What causes wrong height (Z axis)?

What do you think is the reason for a part to come out shorter that it should? The error is about 0.4-0.5mm.
One cause I can think of is the wrong Z_HOME configuration but that can't cause such big error on its own.

Thanks

Is it a fixed offset or does it vary with height?

Is the target height a multiple of the layer height?

---

[www.hydraraptor.blogspot.com]

Not sure right now, I will have to do more tests. However, either the case, it would be good to know the reasons for wrong Z height so I can check them anyway.

The height of an object will be quantised to the layer height, so it can be up to half the layer height out.

If the Z calibration is not correct, so the first layer is not the target height, then there will be a fixed error by that amount.

If the pitch of the lead screws is not exactly 1mm then you will get an error that increases with height. That can be rectified by adjusting the Z steps per mm in the firmware. You would need to print something tall that is a multiple of the layer height and something short. Then measure them and subtract the measurements to separate the offset and the scale.

---

[www.hydraraptor.blogspot.com]

So Chris, do you believe its possible for the lead screws to have the wrong pitch? Do you believe changing the firmware Z steps per mm is good practice?

On another note, could this also attributed to ABS shrinking? Even though I imagine that shrinkage happens only in XY.

Nothing is ever exact, there is always some tolerance. I find using the belt pitch and lead screw pitch is accurate enough for my needs but you can get better accuracy by calibrating all the axes. The difficulty is measuring the error accurately and separating it from plastic shrinkage.

You haven't indicated the size of the object or whether the error varies with size so it is hard to make any diagnosis.

Objects shrink less in Z because the higher layers tend to compensate for any shrinkage in the lower layers. How much it shrinks vertically will depend on how hot the object is when the last layers are added.

---

[www.hydraraptor.blogspot.com]

Quote
nophead
Objects shrink less in Z because the higher layers tend to compensate for any shrinkage in the lower layers. How much it shrinks vertically will depend on how hot the object is when the last layers are added.

That actually makes a lot of sense.

I designed and printed a calibration piece in openscad. This is the result:

Each 'leg' of the part has a different layer height, of equal increment. First 'leg' (the small one) 0.3mm, second 'leg' 0.6mm and so on. Needles to say, I sliced this with 0.3mm layer height.

I measured and came up with this error graph:


I think it is pretty clear that something is wrong with the first layer. It is WAY too close to the bed for some reason (it was OK a few weeks ago!). After that the graph shows a slight increase in later layer which is consistent. I think I should fix the first layer first and then retest. It is, however, important to notice that in some layers the error is well above 0.3mm which means that not all the error comes from the wrongly calibrated first layer. This means there is another source of error. If that is true, I would expect an increasing error as the layers get higher (as nophead noted).

Finally, I suspect that some of my measurements are a bit misleading, as I noticed that I am slightly overextruding, so the top solid layer is a bit rough. I tried to measure only the perimeter height (as the perimeters are unaffected by the overextrusion) but it is possible that I missed some. I will have to reprint the object with more perimeters and measure again

You can get the first layer off by more than the layer height. If it is negative then the pressure of the plastic can actually lift the nozzle giving a finite first layer. Then when it does the subsequent layers gravity will cause it to be lower.

---

[www.hydraraptor.blogspot.com]

Quote
nophead
You can get the first layer off by more than the layer height.

You are right, this seems to be the case! I have changed with Z_HOME_POS (Z_HEIGHT_MM in neildarlow's updated firmware), essentially raising the head by 0.41mm! This is very odd because I am definitely sure that the head was not crashing on the bed. Any ideas here?

It seems we are on the right track. The new piece has a consistent error of +0.03mm which I will add to the Z_HOME_POS in the next experiment (unfortunately in a week ). Now that the measurements are much clearer, I can see that I also need to level the bed. One other problem that popped up now, is that the first first extrusion on the object (after the perimeter is formed) does not stick well. Based on luck, it can either catch on the head and destroy the print or not. Oh well, I will see about that later but any tips are welcome!

Thanks for the help so far, I will update next week.

Quote
demetris
You are right, this seems to be the case! I have changed with Z_HOME_POS (Z_HEIGHT_MM in neildarlow's updated firmware), essentially raising the head by 0.41mm! This is very odd because I am definitely sure that the head was not crashing on the bed. Any ideas here?

Have you moved the machine? Are the Z screw pointers still in sync?

I have known the hot end to appear to get a little longer over the first few prints, but not by that amount.

Quote


It seems we are on the right track. The new piece has a consistent error of +0.03mm which I will add to the Z_HOME_POS in the next experiment (unfortunately in a week ). Now that the measurements are much clearer, I can see that I also need to level the bed.

The way to get it exact is by measuring the skirt and comparing it with the layer height, as suggested in the manual. Measuring the object is not as accurate.

Quote

One other problem that popped up now, is that the first first extrusion on the object (after the perimeter is formed) does not stick well. Based on luck, it can either catch on the head and destroy the print or not. Oh well, I will see about that later but any tips are welcome!

Thanks for the help so far, I will update next week.

When I print ABS I cover the glass with PET tape and use a bed temperature of 130C. That is actually about 110C on the top surface. ABS won't stick to bare glass unless the first layer is too low.

---

[www.hydraraptor.blogspot.com]

After many tests and inconclusive results, I noticed that the indicators on the threaded rod were out of sync. Then I notices I could move the threaded rod (from the top) freely! Almost zero traction! It makes me wonder how I got my prints so far! I unscrewed the mounts and I saw this

nophead, do you have any ebay link for buyng that material?

PVC tubing can be bought from DIY hardware shops and pet shops. I have never known the PVC version of the Z coupling to slip though.

We switched to neoprene tubing and slightly different printed parts to get better centring. That has less grip though but it shouldn't need much torque to turn the studding. PM me if you want some neoprene.

---

[www.hydraraptor.blogspot.com]

Thanks for the information!

Quote
nophead
I have never known the PVC version of the Z coupling to slip though.

Sorry I should have mentioned before that that is most probably my fault as some of the lubricant I applied on the threaded rods, must have moved down with time. The motor shaft was all greasy. I cleaned it now and tried a temporary fix until I get some tube.

I've checked with a good comparator.
8mm move give 7.95mm measured on Z axis.
I'll will adjust the pitch.