Repetier anyone?

[orcinus]

I've decided to give Repetier (FW and Host) and Cura a go, in place of my usual Marlin/Pronterface/Slic3r stack.

Initial installation was a breeze (Repetier is rather well documented, both in-code and in its wiki) apart from one crucial issue - the current stable version (0.7x) has completely broken support for homing to max (instead of min) and doesn't recognize max endstops *at all*. The development branch (0.80) works quite well, though (although it does come with some caveats of its own - the SD support is broken, you mustn't use Repetier's replacement serial library and you have to use Arduino 1.x).

The part that's a pain in the butt is getting everything calibrated to the point where it behaves as well as my previous stack (i'm shelving the idea of getting better results for now). Issue #1 was getting the corners of the calibration cube not to curl. The side that curls up the most is the one where the first layer move starts, which i took to mean i need to play with retract (too much material getting deposited in the corner due to oozing). Doing some voodoo with Repetier's own firmware-based retract (OPS) settings helped and i'm not getting any curling there, although tops of the corners (top as in visible on the top surfaces) are still a bit lumpy.

Now i'm facing some new issues. Namely, the vertical edges of objects are uneven (like wobble, but isn't, because the same artefacts aren't present in prints i did with the M/P/Slic3r stack) and i can't get the top surfaces to print neat (either i get holes or uneven lines, like W/T was set wrong and the filament got stretched too much). Oh well...

Anyone with any Repetier / Cura experiences to share? What are you using for acceleration/jerk/feedrate?

[cvoinescu]

One possible reason for uneven layers is a layer size that is not a multiple of the travel for one step of the Z axis. If you have, say, 1.8 degree steppers and M8 thread, one step advances (1.25 mm/turn) / (200 steps/turn) = 0.00625 mm/step. You want your layer size to be a multiple of that. If you are using microstepping and you really have to do it, make the layer size a multiple of the microstep travel, but it's best if it's a multiple of the whole step.

This is just a guess; it may well be something else entirely.

[orcinus]

Nah, my layers are set to a multiple.
It's not that kind of unevenness...

It's more like a gradual shift in X or Y. Like belt slippage, but extremely gradual and it tends to go one direction for a dozen layers, then another direction for the next dozen etc. Weird. Like there's a fixed error in the X and Y axes that keeps getting compounded.

It's not something drastic, mind you - you can't notice it on the faces of the calibration cube, for example... It's only visible on the vertices.

Took everything apart today anyways, to add a keyboard, tighten up the belts and do some other tweaks, going to see how it behaves when i put everything back.

[Turbo442]

Nah, my layers are set to a multiple.
It's not that kind of unevenness...

It's more like a gradual shift in X or Y. Like belt slippage, but extremely gradual and it tends to go one direction for a dozen layers, then another direction for the next dozen etc. Weird. Like there's a fixed error in the X and Y axes that keeps getting compounded.

It's not something drastic, mind you - you can't notice it on the faces of the calibration cube, for example... It's only visible on the vertices.

Your missing steps, figure which axis is doing it and adjust the current on that stepper driver. This is assuming your belts and pulleys are not too tight. You might want to try increasing the current on x and y 1/8 of a turn in and see what happens. I am betting your problems go away. Some times when you only miss one or two steps you won't hear it. Make sure you have a fan blowing air over your stepper drivers.

[Turbo442]

Another quick test would be to slow the print down and see if the problem goes away. If it goes away your probably missing steps due to the higher speed. Adjust your driver current settings and make sure your bot is running free without any binding.

[orcinus]

Why would switching to a different firmware (with everything else, acceleration, jerk and feedrates included, kept the same) cause missed steps?

[Turbo442]

Why would switching to a different firmware (with everything else, acceleration, jerk and feedrates included, kept the same) cause missed steps?

Possibly a hardware adjustment on the printer changed causing a bit more load on the motors?

Maybe it happened before and you didn't notice it? Maybe the print you are testing with gives the printer a real work out?

Are we really sure the acceleration, jerk and federate run exactly the same between marlin and repetier? It wouldn't take much of difference to mess things up.

I did a print he other day and I watched the print miss 3 steps in Y axis after I tightened the Y belt a bit. I could see it plain as day, printing smooth and then the whole print shifted over one step in Y, it would run fine for 5 minutes and do it again. I loosened the belt back up a smidgen and the missed steps went away.

[orcinus]

Nothing changed mechanically and i was printing the same object (20mm calibration "cube") with the same filament.
On top of that, the acceleration/jerk/feedrates are extremely conservative. I have tested at much higher speeds (3x) with no missed steps. Plus, changing the speed didn't make a difference (which leads me to believe it's not missed steps).

Plus, it would have to miss, then gain steps to achieve the effect i was seeing (the starting point is the same for each layer of the cube).

If anything, i'd say this is an issue with the volume of material deposited. I think there's something wrong with the extruder calibration or retraction. Either that, or some artefact in the dev branch of firmware. I'll do some more testing after i put the printer back together and recalibrate again.

[canadianavenger]

Different firmwares may handle those values differently as they likely use different algorithms for motion, [it's the primary reason we have so many firmware options... everyone says "I can do it better"] so you will likely need to do some tweaking of the values. Also note that how the firmwares calculate the extrusion volume differs as well, so be sure you are passing the expected parameter. There was a good write-up on this a while back, but I can't remember where it was... might have been a post in the forums here.

[orcinus]

Found the culprit.

As it happens, the issue only surfaces in the corners. What looks like wobble is actually too much material being deposited in the corners at times. Ditto for the curling corner (which is actually the start/stop point for the layers). Turning off Repetier's OPS exacerbates the problem.

I'm going to give it a go with OPS off and go back to Slic3r, then try it with the retract settings i know work.

I don't think i like the way OPS works in Repetier at all...
The way it's supposed to work is, doing the retracts during the travel moves. However, the implementation has two issues:
1) the retract happens at the speed that's proportional to the travel speed and time (instead of doing it with a preset acceleration/speed)
2) the re-extrusion happens after the travel move is finished, so the head just loiters over the start point while extruder is being primed (causing blobbing and, more importantly, heating up the part immediately beneath)