Musings and Experiments on the Art and Science of 3D Printing


Mosaic Palette 2 - Understand and Control Color Bleeding

By SublimeLayers Saturday, January 5, 2019

A print showing color bleed of red into the pearl white. Yuck!
Much better! Read on to learn how to get results like this.
I've been having a lot of fun with my Palette 2 over the last few months. By way of disclaimer, Mosaic gave me a P2 as a thank you for being a product tester over the last year or so. But, free or not, if a product doesn't live up to my high demands (and expectations) I will not endorse or promote it.

The P2 is a remarkable machine, completely redesigned from its predecessor, and capable of producing outstanding prints. However, like all things in life, a little understanding will go a long way to get the best results. One problem that new owners complain about - and usually mischaracterize as a splice calibration error - is color bleed. Splice calibration errors result in pure color being printed in the wrong place.

Look at this closeup photo of the salamander's head:

That light pink is not the same as the pure red in the eye. It is color bleed and really doesn't have much to do with the P2 at all - except that you can minimize or prevent it if you understand what causes it and do some quick experiments to refine transition tower tuning in Chroma or CANVAS (I highly recommend CANVAS - it is both a slicer and the Palette's magic software). This post will walk you through the color bleed tuning I've developed to achieve results like the second photo.

What is "bleed"?

First, some comments about why bleed happens. The simple answer is "we don't know exactly why it happens"! Bleed is not unique to Palette - it also affects multiple extrusion hot ends that merge two or more filaments through a single nozzle orifice like the E3D Cyclops. Practically speaking, the outgoing filament is not being completely purged by the incoming filament. That's the simple part to understand and adjust in Palette (or even Cyclops). However, some filaments seem to find hiding places somewhere in the hot end from which they bleed out at the most inopportune moments.

Years ago I noticed that every once in a great while, a streak of black PLA would creep into a light colored PLA print. In some cases, I had not printed black PLA for several prints! This black must have found some nook or cranny to hide in. We have some ideas on this but these can not explain all circumstances. More work and experiments are needed to improve our understanding. And through understanding comes superb prints!

In general, some types of filament bleed worse than others. ABS does not seem to be prone to a lot of bleeding. It does bleed, but is usually purged cleanly. PETG is another one that is not highly prone to bleeding. PLA seems to be the worse of the lot and certain colors like red and black bleed worse than other colors.

In theory, the new, incoming filament, should push the previous filament out the nozzle without leaving behind any remnants. This must not be happening, otherwise there would be no bleeding. So let's think about what's happing in practice.

  1. The filament is smaller diameter than the bore it is being pushed through. Even in the heater block area, the bore is larger than the unmelted filament diameter. The Palette works with 1.75mm (nominal) filament and the bores of most 1.75mm hot ends are actually 2.0mm. This leaves quite a bit of room between solid filament and the inner walls of the extrusion path. Perhaps molten outgoing filament can "circle back" and flow up and fill that space. Retracts - especially long 2mm retracts - might exacerbate the problem by pulling the outgoing filament up and around the cold incoming filament.
  2. The length of the melt zone increases the amount of molten filament. The more molten filament there is, the more trouble it can get into like filling voids and pushing up the extrusion path.
  3. The extrusion path may have hiding places. Consider the E3D V6 hot end. Its nozzle is tightened up against the end of the heat break. This could leave a little gap - especially if the edge of the bore is chamfered slightly. There may be other hiding places depending on the construction of the hot end. Also, the extrusion path may not be polished or slippery enough to keep molten filament from sticking to it. We already have some evidence that Chinese knock-off V6 "style" hot ends are not machined and polished to the same exacting tolerances as a true E3D hot end. Photos of the bore of the heat break looks like the surface of the moon. It's easy to imagine filament getting stuck and collecting in all of the grooves and scratches.
Here is a close up photo of a Chinese V6 heat break. This photo was taken by Bruce Jenkins (used with permission) and was posted on the Palette Facebook group as part of a conversation on bleed. You can clearly see the annular grooves in the bore waiting to trap and hold molten filament.

