Sep 8 2014
In the video you can see Marlin running 1/64 microstep @ 13000 / 16000 mm/min feed rate, using HTD3 with 12 tooth pulley –> apx 77khz / 94 khz –> quad stepped 19khz / 23 khz timer freq.
The first part (before Y move as a delay ;-)) uses same positions and speed like the second part, just a dummy gcode command (G4 1000, but can be any other command as well) is inserted between the moves.
As you can see (hear) there are small interruptions at the end of each move. It sounds like some steps might be lost (maybe caused from processing the serial input buffer or something else…).
This might explain why some people report problems with “shifting layers”, “incorrect sizes”, …
This test shows a serious problem which needs investigation. Updates will follow 🙂
Here is the GCODE:
G0 X10 F13000 G0 X110 G0 X10 G0 X110 G0 X10 G0 X110 F16000 G0 X10 G0 X110 G0 X10 G0 X110 G0 X10 G0 Y10 F8000 ; move Y as a delay G0 Y100 G0 Y10 G0 X10 F13000 G0 X110 G0 X10 G0 X110 G0 X10 G0 X110 F16000 G4 1000 G4 1000 G4 1000 G0 X10 G4 1000 G4 1000 G0 X110 G4 1000 G0 X10 G4 1000 G0 X110 G4 1000 G0 X10 G4 1000 G4 1000
(Icon: “Integrated Chip, SMD 64 pin” © Double-J Design, CC-A-3.0)