Retraction is a key parameter to achieve good prints.

Retraction Tuning

To tune your retraction, you iterate one parameter at a time as follows:

• Maximize (conservatively) non printing travel speed.
• Temperature of the hotend.
• Retraction Length.
• Extrusion Flow.
• Retraction Speed.
• Retraction Acceleration.
• Additional parameters either printer, firmware or slicer dependent.

Slicer settings

At the slicer I enable "fully" the retraction: we want to have as many retractions as possible.
My settings are the following:

• Enable retraction: extrusion length set to 0mm, travel move distance set to 10mm or less.
• Disable all features that affects the retraction or prevent it: combing, avoid crossing perimeters, wipe...

Retraction Benchmarks

Typically the process is to use a few so called "retraction benchmarks" to try to setup this.
You try with some "conservative" value according to the kind of extruder/hotend setup you are running: direct drive (0.5mm to 1mm), bowden (2mm to 4mm) and you iterate over.
I will just dive a bit more into.

I use first a raw retraction test (https://www.thingiverse.com/thing:2465487) that is simply 3 cylinders that are spaced from each other.
Once this one is more or less ok, I move to a more intensive benchmark that are 4 small peaks (https://www.thingiverse.com/thing:909901).

The catch

After some iterations you obtain a correct result, do you think it is ok ?

I will just talk about an issue I had: I was trying to print a few times a specific part and my hotend was jamming after a few hours.
I think first, it was a slicer issue, I checked the toolpath and a few other things, it was correct. So I unjammed the hotend and re-tried, then failure a few hours after.
I know that the hotend that I have on this given printer is prone to heat creep. So, I tried to print during the night when it is colder and however yet another failure.
I spent sometime trying a few other things, but wasting time.
Finally, I spent some time looking at the filament inside the bowden tube and I saw that progressively there was more and more grinding on the filament: the hotend was progressively becoming clogged. So, I reduced a bit the retraction length and not any more a failure.

What is the point ?

Do not spend too much time optimizing your retractions settings for the retraction benchmarks. You need to print a few real parts where you know retractions are quiet aggressive to validate your settings. Those parts need to be relatively long enough (>2 to 4 hours) in order to validate that there is not any progressive hotend clogging. Otherwise, you will loose your time failing, blaming your filament, your hotend, your extruder, whatever else but it was just a bad retraction setting in the first place.

Happy printing !