Got rid of ringing/ghosting almost completely  

Re: Got rid of ringing/ghosting almost completely

You will also require a "G4 S0" before the M205.

Without the pause, the M205 will be actioned immediately upon receipt - before the preceeding "G" commands.

Peter

Re: Got rid of ringing/ghosting almost completely

You will also require a "G4 S0" before the M205.

It works great now already. I will give it a try with the "G4 S0" and look for even more improvement. Thank you.

Re: Got rid of ringing/ghosting almost completely

Maybe you could open a ticket on Github to get something like this added to Slic3r PE?

Re: Got rid of ringing/ghosting almost completely

You will also require a "G4 S0" before the M205.

I just tested that, and it made things slightly worse because then the printer dwells a fraction of a second for the command queue to be empty before continuing. Having read the source code for the Marlin firmware, it appears that nearly all commands, including the M205 command are sent to the planner. That means that the M205 will become effective at the right time, exactly before the next move that is written in the source code. This is confirmed in a thread on https://github.com/MarlinFirmware/Marlin/issues/2125 .

So, a G4 command or a M400 command does not have to be sent before the M205 command.

Here is an update of the script which mitigates another python rounding error that sometimes aborts the script:

#!/usr/bin/python

filePath=sys.argv[1]
outputPath=filePath.rstrip("gcode")+"MOD.gcode"
fw=open(outputPath,"w")
with open(filePath) as fp:
line=fp.readline()
while line:
splitLine=line.split(";")
if len(splitLine)>1:
comment = splitLine[1].strip()
command=splitLine[0].strip()
command=command.split(" ")
if ("perimeter" in comment) or ("move to first" in comment) or ("move inwards" in comment):
modFlag=1
if command[0]=="G1":
if "X" in command[1] and len(command)>2:
x2=float(command[1].lstrip("X"))
y2=float(command[2].lstrip("Y"))
if (x0!=0):
fVx=x1-x0
fVy=y1-y0
tVx=x2-x1
tVy=y2-y1
lenfV=sqrt(fVx*fVx+fVy*fVy)
lentV=sqrt(tVx*tVx+tVy*tVy)
if lenfV!=0 and lentV!=0:
dvr=(fVx*tVx+fVy*tVy)/(sqrt(fVx*fVx+fVy*fVy)*sqrt(tVx*tVx+tVy*tVy))
if abs(1-abs(dvr))<0.00000001: dvr=1 #This is to prevent a python roundoff error where the dvr value becomes greater than 1
#print command, comment, dvr
angle=degrees(acos(dvr))
x0=x1
y0=y1
x1=x2
y1=y2
if modFlag==1 and angle>45:
if lastMod!="ANGLE":
#fw.write("M400\n")
fw.write("M205 X2 Y2\n")
lastMod="ANGLE"
#print angle
elif lastMod=="ANGLE":
lastMod="NOANGLE"
#fw.write("M400\n")
fw.write("M205 X20 Y20\n")
fw.write(line)
#print comment, command, angle
prev2Line=prevLine
prevLine=line
modFlag=0
line=fp.readline()
angle=0
print outputPath


Re: Got rid of ringing/ghosting almost completely

That really cool - can you post before and after print examples of the same model to show the improvement?
Texy

Re: Got rid of ringing/ghosting almost completely

Having read the source code for the Marlin firmware, it appears that nearly all commands, including the M205 command are sent to the planner.

From my reading of the code that is not the case (but I am quite likely incorrect 🙁 )

From Marlin_main:

case 205: //M205 advanced settings: minimum travel speed S=while printing T=travel only, B=minimum segment time X= maximum xy jerk, Z=maximum Z jerk
{
if(code_seen('S')) minimumfeedrate = code_value();
if(code_seen('T')) mintravelfeedrate = code_value();
if(code_seen('B')) minsegmenttime = code_value() ;
if(code_seen('X')) max_jerk[X_AXIS] = max_jerk[Y_AXIS] = code_value();
if(code_seen('Y')) max_jerk[Y_AXIS] = code_value();
if(code_seen('Z')) max_jerk[Z_AXIS] = code_value();
if(code_seen('E')) max_jerk[E_AXIS] = code_value();
if (max_jerk[X_AXIS] > DEFAULT_XJERK) max_jerk[X_AXIS] = DEFAULT_XJERK;
if (max_jerk[Y_AXIS] > DEFAULT_YJERK) max_jerk[Y_AXIS] = DEFAULT_YJERK;
}
break;