There are most likely are other contributors too. But let's look at these three categories and try to understand if there is anything we can do to improve the situation.

Case 1: space between the solid filament and the extrusion path

The extrusion path bore must be slightly larger than the filament diameter to accommodate differences in diameter from one filament to the next - or even within the same filament. It might be possible to reduce the bore to 1.85mm or so but that would require carefully monitoring your filament to make sure it doesn't cause a problem.

Some knowledgeable people have studied this problem in the past and one of them, Jetguy, referred me to some discussion on PTFE lined hot ends like the V6 Lite. The PTFE lining extends all the way to the nozzle and does several things: it's much slipperier than polished metal, it insulates the melt area to more localize the melt zone and as it heats it expands and decreases the inner bore diameter (because the outer diameter is constrained by the metal hot end. I am planning to do some direct comparison tests of V6 vs V6 Lite hot ends to test this.

Case 2: melt zone length

The length of the melt zone is primarily determined by the thickness of the heat block. So, shortening this block could result in less molten filament to bleed. Another contributor to the melt zone is the efficiency of the heat exchanger above the heat block. In an all metal hot end like the V6, the heat break helps minimize the flow of heat up into the exchanger but if you've ever had a heat exchanger fan die or disconnect, you'll know first hand that the filament will melt for a cm or more up into the exchanger. So, more efficient exchanger cooling could also minimize bleed. The most efficient cooling is water cooling. I've been an advocate of water cooling for many other reasons but it might also have a positive effect to minimize bleeding. Again, another experiment to conduct!

Case 3: extrusion path hiding places

Removing hiding places may be as simple as using a high quality hot end that's been machined and polished to precision or replacing inferior components like a heat break with a quality component. PTFE lined hot ends are certainly smoother than metal hot ends and worth conducting some experiments to see how they perform with respect to bleeding.

Now that we have a little better understanding of what causes bleed and some things that might contribute to it, there is still no guarantee that you can completely eliminate it. But don't despair, you can greatly improve your print quality with some simple tuning - color bleed tuning - as described next.

Color Bleed Tuning

To perform this tuning you'll need to download two models:


And it is best if you follow along using CANVAS. If you prefer to use Chroma and your favorite slicer, you'll have to translate the slicing parameters to your slicer.

Bleed Test Print

  1. Load both models into CANVAS and set the bottom part to the darker color and the upper part to the lighter color you want to test.
  2. Slice in Vase mode, DO NOT use a transition tower (select No handling).
  3. Slice with your standard layer height and extrusion width. Make note of extrusion width you'll need this later.
  4. Print the model.

Now, measure the the distance from the bottom of the part to an area where the upper lighter color is clean and not tinted. It is better to over measure a little than under measure. Make note of this height. (the example shows 12mm to very clean white).

Now some SIMPLE math:
  1. subtract 5mm from your measured height - this is your “clear height” in the formula
  2. calculate the transition length using your extrusion width and clear height using the following formula:

65.314 * extrusion width mm * (clear height mm - 5mm) = transition length

I found that this works great with a 20% transition but you need to have a well calibrated extruder, P2 and ability to precisely start the filament when initiating a print. If you need to maintain a 30% or 40% transition, you will likely need to increase the transition length by 10% or 20%, but this will get you in the ballpark. I am conducting further experiments and also working on studying the effect in various hot ends to test a hypothesis on where the bleeding derives.

PLEASE NOTE: this formula has been developed and tested with CANVAS with its integrated slicer. I am doing more research on Chroma and its supported slicers but some Facebook group members have had success with it on Chroma too.

The final step is to test your new transition settings by printing the model again, but this time enable the transition tower with your calculated Transition length.
  1. Load the model into CANVAS and set the bottom part to the darker color and the upper part to the lighter color.
  2. Slice in Vase mode, USE a transition tower (select Transition tower). Set your Transition length to the value calculated above and set Transition target to 20% (or 30% if you must).
  3. Slice with your standard layer height and extrusion width.
  4. Print the model. The top section should be clean with a crisp separation from the lower dark color.

No Comment to " Mosaic Palette 2 - Understand and Control Color Bleeding "