Weird z-height variance

I've been using my Mendel90 much less regularly recently and have spotted something weird with my first layer height. All of the following observations are from the same .gcode file, so no slicer settings have been changed between runs. The job takes about three hours and I let the bed cool completely to room temperature before removing the job and starting over.

The first print I do seems to have a thicker-than-desired first layer – I can see the infill clearly and there are definite ridges that catch my fingernail when I run it over the surface. As this first layer is clearly not squished down as far as it should be, there is some lifting at the corners and edges, especially near the edges of the bed where it is cooler than the middle.

Subsequent prints gradually improve, and I would say that after about four jobs the first layer is back to its glass-smooth finish. In fact, by job four I think the first layer is now too thin and there is some flaring around the edges that indicate I am squishing it down too much. Print three is definitely the best one in terms of side-wall straightness and first layer smoothness.

The weird thing is that the job takes three hours to complete, and I don’t start the next one until all temps are back to ‘normal’. In fact, if I leave the printer off overnight and resume the following day, the pattern still holds true. The first layer height only reverts to being ‘too thick’ if I leave the printer off for an extended period – a week or more as has often been the case recently.

My initial conclusion was that something was expanding and contracting with the heat but I can’t think of anything that would retain its heat for such a long period of time. I can’t however think of any mechanical explanation for why the first-layer height would increase and decrease in such a way – surely any stepping issue with the motors would gradually compound and cause the trend to continue in the same direction?

Any ideas?

Hi,

There are only two things that I can think of:

1. First time bed cleaning differs from inter-print cleaning
2. Effects on the filament when left in the extruder for a long period

My logic for the first case is that bed adherence might not be as good for the first print and the whole bottom layer might be lifted somewhat giving the impression of a thicker first layer.

In the second case I have noticed that when I have left my printer for a while the initial extrude is less viscous. I do not know whether this is due to the material structure changing over time after it has been heated and then cooled in the extruder. If you perform a number of prints in succession the effect is less pronounced.

As an aside, I have noticed that my initial ooze-free startup characteristic has changed recently. I used to get a small blob followed by a thin line before the fat wipe line was printed. Now I seem to get the small blob followed by nothing and a partial fat wipe line. This seems odd because,like you, I have not changed my start/end sequences.

It appears that odd effects like these do happen and I suspect it is material related rather than mechanical.

Regards,
Neil Darlow

---

I try to write with consideration for all nationalities. Please let me know if something is unclear.
Printing with Mendel90 from fedora 25 using Cura, FreeCAD, MeshLab, OpenSCAD, Skeinforge and Slic3r tools.

Quote
neildarlow
As an aside, I have noticed that my initial ooze-free startup characteristic has changed recently. I used to get a small blob followed by a thin line before the fat wipe line was printed. Now I seem to get the small blob followed by nothing and a partial fat wipe line. This seems odd because,like you, I have not changed my start/end sequences.

Regards,
Neil Darlow

I have started to get the same Neil. I did wonder whether my filament had absorbed some wateras it is about 8 months old but I read that this would result in popping from the extruder and my prints do not have any unwanted holes in the walls.

My other theory is that the hobbed bolt slowly digs deeper into the filament when it is left for a while which may cause more or less to be extruded (I cannot decide which). However this should sort itself out quickly as the hobbed bolt bites into new filament.

I do sometimes get the first layer squished a fraction too much like QuackingPlums and I also do not use my printer every day.

Richard

Ok, this is getting frustrating now. I've loaded some new filament (freshly opened from its vacuum sealed pouch) to run some test prints and it would appear that the OPPOSITE occurs from what you would normally expect.

The following tests were done with the same job: 9 square objects (rounded corners) placed in a 3x3 grid, about 50mm square and 5mm high, hollow interiors with 2mm thick walls. There is 2mm spacing between the objects and the job takes over an hour to complete.

Print #1: Very poor adhesion, ridges appear on the bottom face indicating that the first layer isn't sufficiently squished into the glass.

Print #2: Better adhesion than the first, but I can still feel the ridges if I run my fingernail along the bottom face. The ridges are visible but there is clearly better 'squishing'.

Print #3: Perfect adhesion and glass-like bottom. Very few ridges on the bottom face.

Print #4: TOO MUCH squishing! The objects are now flared at the bases causing them to close the 2mm gap between them. Totally glass-like bottom face, no ridges at all.

It sounds like something is expanding but the glass is still level, there's no bowing upwards of the bed (the level of 'squish' remains constant across the bed) and I can't see how the nozzle could continue to elongate over several hours. I'm also allowing everything to cool for about half an hour between jobs and the display reports that everything is at room temperature before each new job.

If I turn off the printer and repeat this tomorrow then I suspect I will get exactly the same results. Whatever the problem is, it resets overnight.

I'm totally at a loss as to the cause and pulling my hair out in frustration!

Hi,

Ignoring the adhesion aspect of these tests, can you measure the thickness of the surrounding skirt and assess any changes in its width between the print runs?

