H2D Printig supports without prime tower - problem

[ProtosOne]

Good morning guys, I am writing to share with you the problematic results of a test:

Fw: 01.01.01.00 - 17/04/2025

I am referring to the multi-material and double nozzle management.
Let’s start from the assumption that the H2D is certainly a printer that aims at a user with technical needs that perhaps for the first time are seriously considered in the consumer / prosumer field, the print volume, the heated chamber, steel nozzles… everything suggests the desire to be able to efficiently and functionally deal with the printing of technical and therefore expensive materials…

I tried to verify the case in which I want to use an expensive material for my model and a cheap material for ALL the support. The idea is to use a cheap material for the entire support and not just as a separation layer, in order to have as little waste as possible of expensive material.

I used a design cat as a test model (if needed, I’ll post a project) and I initially simulated 2 “normal” conditions:

CONDITION 1 - The cat printed with expensive material / supports printed with expensive material / no prime tower (not needed as it’s the only material)

CONDITION 2 - The cat printed with expensive material / supports printed with cheap material / prime tower active

The result between these two simulations was not comforting: due to the prime tower, the amount of expensive material used overall is excessive, equal to 50% of the expensive material used for the cat.
This is also due to the ratio between the measures of the prime tower (too big by default) and my model. I then tried to reduce the size of the prime tower, reduce the flush coefficients but there is some parameter that is currently uncontrollable that forces me to have that amount of material in the tower in any case. At the moment, therefore, if I want to use the tower and print a medium model, I have to accept the fact that 30% of the material used will be for the tower. BUT in reality it was not the focus of the test, the idea is to use the two nozzles to their full potential, without prime towers and not having to change filament:

CONDITION 3 – Cat printed with expensive material / supports with cheap material / no prime tower

Clearly it is the ideal solution to avoid waste and reduce times, but how will the printer behave in reality? Will it clean the nozzle from the poop that could possibly form on the hot nozzle while waiting? It has a dedicated camera with AI, why would it fail?

The simulations of weights and material times were obviously much better than the previous two and so I tried to launch the print…

I used black PETG as the “expensive” material (nozzle 2 - AMS 2 PRO) and white PLA as the “cheap” support material (nozzle 1 - external spool). The cat model is small, about 10 cm high, base diameter 5 cm. Since the model is small, the supports (automatic organic) are also small and this will be the focus of the discussion:

I ran into an annoying problem (a bug?), I think due to the retraction of nozzle 1 where in this case there is white PLA. This extruder is the one used to create supports and unfortunately in my test it does not do them.
It seems in fact that when the retraction occurs, just after the material change, this is excessive for nozzle 1 - let’s remember that in this case there is no purge tower so when it is the PLA’s turn the extruder immediately goes to print the support without any purge operation first.

Since the support is small and requires little material to be completed, the nozzle does not have “time” to fill (due to the large retraction) so nothing comes out, and the supports are therefore not created. It can only create the first layer because the brim certainly requires more material, but I clearly noticed how even during the first moments of printing the brim did not come out material, only to start coming out shortly after and finally start to deposit optimally.

This condition does not concern nozzle 2 where the black PETG is loaded. It remains well loaded and when it prints it immediately starts to deposit material creating the model in an optimal way.
In both nozzles if a material leak is detected before the printing turn there is a quick cleaning on the silicone pad (so excellent feature).

However, I believe that there is a problem in the “waiting” retraction of nozzle 1 which occurs too deeply. Nozzle 2 on the other hand has no problems with retraction being too high and in fact it is always ready for use.
Maybe if there was a prime tower, perhaps this problem wouldn’t exist since flushing the tower itself before printing would also allow nozzle 1 to fill properly and be ready, but this is just a guess since I refuse to have to create a tower when I have 2 ready nozzles available.

Below is a short video:

Bambulab H2D fail to print supports without prime tower

[Floppy525]

I have exactly the same problem.
Is there a solution so that the nozzle stays full even without a prime tower?

[3DTech]

I’ve noticed this as well, but I haven’t had the time to look into a G-code modification or whether the extra length on restart setting can be used to fine tuning this setting in the printer settings and extruder tabs.

If you get a chance, let me know what you discover.

This is what I do when using a prime tower with support interface filament you may need to fine tune it for your interface filament.

