Major interlayer strength differences when multi-material printing

I have been printing production parts daily from PC, PA-CF, and PAHT-CF with our two X1C’s and AMS’s for over 6 months now. It has become clear that the mechanical properties of parts, specifically the interlayer strength, reduces dramatically in layers where a material change has occurred. When printing a single material I am able to reach the material properties quoted in the Bambu filament tech sheets but fall far short (it is difficult to quantify because failures tend to move from ductile type material failures to brittle interlayer failures) when using a second material for clean support removal.

I have conducted a large number of impact tests on standard test coupons and both the failure mode and energy to failure is significantly different when the printer has changed materials during layers or just continuously printed. I believe there are a number of reasons for this, chief among which is the additional cooling that happens to the previous layer before it is printed on again. I have tried almost everything to speed up layer changes including implementing many of the suggestions in this thread (Ideas on boosting speed for multi-color printing with AMS) an have managed to get the load and unload times down to 6 and 8 seconds respectively with the entire material change including purge being under 30 seconds print-to-print. I have also played with both bed and nozzle temperatures as well as active enclosure heating.

I would be interested to hear if anyone has managed to overcome this challenge? I have tried annealing at a range of temperatures and times and, although material properties are indeed improved, the underlying issue is still there. My next plan is to implement a focused infrared light which shines on the previous layer while the material change is happening with the hope that this decreases the temperature delta between printed layer and substrate. To be clear, the printer is doing its job perfectly, the parts look great and support comes off super easily, I just believe there is an inherent problem with trying to print really mechanically strong parts with this style of single print head material changing printer.

I thought I would just post this here so that others are aware and maybe someone has come up with a clever solution for this problem. I certainly wasn’t aware when I made the decision to use this machine for production.

2 Likes

I would suggest increasing your purging when going from support material to the material your printing. If all the support doesnt get purged out it would make the layers weaker.

I remember reading a thread recently where the use of PLA to support PETG and vice-versa was discussed. Several users found that parts ended up weak at the layer where the filament change happened and there was discussion about contamination being left in the nozzle and affecting the subsequent printing. Interestingly, the observation was that using PETG to support PLA works but using PLA to support PETG doesn’t fare well 'cause the PETG ends up weak at the filament change layer. IIUC PETG needs the previous layer to remain warm in order to fuse well, so now I wonder whether the issue there wasn’t contamination but the same you observe here, which is too much cooling of the previous PETG layer…

Something you could test is to construct a print using all PAHT-CF with an auxiliary structure (similar to the prime tower but perhaps larger) that has a time-consuming layer somewhere to simulate the filament change without actually changing filament. E.g. minimal infill for almost all the tower but one layer solid and tweaked to print at the lowest possible speed so it takes about the same amount of time as your filament change. If that produces the same weakness you have your proof… (I can also understand that you may not need any further proof at this time…)

Another idea: would it be possible to perform an ironing pass over the layer previous to the filament change to reheat it? Dunno whether ironing can accomplish that, I’m just wondering…

Yet another idea: observe the print with a thermal camera to verify the hypothesis…

1 Like

You could just print a small test object and hit Pause for a couple of minutes to see if it really is time causing this issue.

I have some doubts about this as it would cause large and complicated parts with minute-long layer times to be quite weak as well. I have not printed such parts in PETG yet but I have never heard complaints about such issue. On slower printers (basically almost every printer to date except Vorons, including Prusas) this would be even more pronounced and, therefore, probably be common knowledge by now. Hence, I tend to believe that it rather is contamination by the support material. If you have ever printed a white part after, say, a red one, you probably know that it takes waaaaaay more purging than the printer uses by default to get rid of the last bit of rose shade = contamination.

Yes, purging volumes are an interesting one. I have not tested a range of purge volumes but I have tested the extremes. From what I can tell (and this seems to be the general consensus online on other forums and with other users: Do Flushing Volumes Matter on the X1-Carbon? - YouTube and Reddit - Dive into anything) the standard “auto-calc” Bambu flushing volumes are very conservative. It seems from my testing at least that you only start running into trouble when you are going well below 0.5 of the recommended flushing volume. In terms of interlayer strength I didn’t see any difference between 0.5recommended and an extreme, 2recommended flushing volume. In order to further reduce the effect of contamination I am printing the infill first and then the walls. Since each of my layers takes around 30s of printing I’m pretty confident there is no nozzle contamination.

I actually tried this by implementing a “G4 S60” into my G-code at every layer and then printing my part with a single material. This pauses for 60 seconds at each layer. It is not really a fair test since the print head pauses over the part wherever it ends and it also pauses on layers that wouldn’t otherwise have a material change so it makes your print suuuuuuper long. The prints also did not come out very well and although they were weaker I don’t think it was a fair test, I’ll try to re-do it with some improvements.

