. Adjust the flowrate in Slic3r to get it close.

The 100% cube looks like you are getting inconsistent extrusion, for now, slow everything down (including infill) to whatever rate the single wall piece was printed at. Once you can get decent quality pieces printed, you can start increasing the speeds, and fix whatever mechanical issues come up.

[dmpalmer]

Sorry, the white stuff is ABS. (The green is PLA)

[dmpalmer]

Just to close up the loop for people who come to this thread, I no longer have problems with the first layer sticking with PLA. The most important detail is getting the bed height right and the Z axis parallel to it. If it is 200 microns high or low then it fails to attach. I measure at both the left and right sides of the bed, and if necessary pop the timing belt off of one of the vertical Z axis and rotate it by a precisely calculated amount. Unfortunately there is no adjustment on the H1.1 to level the bed fore-aft, so I have managed to live with that so far.

I am still using blue tape (three layers of 3M #2090) as the bed. If I wipe it down with isopropy alcohol then it sticks so well it is hard to remove, but I don't have to do that anymore.

Slic3r often fails, giving slices that don't look like the STL. I am having better prints with KISSlicer and Cura. Cura doesn't know how to talk to the Repetier firmware I have burned onto my Rambo, so the workflow is something like:
openSCAD->(stl)->KISSlicer->(gcode)->Repetier