microstepping / layer height relationship

I recently made the switch from .3 to .2mm layer height on my mg prusa. The difference has been HUGE.

What I mean by that was - I thought I had a BIG z wobble issue but it turned out to be some weird artifacting resulting from 1 step not always equalling a z movement. I'm not 100% sure, but I think my ramps z axis is set up for 1/16 step. There must be a mathematical way to figure out what the bad layer heights are - does anyone know what it is?

You want your layer height divisible by the steps per mm as evenly as possible.

---

FFF Settings Calculator	Gcode post processors	Geometric Object Deposition Tool Blog
Tantillus.org	Mini Printable Lathe	How NOT to install a Pololu driver

I don't understand why it's necessary to do this when at 1/16 steps using a M8-1.25 threaded rod, a step amounts to only 0.00039 mm. Surely, an error this small in the absolute positioning of the z height is insignificant compared to the 0.3 or 0.2 mm layer thickness.

There is this tool by L4nce0

Punch in your nozzle diameter, Z steps per mm, and a "test increment" and it will calculate which layers line up best with your hardware. From seeing Nophead's comments in this thread, it seems the conclusions from that applet are somewhat controversial, and with the quality of prints he produces I'm likely to believe him!

That said, the layer height I use the most, 0.2mm, is close enough to one of the sweet spots in the applet (0.199mm), so, maybe I stumbled into an ideal situation? On the other side, though, my prints still exhibit layer alignment issues, which gets more noticeable as the print complexity goes up but remains geometrically simple. For example, my test cubes are fairly wobble free (but not completely), but printing Greg's Extruder shows "jumps" in X+Y positioning that correspond to major changes in the geometry as it goes upwards a layer at a time.

I'm not saying that I understand it, just that I observed it. I'll post some pics when I get home, but the .3mm print looks like it has almost a mm of z wobble in a consistent, repetitive shape rippling up the print.

All else being the same, the .2mm print doesn't have that. To me that means that it has something to do with the layer height. This is a complex beast so it could be many other things, but for today correlation is almost as good as causation.

brnrd Wrote:
-------------------------------------------------------
> I don't understand why it's necessary to do this
> when at 1/16 steps using a M8-1.25 threaded rod, a
> step amounts to only 0.00039 mm. Surely, an error
> this small in the absolute positioning of the z
> height is insignificant compared to the 0.3 or 0.2
> mm layer thickness.

If I am figuring this correctly 1 step in my z access is 0.078125mm

In your example I think the pitch of the rod would be have to be 0.00624 not 1.25.

I am not 100% sure of this but about 90% sure. Can someone else do the math to see.

A step is 1.8 deg. This gives 200 steps in a complete revolution for a full stepping controller. For a 1/16 microstepping controller, there are 3200 steps per revolution.

So, divide 1.25 mm by 3200 steps to get 0.00039 mm per step.

I read somewhere that with Z lift, these small errors accumulate and will eventually mess up the Z movements.

Would it make sense to use 1.5/3200 * 400 = 0.1875mm as layerheight on a 0,35mm hotend? instead of 0.2mm, just as an example.
My leadscrews pitch is 1.5


Jonas

jonasl Wrote:
-------------------------------------------------------
> I read somewhere that with Z lift, these small
> errors accumulate and will eventually mess up the
> Z movements.

It depends on how slic3r handles z lift. If z movements is done in absolute coordinates, then it shouldn't make a difference. I never use z lift since it introduces errors if you have backlash in the z axis.

>
> Would it make sense to use 1.5/3200 * 400 =
> 0.1875mm as layerheight on a 0,35mm hotend?
> instead of 0.2mm, just as an example.
> My leadscrews pitch is 1.5
>

I see that slicer generates g-codes with three decimal places for the x, y and z moves. So, even if it makes a difference, I don't think this would accomplish your goal of making the layer thickness a multiple of the z step. But, what is the evidence that it really makes a difference to make the layer thickness a multiple of the z step when the step size is so insignificant?

brnrd Wrote:
-------------------------------------------------------
> I see that slicer generates g-codes with three
> decimal places for the x, y and z moves. So, even
> if it makes a difference, I don't think this would
> accomplish your goal of making the layer thickness
> a multiple of the z step. But, what is the
> evidence that it really makes a difference to make
> the layer thickness a multiple of the z step when
> the step size is so insignificant?

It doesn't make any difference. Someone just has the misconception that Z position not falling exactly on an step position would introduce rounding errors, which it doesn't since the firmwares just drive the steppers to the absolute Z position times the Z steps per mm value. The possible transient error is microscopic and since it doesn't accumulate, it has no practical effect. (Perhaps someone saw an improvement due to some other reason when changing layer heights and interpreted the cause being the layers falling on exact step values.)

I could maybe see that the actual micro stepping offsets from a given full step are not the same as offsets on other full steps. Depending on which full step you land in your offset might be slightly different. However, I don't see how that's going to be worse than the 5% error rating (0.09 degrees) of the full steps.

Yep.

Sorry that I have been absent for a bit. I actually think that I was wrong in my assumption.

I went back and did a few controlled tests. Believe it or not, I think that the reason why I suddenly had better results is because I turned _off_ my cooling fan. Weird, huh? If it proves anything, it is that there are a lot of variables at play for any given symptom, and something can make it appear better without actually curing it. I.E. I think that the cooling fan is a good thing but with it not there the plastic cools slower, and the vertical edges smooth out somehow. The fan didn't get rid of my z wobble, just masked the symptoms of it.

In short - I dont think that the z layer height has anything to do with my symptoms, but hopefully someone else learned from my question.

The reason that we used to worry about layer height being divisible by the Z full steps is because the Z motors used to be disabled between layer height changes. With old motor drivers, and more complex, heavier and more sticky axes of the original Mendel, the amount of current needed caused them to get hot if left enabled. But when the motor is re-enabled, it is always to the nearest full step. So you tried to make sure it moved in full steps, or you got inconsistent layers. Nowadays, most firmware leaves the Z axis enabled, so it isn't an issue; it stops at the current micro step, and stays there until it moves again. That's why it's possible to accurately print at 50 micron layers now.