As for the reason other, slower printers don’t have this issue, this is a really good question and it has been on my mind throughout this whole process. We have a number of Prusa’s (MK3S+) as well and have been able to compare the strength of parts between the Prusas and the Bambus. When we print with no material changes the strength of the two is quite comparable (printing with Prusa PC blend). In fact the Bambu seems to produce slightly stronger parts, all things equal except for print speed. The Bambu prints the part in about half the time of the Prusa (makes sense because it is moving about double as fast).

When we introduce material changes the total print time (for this specific part) is about the same between the Bambu and the Prusa. What is important to note is that the Bambu is changing materials about 100 times in this specific 400 layer print, and that takes the total print time from half of the Prusa time to the same time as the Prusa. This means that in those material change layers the layer time is a lot longer than the Prusa layer time and the Bambu “catches up” in the layers that it does not change material. All of the interlayer failures we get are in the layers where material changes occur. Basically what I think is happening is that there is a drop off point in the graph of layer time vs strength. I think most printers are running just under that drop off point and when you start to change layers you are on the other side of the drop off. For example the Prusa is printing layers around 30 - 40s a layer for this print, the bambu is printing around 15-20s a layer with no material change and about 60-70s a layer with a change.

One other point to consider is the radiative heat transfer from the super hot nozzle and molten filament to the substrate near the area where you are printing. If you print really slowly you actually re-heat the substrate right before and while you are printing on it, if your nozzle moves really fast you have far less time to heat both the material coming out of the nozzle and the substrate. Although not exactly the same scale this paper (Infrared preheating to improve interlayer strength of big area additive manufacturing (BAAM) components - ScienceDirect) has quite a concise explanation of the problem with some indicative test results. (Also the inspiration for trying IR heating on the X1C)

Depending on how much flexibility you have with the designs could you perhaps take advantage of the fact that at least one colour or material tends to span two layers.

If material changes are moved as close as possible to the middle of the print time of each layer then almost half of each layer should be well bonded to the previous one.

1 Like

I think I am understanding you correctly, and as far as I am aware the slicer already does something along these lines. The material change is not always performed at the beginning or end of a layer, in fact the material change occurs sometime during each layer. As far as I understand (and I may well be wrong) the slicer has the following behaviour:

So lets say you have two materials A and B both take up 50% of every layer of your part. The slicer will start on layer 1 with material A and then half way through layer 1 switch to material B, it will then complete layer 1 and when it starts on layer 2 it will start with material B (no material change) and then switch to material A and complete layer 2 (for material A it has waited 2 material changes and the time to print layer B). For layer 3 it will start with material A (no material change) and switch half way to material B. The time that a layer cools is not constant when you are changing materials, for one pair of layers it has very little time to cool (since there is no material change and it ends one layer and immediately starts the next with the same material) and then the next pair of layers it waits a really long time (since it essentially waits for two material changes and the print time of the other material). The result is that you actually get alternating strong/weak layers in the region where material changes are happening since you alternate starting and finishing layers with the two different materials. This also means that your total layer time is not really a good metric to use to figure out how long the material is cooling on the “slow” layers.

You don’t say here anywhere what filament you are using for support/support interface?

I’ve run piles of tests for this problem when printing with PETG or ASA and XYPolyer Breakaway-80. In my case it was not the layer time but the insufficient purge of the support material that was causing the issue. The way I tested is to lie to the AMS and replace the support filament with whatever filament you are using (PETG/ASA in my case). This ensures the layer times are exactly the same. Obviously the supports are a PITA to remove from this part.

The filament flush calcs are entirely based on COLOR, not material type so are wholly wrong for the support filaments. I had to set flushing volume to 1,450mm³ to retain the similar layer adhesion as the print with all one material. Bambu calculates a flushing volume of 420mm³. Due to this I only use it as a support interface on flat horizontal sections where I need a nice surface.

In my opinion you need an IDEX printer to use support filaments for any type of production printing. I’d kill for an X1C with dual heads and 2 AMSs. You could print with 8 different filaments with virtually no pause when switching.

Interesting, I will definitely run this test, that seems like a crazy amount of purging but maybe I am totally missing the mark here. My logic was that purge volumes required to change colour (especially drastically different colours) would equate to purge volumes to change material (support vs part) maybe my “extreme” purge of 2x"flushing volume" was not actually enough. I struggle to see how this is true though but I definitely could be wrong.

In my case I am printing PAHT-CF parts with PC support. Interestingly marking the filaments as “support” or “soluble” changes the automatically calculated flushing volumes so there is some logic other than filament colour that Bambu applies to flushing volumes.

I totally agree IDEX is the way to go for strong production parts with easy support removal.

I’m using these knockoff CHT nozzles that have an insert in the filament path which probably has more nooks and crannies for filament to hide out in? The whole experience was disheartening and I do everything to avoid supports now. Please keep us updated if you figure it out!

