Delta + RAMPS: shifting layers, loosing steps

Hi there,

we recently moved from Marlin to Repetier for our Kossel Mini driven by a RAMPS 1.4 board. Everything seems to work fine, the nasty speed deviations we had with Marlin have vanished (see [1] for details).

However, it now appears that we are loosing steps (see attached image for failed prints) and I don't know the reason. First I thought the head hits the print but it seems that steps are lost without influence of force (indicators: bed plate was not displaced, I raised DELTA_SEGMENTS_PER_SECOND_MOVE and turned on lift-z-on-retraction as test). The stepper currents never have been a problem as Marlin did not loose any steps. The frequency and acceleration settings should not be unusual.
We also noticed that discrete vertical test structures become slightly oblique which also could indicate a loss of steps.

Could someone please look through our configuration file attached and tell if there are some unfortunate settings? I don't know the quadstepping feature well, maybe there is something screwed?

[1] [forums.reprap.org]

FYI: tried a few things: turned off endstop check, raised STEP_DOUBLER_FREQUENCY to 15000, set DOUBLE_STEP_DELAY to 1 (as seen in this post [github.com]), lowered all accelerations to 1000 max, lowered ZJERK to 0.1. However, it just seems to get worse. I attached an image of two test prints, with the right one after parameter change (same gcode).

Edit: are there any such facilities as to check for steps lost after homing? (i.e. compare is-position from home switches with counted must-position). And related: are there benchmarking scripts which test different speeds, accelerations and circumstances for limits of the printer hardware?

Edited 1 time(s). Last edit at 04/24/2016 01:31PM by ribaldos.

Here our temporary solution: switched back to Marlin using 30 mm/min minimum speed and .3 mm first layer height, here the bug reported previously does not occur.

Another two things we noticed with Repetier:

The motors are singing songs when printing (C, D, A, G#,...) while Marlin does not.

Low speeds (about less than 300 mm/min with 1000 max acc.) lead to severe problems, partially resetting the firmware at the end of move.

Your max. speed in much too high. Deltas can not compute 40000 steps per second. Think more like 20000 steps per second a sa realistic speed with AVR board. Use Due based boards for higher speeds.

Also make sure to have enough free ram. Deltas on avr use every byte and you need 900-1000 bytes ram to not crash firmware or do unexpected things. Arduino compiler shows ram usage or firmware on connect. Reduce line subdivisions if ram usage is too high.

---

Repetier-Software - the home of Repetier-Host (Windows, Linux and Mac OS X) and Repetier-Firmware.
Repetier-Server - the solution to control your printer from everywhere.
Visit us on Facebook and Twitter!