Peter

Re: Got rid of ringing/ghosting almost completely

From my reading of the code that is not the case (but I am quite likely incorrect 🙁 )

What version of Marlin are you looking at? The one on github (1.19) shows the following code:

inline void gcode_M205() {
if (parser.seen('Q')) planner.min_segment_time_us = parser.value_ulong();
if (parser.seen('S')) planner.min_feedrate_mm_s = parser.value_linear_units();
if (parser.seen('T')) planner.min_travel_feedrate_mm_s = parser.value_linear_units();
#if ENABLED(JUNCTION_DEVIATION)
if (parser.seen('J')) {
const float junc_dev = parser.value_linear_units();
if (WITHIN(junc_dev, 0.01f, 0.3f)) {
planner.junction_deviation_mm = junc_dev;
planner.recalculate_max_e_jerk();
}
else {
SERIAL_ERROR_START();
SERIAL_ERRORLNPGM("?J out of range (0.01 to 0.3)");
}
}
#else
#if ENABLED(HANGPRINTER)
if (parser.seen('A')) planner.max_jerk[A_AXIS] = parser.value_linear_units();
if (parser.seen('B')) planner.max_jerk[B_AXIS] = parser.value_linear_units();
if (parser.seen('C')) planner.max_jerk[C_AXIS] = parser.value_linear_units();
if (parser.seen('D')) planner.max_jerk[D_AXIS] = parser.value_linear_units();
#else
if (parser.seen('X')) planner.max_jerk[X_AXIS] = parser.value_linear_units();
if (parser.seen('Y')) planner.max_jerk[Y_AXIS] = parser.value_linear_units();
if (parser.seen('Z')) {
planner.max_jerk[Z_AXIS] = parser.value_linear_units();
#if HAS_MESH
if (planner.max_jerk[Z_AXIS] <= 0.1f)
SERIAL_ECHOLNPGM("WARNING! Low Z Jerk may lead to unwanted pauses.");
#endif
}
#endif
if (parser.seen('E')) planner.max_jerk[E_AXIS] = parser.value_linear_units();
#endif
}

Re: Got rid of ringing/ghosting almost completely

What version of Marlin are you looking at? The one on github (1.19) shows the following code:

Well, this is the Prusa Research forum, so I am looking at the Prusa Research Marlin branch version 3.4.0

Peter

Re: Got rid of ringing/ghosting almost completely

hans.d3 wrote: ↑
Sep 25, 2018 4:41 pm
What version of Marlin are you looking at? The one on github (1.19) shows the following code:
Well, this is the Prusa Research forum, so I am looking at the Prusa Research Marlin branch version 3.4.0

Peter

Ah, I see. I did not realize that there is a separate Marlin for the Prusa printers. I use the the latest main Marlin firmware on my RAPMS1.4 board (with my Prusa I3 box version printer). The reason I use the Prusa Slic3r is that it is just so much better than the original Slic3r, especially where support material is concerned.

So, if you are using the Prusa Marlin firmware: Yes, you will need the M400 command before the M205 command. It is already in my script. Just uncomment the two lines that contain #fw.write("M400\n"). You should still benefit from the ghosting being almost gone, but suffer slightly from the short dwell that happens after each M400 command in the gcode. It is a pity that the Prusa Marlin did not include the changes that were incorporated in the main Marlin around 2016 because not forwarding the M commands to the planner makes no sense whatsoever, except for a very few commands that need to be executed instantly such as an abort command.

Re: Got rid of ringing/ghosting almost completely

That really cool - can you post before and after print examples of the same model to show the improvement?
Texy

Ok, I try now to add two pictures. If this post succeeds, you will see three Marvin statues in the pictures. The most left one is the one that was processed with my script. The middle one is with the standard jerk value of 20. The most right one is one where the jerk value is lowered to 2, but for the whole file.