CHTNozzle

Yes I agree that between individual materials the cooling time will be inconsistent across layers - but my thinking was that a long layer cooling time between material A layers would be balanced out by a short cooling time between material B layers. And if material A and B were close to each other, or potentially overlapped somewhat then overall you might get some increased strength.

However re-reading this thread I think the objective for having two materials may be for one of them to be removable support. In which case my suggestion would not work - because only one of the materials with the inconsistent layer cooling would be left in the final part.

If that is correct then an alternative suggestion would be for the AMS to be loaded with two rolls of the final target filament, and for if possible the layers with support material in them to be arranged in such a way that the support interface material is created mid layer.

I.e. where final materials are f1 and f2, and support material if s.

Layer 1 - start with f1, switch to s mid layer, then switch to f2 to complete the layer and half of layer 2.

Layer 2 - continue with f2, switch to s, then switch back to f1 for the 2nd half of the layer and first half of layer 3.

Layers 3 - f1, s, f2

Whilst this approach is wasteful in terms of requiring extra switches - these would only be needed on layers with support interfaces.

Plus those support interface layers would only have half of the f1/f2 material having a short time interval to the above or below layer, when compared to layers with no support interface in them.

I agree however also with the suggestion above about dual AMS idex being the perfect solution for the next Bambu lab printer.

What I would like to see is a slightly deeper printer - with longer metal Y rails, which then have two separate sets of carbon fibre X rails mounted on them with one extruder per X rail. Would probably need a purge bucket at the front for one and the back for the other.

1 Like

So I “lied” to the AMS and printed a part with material changes but changing to and from the same material on two different spools, this was an excellent suggestion because it really isolated the purging variable. From my single test so far the strength definitely increased significantly! So if anyone is having the same issue I would strongly suggest trying this test to isolate the “contamination” variable. In saying this the part is still weaker than the non-material change part.

Basically (bear in mind this was just a single test part) I think I was overestimating the effect of the previous layer cooling and underestimating the effect of purging and material contamination. Both seem to definitely have an effect but maybe not in the same proportion as initially thought. I am now running a test with dissimilar materials with the max flushing volume possible (999mm^3, for some reason my slicer doesn’t allow more.)

It would be great to hear from someone at Bambu as to their logic with flushing volumes, if it is the case that their flushing volumes are far too low (for strength and not colour) plenty of people are going to be struggling with incredibly weak parts when using expensive engineering materials.

While not scientific, I did have issues using auto calculated flushing volumes when using PLA and the W support material. The layer bonding at the auto calculated volume was extremely poor, many times the part broke while removing the supports. I changed the multiplier to 2X and the problem seems to be resolved.

Curious, have you tried printing too hot? That’s a little difficult with PAHT CF as the printers are bumping right up against the limit, but that 10 degrees still may help. The idea being, the cooling is going to happen regardless, but printing the material as high as it allows may gain back some of what it lost when reattempting the bond. Bridging and overhangs aren’t an issue as they normally would be because of the supports. Stringing will still be a concern, but I find that not so much of an issue for most of my prints.

Just something to consider.

Yes, good suggestion. I am printing as hot as possible both on the nozzle and bed side, and using a home made enclosure heater to keep the enclosure temperature right at the upper limit before the printer starts having issues. In my strength tests I definitely saw an increase in interlayer strength with an increase in nozzle temperature.

I still had no improvement at 2x but I just finished printing and testing a part at 4x flush volume (bumping up to the max purge volume allowed by the slicer) and the strength was definitely improved. I am starting to suspect that materials with an additive (carbon or glass fibre) are much more difficult to purge than straight homogenous materials. I am implementing some custom G-code in the purge step to purge at the max nozzle temp as well as the max volumetric flow rate of the material to try and get a more effective purge. One thing to note is that the machine can’t handle this amount of purging effectively and the poop-chute is continuously getting clogged (not to mention the enormous amount of material waste)

I was even considering trying to do some kind of cold pull in the g-code for the support interface filament change. This has worked quite a few times for me when the nozzle is clogged so may help remove more material? Never got around to testing as I just used max flush volume and it worked for that particular problem and I moved on.

FYI
You can type in whatever number you want in the purge table instead of using the multiplier. It maxes at 999mm³ though as you have observed.

You’re going to have it because of the inherent cooling that occurs.
Come back later and lay new plastic down, previous is cooled beyond majority bonding. Slinky time…
So if it’s critical, I would suggest perhaps slowing printing down overall, or keep the chamber as hot as possible without causing a goop-session, or perhaps you might try a heat gun over the print slightly while it’s changing.
I’m finding the fast printing is nice, but the bonding just doesn’t happen. So it’s gotta be a compounded scenario for multi-changes.

First of all, thanks to everyone for an interesting discussion. I am at some point going to have to dive into some of the more exotic filaments so it is very helpful.

I wonder if approaching the problem sideways might help. Instead of using support filament, design the support as a component of the part that is easy to remove with minimal processing, then print in a single material.