Skip to main content
User avatar

What is the maximum print speed the Palette 3 Pro can support?

What is the maximum print speed the P3P can support? I ran into an issue last night and I believe it is because the P3P couldn’t keep up with the speed of the printer.

I normally print at 75/mms with a 0.6mm nozzle with surface finishes at 90% of print speed. I was printing a simple 3 color badge with only 3 splices. On two occasions during the print I heard skipping from the extruder. On both occasions, the P3P was still working on a splice and the buffered filament was all used so there was a small tug of war going on between the extruder and the P3P. While the print completed there are visible lines missing from the top surface where the skipping occured and a little more than normal filament was left over after the print (expected due to skipping). I am able to recreate the issue so I am pretty sure that this is the case. I tried it again with a slower print speed (10% reduction) and the same thing occurred. Skipping only occurs when the P3P is still splicing and the buffered filament runs out. As soon as the splice is completed the P3P releases the filament and the extruder is able to resume. But the print quality suffers as no filament was layed down for the 2-3 seconds while the skipping occurred.


Can there be an option in Canvas that will allow speed reductions during printing when the P3P has to splice?

Since Canvas is the slicer and the P3P controls the splicing why can’t Canvas set a lower print speed before/after a splice to ensure the buffer doesn’t run out and then set the speed back to normal when there are no recent/upcoming splices? During slicing Canvas can set the correct print speeds in the gcode/mcfx file. This would allow us to print at our normal print speed and have Canvas automagically lower the speed to ensure we don’t consume the buffered filament while the P3P is still splicing.

Of course, I can lower the overall print speed but to have Canvas handle it automatically would be a huge improvement (and time saver) as we could both leverage the ability of our printers and the P3P with no manual intervention (while achieving high quality prints).

Answer this question I have this problem too

Is this a good question?

Score 5


I have had this happen often, at least the Palette 2S would slow down or pause the printer and not let hours of printing and meters of filament go to waste!


I have had this issue as well, ended up having to reduce my overall print speed by 50% from what I use normally, and had to very seriously slow down the transition purge moves - I use a dual-gear extruder that can push massive volumes of filament, so purging at full speed was almost always running the Palette out of buffer (especially if the print had a transition between black to white which requires 240 or so mm of filament)


@zinventor can you think of a reason why Canvas couldn't be updated to set variable speeds around splices to ensure the buffer isn't consumed? From what I have observed, the P3P doesn't have any issues PULLING/PUSHING filament when I print with my normal setup at 75mm/s through a 0.6mm nozzle with 0.46mm layer heights and a 0.62 width. Filament travels through all paths without issues and splices remain strong. So this gives me hope that Canvas could slice the project with a user defined maximum print speed and then slow the print down based on Mosaic's volumetric flow rate algorithm to guarantee the buffer is never fully consumed. This would allow us to print closer to our desired max print speed while keeping the printer and P3P in synch further eliminating skipping and ruined prints.

It would require some mods in Canvas' UI and the slicing engine, but from a technical standpoint I can't see why this couldn't be implemented fairly easily.

On a side note: During Mosaic's livestream tonight, it was mentioned PETG cooling doesn't need to be as long as for PLA. I have been using the default settings that came with P3P which is tuned for PLA but I print almost exclusively in PETG. Out of the box, the P3P splices were rock solid so I had no reason to change them. I hadn't thought that I might be able to "cheat" an extra 1-3 seconds back per splice through a simple tune. I'll update the tune profile and see if that buys enough time to mitigate the complete buffer consumption/skipping issue. Of course, not everyone prints with PETG but hopefully, it can help or increase print speeds for those that do.


I definitely think it would be possible, however it may be a challenge.

My first thought was just to have the palate send feed rate override commands to the printer, but that could be problematic if the user wants to set a feed rate over ride independently for some other reason. That leaves actually changing the feed rate in G-Code, which may be a bit more complicated.

In the G-Code, the Slowdown would actually need to happen substantially before the splice occurs, because of the outgoing tube distance, which means that the software would need to account for variable length outgoing tubes and synchronizing. That being said, the software already accounts for that in generation of spice positions so I think it should work.

Continues Below


The other challenge would be that certain printers behave differently at different Sprint speeds, and this may cause problems with printing of small parts or parts that can get overheated easily oh, such as wood fill (which will darken if heated more) or other similar color change filaments, and even certain pla materials that will over melt if the nozzle sits in one place. Some materials (like pearl/shiny pla) also swell or shrink at different rates depending on the pressure in the nozzle, so consistent volumetric flow rate is likely to produce a much better result of a part where an inconsistent flow rate may create problems in the final print quality.

It may also cause additional problems with firmware is that handle pressure advance and other path planning algorithms.

Continues below (again, curse the 1024 limit and trying to split on paragraphs )


Show 11 more comments

Add a comment

3 Answers

User avatar

I’ll take a WAG at this. Offer a wrong answer and someone will correct it right?

Measured data/assumptions:
Filament provided during the compression step is negligible.
P3P max drive speed: 200mm in 5s, or 40mm/s (raw 1.75mm filament speed)
P3P USABLE buffer length: ~70mm- Measured available filament in buffer from when splice core starts heating.
0,0,0 P3P splice time: 28s
Printer is continuously consuming filament, AKA no travel moves, no features that vary the flow rate.

