Retraction Tuning Guide

Phaedrux

T lets you specify an unretract speed different from the retract. This is handy because molten filament behaves a bit like a non Newtonian fluid. Kinda like cornstarch and water mixture if you have ever seen that. If you press softly on it it flows like water. If you tap it hard it feels solid. Retracting fast helps pull the whole bit up the nozzle in one piece and then unretracting more slowly lets it flow more easily out.

R let's you modify the amount of filament that gets pushed back out after a retract. If your idler tension is very high it can deform the filament as it goes through causing it to have a slightly different effective diameter which slightly changes the extrusion rate. So in times wirh lots of short retracts close together the filament can get compressed and less will come back out leading to a bit of local under extrusion.

Alternatively if you have a very oozy material like PETG that bulges out when heated it can be handy to reduce the amount that comes out after a retract to try and reduce the z scar on the outter layer.

Danal

@ricky_grade10 said in Retraction Tuning Guide:

Through denial and error I was able to get some decent settings...

Ok, I know it was autocorrect.

Still made me chuckle.

Phaedrux

@danal I dunno about you but I read that and didn't even notice it as an error. I tend to make my way through life by denial and error.

Ricky_Grade10

I never went to university to get a fancy PFD or whatever, I only got my grade ten.

Any suggestions on where to start with the T and R parameters? Are these usually small values?

Thanks for the help boys

Phaedrux

I usually use 20-40mm/s for unretracting. 90-133mm/s for retraction.

The R value is usually a very small positive or negative decimal value.

Beauty of firmware retraction is that you can adjust the values during a print. So you can start a retraction test print and make changes on the fly to see how they effect it.

wilriker

Something I wanted to share here because the Printer Settings plugin for Cura came up: currently there is a bug in CuraEngine (the part the does the slicing in the background) that will create broken GCODE if you enable FW Retract, Relative Extrusion (as recommended for RRF) and Coasting at the same time. This is fixed in source code but will not be part of Cura 3.4 but only the version after (probably Cura 3.5 - not sure about minor fix versions).

Just so you know, do not use this combination right now.

Dino

As a slight aside I spent most of my Saturday setting up pressure advance on my delta, part of that included using firmware retraction.

I use S3D as my primary slicer on the delta and for $150 software it doesn't have a nice checkbox to enable firmware retraction I used post processing scripts I found here:

http://thrinter.com/using-firmware-retraction-with-simplify3d/

I had to turn off relative extrusion distances to get the script to work properly but now that it's setup it seems to be pretty nice.

Only thing I noticed was if I sent all the parameters for retraction distance, speed, unretract extra distance, and zhop together in the same command the zhop wouldn't apply. I added a second command and moved the zhop command to it which seems to have fixed it.

I have all my default retraction gcode in my config.g and have alterations of it in the start gcode commands for different filament if different retraction is needed. Plus the ability to make on the fly changes to retraction is fantastic, I can disable retraction mid print or add more shop if I need, or even print multiple retraction tests at once with different settings easily.

Ricky_Grade10

So if the R value is neglative does that mean it will retract the filament even more?

I guess Im having problems trying to figure out the order of the retraction process.

In my brain compartments, I have been thinking it would retract first, then z hop and finally un-retract. Is that the way she goes?

wilriker

@ricky_grade10 said in Retraction Tuning Guide:

So if the R value is neglative does that mean it will retract the filament even more?

No, that would mean it would unretract less than it has retracted, e.g. if you set
M207 S5 R-0.5
that would mean it retracts 5mm but only unretracts 4.5mm (whyever you would wanna do something like this).

I guess Im having problems trying to figure out the order of the retraction process.

In my brain compartments, I have been thinking it would retract first, then z hop and finally un-retract. Is that the way she goes?

That's exactly what it does.

Retract amount specified by Snnn at speed of Fnnn
lift Z by the amount of Znnn
do whatever slicer has put in between
lower Z by the amount of Znnn
unretract Snnn + Rnnn at the speed of Fnnn (or Tnnn if that is given)

burtoogle

@phaedrux said in Retraction Tuning Guide:

@burtoogle seems so. I don't tend to use the betas. Experimental features is as far as I'm willing to go. But glad it will become widely available. Though I still think firmware retraction belongs in the main settings.

Thanks for the heads up.

FYI Cura 3.4 has been released today.

wilriker

@burtoogle Any idea when this release will make it to GitHub? As that is the base for the package manager in Arch Linux.

burtoogle

@wilriker said in Retraction Tuning Guide:

@burtoogle Any idea when this release will make it to GitHub? As that is the base for the package manager in Arch Linux.

I don't know but the tags normally appear about the same time as the release so it shouldn't be long now.

Ricky_Grade10

So if the R value is positive, it will unretract more and act like a primer of sorts? Im guessing I should start with a small value for this parameter?

mrehorstdmd

@ricky_grade10

yes, start small.

I was using a volcano hot-end with a 0.8 mm nozzle and found that after I tuned the extra priming, it would work for the parts I tuned it for, but not for other parts/arrangements on the bed.

There is a plugin for Cura called "scalable extra prime" that adjusts the extra prime amount as a function of the travel distance. It's not usually needed with "normal" sized nozzles, but can be a big help with large nozzles.