You can clearly see some bad ghosting in the middle (unprocessed) one. And you can see some terrible vertical banding all around (but no ghosting) in the most-right (low jerk values) one. My script solves the banding issue by only applying a low jerk value where it is needed, and resetting it to a high one where it is not needed. Thus no ghosting and no banding.

All three were printed in PETG. I apologize that the most-right one was printed in a different colour PETG than the other two. But the banding shows up on all curved objects when jerk is low for the whole print, regardless of the material.

Re: Got rid of ringing/ghosting almost completely

doesn't filament pick the worst possible time to run out?

Re: Got rid of ringing/ghosting almost completely

It is a pity that the Prusa Marlin did not include the changes that were incorporated in the main Marlin around 2016 because not forwarding the M commands to the planner makes no sense whatsoever, except for a very few commands that need to be executed instantly such as an abort command.

How right you are! Until recently, there was also a requirement to include a pause immediately before a tool change, otherwise the tool number would just be a spurious character from a subsequent line of G-code 🙁

Peter

Re: Got rid of ringing/ghosting almost completely

The middle one is with the standard jerk value of 20.

Currently for the Mk3, the XY Jerk value is set to 10.

Peter

Re: Got rid of ringing/ghosting almost completely

Currently for the Mk3, the XY Jerk value is set to 10.

Peter

A value of 10 is not low enough to get rid of ghosting on my printer, but it is low enough to cause banding on curvatures. I tried that by printing hollow cylinders with a radius of 45mm. Ghosts could be seen at the layer change seems, and vertical bands could be seen all around the cylinders (though less noticeable than with lower jerk values). With jerk values above 20 I get really smooth curvatures without bands.

Re: Got rid of ringing/ghosting almost completely

On a side note, I am able now to call the script from the post processing scripts option in Slic3rPE. It turns out that the Appimage of Slic3r that I was using, somehow does not play well with python. I am now using the pre-compiled version, and it calls the script without issues. After clicking on "Export G-Code", two files are saved: The one that is generated by Slic3r and the one that has been modified by my script. So behavior is now just as if it is a built in feature of Slic3r. Really cool. I am so happy. It feels like I can squeeze some more years out of my (much modified) Prusa I3 box version now.

Re: Got rid of ringing/ghosting almost completely

[...] Currently for the Mk3, the XY Jerk value is set to 10.
A fresh install of Slic3rPE 1.41.0 has X & Y jerk set to 8. They seem to be ratcheting the acceleration and jerk settings back a bit with each release.

Re: Got rid of ringing/ghosting almost completely

A fresh install of Slic3rPE 1.41.0 has X & Y jerk set to 8. They seem to be ratcheting the acceleration and jerk settings back a bit with each release.

Yes, I noticed that and have the same setting already in KISS. But the firmware setting is 10 😉

And yes, they are playing around with the settings quite often (example: E acceleration of 8000 is just way too high, but is thought to be "required" by the MMU - I use 1000 with the MMU for that setting).

Peter

Re: Got rid of ringing/ghosting almost completely

That really cool - can you post before and after print examples of the same model to show the improvement?
Texy

Ok, I try now to add two pictures. If this post succeeds, you will see three Marvin statues in the pictures. The most left one is the one that was processed with my script. The middle one is with the standard jerk value of 20. The most right one is one where the jerk value is lowered to 2, but for the whole file.

You can clearly see some bad ghosting in the middle (unprocessed) one. And you can see some terrible vertical banding all around (but no ghosting) in the most-right (low jerk values) one. My script solves the banding issue by only applying a low jerk value where it is needed, and resetting it to a high one where it is not needed. Thus no ghosting and no banding.

All three were printed in PETG. I apologize that the most-right one was printed in a different colour PETG than the other two. But the banding shows up on all curved objects when jerk is low for the whole print, regardless of the material.

That is really impression - thanks for sharing.
Ian

Re: Got rid of ringing/ghosting almost completely

I have made a new version. This time written in perl. Anyone should be able to use this version. Just copy and paste the code below into a text file, and name it: "deghost.pl"

If in Linux or Apple, make the file executable by typing: "chmod +x deghost.pl"

Then in Slic3r in Print Settings->Output Options, type the location and name of the script file in the Post-processing scripts box, and you are done. This also works with the appimage of Slic3r.

Here is the code below:
#!/usr/bin/perl

}