1: Single splice: Buffer length and splice time-

The thought here, if there is a single splice with a constant (maximal) rate of filament used, there is a maximum filament the buffer can make up while the filament is paused. Bonus points if someone else factors in the filament fed while compression is happening gets accounted for.

This would just be the usable buffer(mm)/splice time(sec), or about 2.5mm/s. or about 4.4mm^3/s
This should be a lowerbound given the buffer will be fed during the compression and cooling steps.

2: Infinite splices: Minimum transition length and splice time-

Assuming there is a continuous series of splices, what is the fastest the P3P can push out filament. I’m assuming an infinite buffer here, so the speed limit of the P3P will be whichever limit is less

This should be: minimum (transition tower, 130mm default) splice length(mm)/(time to drive that length(sec) + splice time(s)). I get 130/(3.25+28) = 4.16mm/s or about 10mm^3/s

Looks like we’re mostly limited by the single splice rate, unless transitions are shorter than ~75mm (Where speed limit 2 becomes slower than speed limit 1). Ouch. I print my slowest features at more than 5mm^3/s. Challenge I see in a slicer script slowing the printer down during a splice would be knowing what the output tube length is to know where to insert the slow periods. But I don’t see any reason why the P3P in connected mode couldn’t drop the feed rate real time during the print when needed.

Please feel free to poke holes in my assumptions or approaches, and or offer your own data! Other tests I can run? I have a sh*t load of black PLA, might try multi-spool mode and just push plastic at known rates to see where the buffer runs out.

Staff- It looks like a little bit of the buffer goes unused? I get that the compression step needs some space to push filament into, but even using 0mm after the splice initiates the buffer isn’t full when the splice is done. Do you have any of your own data to confirm these speed estimates?

Was this answer helpful?

Score 0
Add a comment
User avatar

Test results!! Everyone likes data right?

TLDR for your average user: 0.4mm nozzle shouldn’t print faster than 150mm/s with “typical” settings printing PLA. See results for the proper volumetric machine limit I measured.

Set up:

Printing a large cylinder in vase mode. 0.2mm layers, 0.45mm extrusion width, constant tool head speed of 60mm/s. Goal is constant and controllable extrusion speed. Multi-spool mode KVP PLA spliced with 2,2,0

Block Image


I increased the feed rate multiplier until the P3P ran out of buffer and caused the extruder to skip. Got up to 250% feed rate, 270% caused skipping. Even showed the P3P do a ‘worst case’ filament run-out at a pong event at 250% feed rate which looked very tight to the end of the buffer, but didn’t skip. Worth noting I do have a very long output tube ~120mm, shorter output tubes may have slightly less forgiveness.


Machine limit is about 13.5mm^3/s, or 5.6mm/s filament speed.
Phew! I got worried in my previous answer that I was going to need to slow WAY down. Still need to drop my max volumetric flow rate to a third of what it was. Keep in mind this will change with splice settings. I will be setting my SuperSlicer profile for Pallet to 10mm^3/s to give some margin. My ABS profile (8,3,-2) should only take 1s longer to splice than this PLA test (2,2,0).
I believe my method for measuring minimum usable buffer length was incorrect in my previous answer which lead to a fair amount of error. Additionally my assumption to neglect the feed rate during compression is definitely unfair!

Again, feel free to critique my methods!

Was this answer helpful?

Score 0
Add a comment
User avatar

I just got my P3P, and wished that I had considered the volumetric flow rate. I have a large volume machine, with a 1mm nozzle. During my first test print, it was splicing while the machine was laying down a flat surface - twice. The buffer was exhausted, extruder started skipping, and it started compressing the PTFE tube. Result was sections of the flat top layer didn’t print at all. I came here to see if there is anything I could do to ease that out.

A volcano hotend can do upwards of 60mm3/s; so this is a major bummer to read.

I agree with some of the comments, I would hope Canvas could slow down the gcode that is running while a splice is happening. Or worst case, pause when the buffer is exhausted while splicing is still happening.

Was this answer helpful?

Score 0


I designed my own buffer.

I use a 1.2mm nozzle

Filament feed, Palette 3, and Buffer system.


Yeah, I've seen that video, it's very impressive. But I'll be frank, I don't want to invent/build a machine to make the device I bought work.

I'll slow down, but this should be a software fix, one way or the other.


I hear ya the palette 2s pro had a small buffer and the palette 3 isn’t to much bigger. The palette 3 buffer should be bigger. But want we want needs a crazy big buffer. Mine adds about 500mm of extra filament. That’s 20 inches! On top of that their are times when I still use up the palettes buffer and my buffer! Most of the time it comes petty close. I was printing slower in the video. Now I wish I designed a bigger buffer that had a 50mm longer rail for another 100mm of filament. If you think about it that’s a lot of extra filament. Palette 3 would have to somehow fit all that in that small package. It would be very hard to do. So even if the palettes buffer was bigger you will still run into the same problem. So I knew I had to design something. I believe they should sell something separate like my design for the big nozzle club.

The palette 3 splices pretty fast and I don’t think you can go to much faster. You have to heat up and cool off the filament.


Add a comment

Add your answer

Author avatar Larry Schack will be eternally grateful.
View Statistics:

Past 24 Hours: 1

Past 7 Days: 12

Past 30 Days: 48

All Time: 894

Contact Mosaic Support Team

© 2023 Mosaic Manufacturing Ltd.