My custom Cartesian

wilriker

@deckingman But what to do with my luck? Driving Z all day up and down... I could write a macro for it, it's RRF after all.

But still your motor seems to be better than mine plus you probably use 24V.

My X gantry weighs about 2kg (probably much less but easier for the following math) and I have two Z motors and therefore only 2 leadscrews. That makes a quarter of the weight, double motor power and only 2/3 of the leadscrew friction. All together my Z axis should run 8-12x as fast as yours and I only have it about 4 times faster.

But as you also mentioned: when does one need speed in the Z axis?

deckingman

@wilriker said in My custom Cartesian:

@deckingman But what to do with my luck? Driving Z all day up and down... I could write a macro for it, it's RRF after all.

But still your motor seems to be better than mine plus you probably use 24V.

My X gantry weighs about 2kg (probably much less but easier for the following math) and I have two Z motors and therefore only 2 leadscrews. That makes a quarter of the weight, double motor power and only 2/3 of the leadscrew friction. All together my Z axis should run 8-12x as fast as yours and I only have it about 4 times faster.

But as you also mentioned: when does one need speed in the Z axis?

Ahh, but what is the lead of your screws? Mine are fine lead (1mm) so I've probably got better gearing than you.

There is one situation that I could think of where you might need faster Z and this is if you homed to Z max - for example if you want to recover from a power loss on a CoreXY. I don't do that myself as the bed would cool during a power fail and the part would fall off in any case. For me to home to Z max would take 200 seconds (100 seconds to go down 750mm and 100 seconds back up).

wilriker

@deckingman said in My custom Cartesian:

Ahh, but what is the lead of your screws? Mine are fine lead (1mm) so I've probably got better gearing than you.

Mine is really coarse at 8mm. But that gives me also fewer steps/mm what makes it less likely to get to the speed limit of the motor.

There is one situation that I could think of where you might need faster Z and this is if you homed to Z max - for example if you want to recover from a power loss on a CoreXY. I don't do that myself as the bed would cool during a power fail and the part would fall off in any case. For me to home to Z max would take 200 seconds (100 seconds to go down 750mm and 100 seconds back up).

Yeah, you're right about that case. Luckily and solely by coincidence I can re-home to regular Z min if X is at min and Y at max position. I could only hit something if I would print in the very back left corner and I never did that so far.

So, no real need for a Z faster than needed at printing. We could invent vertical layers.

deckingman

@wilriker said in My custom Cartesian:

We could invent vertical layers.

I already have a technique for that - it's called printing something on it's side.

wilriker

@OBELIKS Printed a new Benchy with walls before infill today. Will check the result in detail when I get home. Based on what I can see over the webcam the issue with very visible infill patterns seems at least to be reduced but not completely gone. But images can be misleading. Will report more details tonight.

Phaedrux

@wilriker You can also modify the amount that infill crosses over into the walls.

wilriker

@phaedrux I know but that already is as low as (Cura's default of) 10%. Going lower doesn't make sense IMHO, does it?

Here are the results of printing walls before infill. The backside is totally smooth now. The front side still some infill is visible and it can be felt when sliding a finger across the surface. To my surprise it is rougher though than previous ones.

Front side. Previous result on the left, today's Benchy on the right.

Back side. Again old Benchy to the left, new one on the right.

Phaedrux

How many walls are you using? Adding another wall would be my suggestion. With a 0.4 nozzle I usually use 3.

wilriker

@phaedrux With most of the settings I go by Cura's defaults. In case of perimeter count it is 2. I only increase this for parts that need additional strengths e.g. to directly tap screws into the plastic.
I still have a setting to test called "Connect infill lines". It basically is an additional inner perimeter. I guess this will solve the issue finally.

Phaedrux

@wilriker Connect infill lines isn't quite the same as an extra full perimeter. It does the same thing with infill that slic3r does by default. A full perimeter is a continuous lines from start to finish, connected infill lines are sporadic.

I use 25% for infill overlap and don't get such pronounced infill pattern. What is your infill wipe distance?

wilriker

@phaedrux Again the default of 0.1125mm.

Phaedrux

2 perimeters is rather thin, especially on curved surfaces like the hull. I'd bet that adding a third perimeter will take care of it.

In my opinion the Cura defaults are some of the worst defaults. They are tuned for an ultimaker and don't always translate well to other styles of printer. And so many of the useful settings are hidden by default. That's not to say I dislike Cura, but it definitely takes a lot more fiddling.

In your case, with an I3 style printer, I think you'd probably get really excellent results using Slic3r PE and one of the prusa profiles for the MK2.

https://github.com/prusa3d/Slic3r-settings/tree/master/old/Slic3r settings MK2S MK2MM and MK3

OBELIKS

Just to test, try this one:0_1536176181631_3DBenchy_0.2mm_PLA_MK2S.gcode

wilriker

@phaedrux I am looking into other slicers every once in awhile but at least until now always returned to Cura. Maybe because it was what I started with.

I will try to slice and print a Benchy also with Slic3r and maybe also ideaMaker. Will see. I will just create an army of Benchies I can use to conquer the world.

wilriker

@obeliks Is it sliced with (0, 0) at bed center?

OBELIKS

You have it set at center now?
It's at 0,0

wilriker

@obeliks Yes, I changed it with PA tuning.

I looked into the GCODE with an editor and a lot of movements are outside my printable area. So I guess this was sliced with (0, 0) at front left.

Also I doubt that my motors could handle the speeds given in there. 144mm/s is not possible - while proof-reading this sentence I just thought "actually I never tried but only used EMF Calculator values to set the limits". Usually I wouldn't care as it is capped by my settings but these are overwritten at the start of the file.
I won't print this before Friday anyway as tomorrow is something else already on the list. So enough time to setup Slic3r and slice it myself.

Also it does not use FW retract and the retractions in there are way too short for my MK8.

OBELIKS

I recommend you start with MK2 profile and work from that.
Even if you just change the dimensions of print volume and remove the acceleration and other things it should work.

wilriker

@obeliks Thanks. Currently fighting through all of the options.

wilriker

OK, some interesting discoveries with Slic3r PE so far.

I started with MK2 profile but ditched it and started with a completely fresh profile based on a custom printer. Also I left nearly all of the settings at their defaults as this will compare better to how I use Cura
The Gcode file for the Benchy is considerably smaller, i.e. nearly 40% smaller
It uses about 70cm more filament for the Benchy than Cura - this is probably due to larger line width and 3 perimeters
With all speeds set to default (except travel speed that I reduced to match my printers current set limit) it will take 1:46h compared to 1:28h when sliced with Cura (both values taken from Duet simulation result)
Slic3r said it would take 1:25h but I have not found where to set maximum speeds for axes, accelerations and jerk values. It can be entered in the advanced acceleration control fields if they are enabled by using a Prusa profile but not for custom profiles. But I don't want it to control the acceleration but only take it into account when calculating print time. Before the "Printer Settings" plugin existed these values had to be entered in configuration files with an external editor. ~~Is there something similar for Slic3r possible?~~ Just found it right after posting here.
Modifier meshes are a bit more hidden than in Cura but have the nice feature of adding generic shapes (cubes, cylinders, etc.) where I had to create those in advance for Cura
I like that Slic3r has Gyroid infill
I don't like the UI - it looks like from the 90s