I am trying to determine whether this is an extrusion variance instead of a mechanical effect. I don't often print multiples or repeat prints so I cannot say I have seen similar effects but it sounds like an intriguing problem.

Regards,
Neil Darlow

---

I try to write with consideration for all nationalities. Please let me know if something is unclear.
Printing with Mendel90 from fedora 25 using Cura, FreeCAD, MeshLab, OpenSCAD, Skeinforge and Slic3r tools.

Quote

The first layer height only reverts to being ‘too thick’ if I leave the printer off for an extended period – a week or more as has often been the case recently.

Quote

If I turn off the printer and repeat this tomorrow then I suspect I will get exactly the same results. Whatever the problem is, it resets overnight.

Which is it?

I get a similar effect with my sturdy that is made from unsealed MDF. If I leave it off for a week the MDF expands due to water absorption and then over the course of a day's printing it will contract again. I don't see how you could get that with dibond though. My dibond machines remain very stable after the initial settling in period, where I suspect the hot end does slowly stretch a bit.

What sort of environment is the printer in?

Is it the same spec as the original kit or have you changed anything?

---

[www.hydraraptor.blogspot.com]

When I first noticed the problem I had stopped using the printer as regularly as I used to, so assumed it was somehow connected to the lengthy period of sitting idle.
Looking back at my first post I suppose it's possible that I just assumed that the period required for the symptoms to 'reset' was a week or more, though it does appear that it at least took longer than a day.

Recent activity (I had to do a whole batch of prints consisting of multiple plates of the same objects for someone which took several days) indicates that the reset interval appears to have reduced. Or maybe was always some arbitrary period around a day or so.

My printer is indeed the dibond kit - one of the early ones and I still haven't got round to making the new style X-ends. It has been sat in my den all the time I've had it which is dry, warm and draught free. I have changed nothing in the structure (why should I? it's perfect! ) and the only thing I've added is the OctoPi with its accompanying webcam.

The thing that confuzzles me most is the duration over which this effect goes from one extreme to the other - we're talking a couple of hours at least, with a full cycle from room temperature to ABS printing temperatures and back, using the same identical gcode file loaded into OctoPrint. I distinctly remember a time when I did suffer the problem described by Richard above, where the first job has a flare at the base but is fine for all subsequent jobs once everything has had a chance to settle but this way round is just plain odd when there's very little chance of the main chassis expanding.

Eric,

Have you changed any of the start gcode script recently? At the AP meet you mentioned that you left the extruder high during bed heating so that it could warm up quicker under a cover. I use the gcode that Chris supplied where the extruder decends to first layer thickness above the bed immediatly the bed heater turns on. The extruder then starts heating when the bed is about 20 deg C below the target. Could you problem be due to the extruder being at different temperatures and therefore changing in length. Worst case would be a faulty thermistor but they normally work or fail.

Richard

My start gcode hasn't changed in years - the initial pre-heat where I keep my extruder high is for the bed only - the extruder isn't on at all during this phase. Once the bed gets to temperature the rest of the gcode is the same; start the E pre-heat and lower to the glass to prevent ooze, wait for E temp to stabilise for a couple of minutes, wipe, start the job. The bed pre-heat phase just lets me keep a cardboard blanket on top of the bed to help it warm up quicker and give me time to remove it before the extruder pre-heat phase, which still includes the time allowed for the nozzle length to stabilise.

The first layer Z variance occurs from job to job, and the length of time I allow for the whole printer to cool down between jobs is sufficient for all the temperature readings to return to 'room' values. In any case, with the exact same job, using the same gcode, I can't see how the amount of expansion in any component can vary - it's not like the various phases are taking differing lengths of time. It's also a definite linear relationship between job number and amount of squish.

EDIT: PS, I've also been working on an automatic bed blanket removal mechanism on Chris' recommendation, though that mini-project has stalled as it's flying season. The blanket method isn't something I dreamt up - it's on this forum somewhere. It would be nice to be able to do the whole thing unattended though.

Edited 1 time(s). Last edit at 07/10/2015 04:58AM by QuackingPlums.

Just a wild thought but could it be that the printer is not homing the Z axis the same height every time? This could happen if the limit switch was loose or worn. You could check whether the pointers on the Z axis lead screws were in the same position at the start of each print.

I've set my Z-pointers to be pointing straight inward at the point where my nozzle is pre-heating above the bed, rather than in the idle position at the top. I figured this is the point where I usually remember to check them, which I do before every print. I guess however that ~200mm travel downward from the limit switch has no bearing on the bed height, if that has somehow moved.

In any case, I've just checked the switch and bar clamps and everything is tight with no play. Could there be enough expansion in the smooth rods to push the switch upwards relative to the bed? I'm thinking not but we are talking about fractions of my layer height (0.2mm) which isn't that much. If this is true however then whatever is causing the expansion is reducing linearly over several jobs.

The switch is the same age as the printer - about three years old. I'd be surprised if it has worn already (in a way that makes it actuate slightly later on each subsequent job?) as these things must be rated in the tens of thousands of actuations. I can swap it out if it'll eliminate a variable but I think this might be a straw-clutching exercise. ;-)