[hotellonely]

The prime tower serves its purpose. There’s simply no way that you can stop and hold the flowing material in the nozzle that long without it leaking or dripping. Not even a long retraction. Brushing the nozzle on the silicone brush doesn’t work well to remove small blobs on nozzle, which is why before every bed mesh, the nozzle would do its mopping move on the cleaning area of build plate.

What you can do instead is to manipulate the size of prime tower, like what @3DTech has shown above which is quite nice.

It’s just the limitation of FDM, or if you want, build a “cleaning pad” for H2D that does the same tip cleaning job, place it on the purge chute and avoid any collision… well that’s a lot of work to do Id just print a prime tower.

[ProtosOne]

I agree, it is not possible to imagine having the nozzle ready to print as if it had just finished a layer a few tenths of a second before.
Indeed, the prime tower does its job well.
At the time of creating the post it was not possible to customize the prime tower too much, it really required too much useless wasted material in any case. I will do a check now to evaluate if after some updates the measurements of the prime tower can be manipulated better.
It seems that at the moment the small models do not lend themselves well to material/color changes as there will always be a bit of wasted material on the prime tower… the secret therefore is to print large models!

[hotellonely]

This discussion does inspired me a bit about why not building a cleaning pad for nozzle on the purge chute. It should be doable with somewhat flexible materials and the lever mechanism it’s already using.

[ProtosOne]

However, I have a doubt, which can also be viewed in my short video… Why is the right nozzle with black PETG always ready even without prime tower, while the left nozzle with white PLA was not ready? ok that the two materials are different and could have different behaviors during the “waiting” but if the right nozzle with black petg is always ready it means that with careful tuning of temperatures and retractions you could have the nozzle always charged at the right time, without the need for prime towers.

[3DTech]

Just thinking out of the box it would be cool to have option for a dedicated support pad specifically for priming the interface layer, or even a tree branch with an interface priming pad. This way, you wouldn’t need a prime tower just for basic interface layer supports.

Due to certain PVA requirements, a ramming volumetric speed setting is needed.

[Qjo]

Just leave the prime tower there. Its purpose is to prime the nozzle so that you don’t get printing defects like the mentioned failed supports.

It’s a small price to pay to make sure the model you want printed comes out defect free.

[maventech]

I noticed the same issue.

On my X1E, I never used a prime tower and I always had great results.

I understand that the H2D has to deal with 2 print heads, but for me, the reason why I purchased this printer was to avoid spoiling material (the environment is important for me). Using a prime tower is not acceptable for me. I’m very surprised that Bambu Lab didn’t consider this when creating the H2D.

On my side, I needed to use 2 different material colors. When switching to use the other print head, some material gets out before moving back to the model actually being printed… Leaving a lot of spaghetti…

[Keybaudio]

Having similar frustrations (disabling towers leads to printing without filament on the second material). From what I can tell the reason is that after it purges, it looks like it may retract, and then seems to make no effort to de-retract? This is hidden by a prime tower, or the lead line (front right corner of the plate) for the first nozzle. The second nozzle doesn’t get a lead line, and with the prime tower disabled goes straight to printing… where it needs to attempt to push enough filament to overcome the retraction before you’ll see anything actually print.

I’m trying to read the machine g-code to confirm if this is actually what’s happening or not… if it is, I’d imagine that a larger retraction on that particular filament would make the problem worse, and that perhaps setting the retraction to zero would work around the issue, but likely cause other problems.

[Gerrit]

Why do you think there are prime towers?

What do you think of the quality of that half cooked filament?

You still have the possibility to prime in the infill, but that might have negative impact on the structural strength of your part.

[Keybaudio]

Not debating that, I’d rather have prime towers enabled too. But even with prime towers, I do think there should be a de-retract that occurs before printing the tower. I have a reason to disable it for a particular situation that isn’t really relevant to what’s happening.

[Gerrit]

It could be a feature request towards Bambu Labs.

You might also be trying to play with the “Multi Filament” settings in the Filament profile, if you haven’t done so.

[Keybaudio]

I’ll continue with more tests tomorrow and see if there is anything useful in the multi-filament settings, but I tried setting the retraction to zero on the second filament and it didn’t seem to resolve the issue. I also changed the printer profile retraction settings for left and right nozzle to zero on material change. That didn’t help either.

The retraction looks like it might be part of the purge wiping cycle, right before the fan kicks on and it jolts back and forth to drop the filament down the chute. It’s strange that one material on one nozzle will print without this gap in filament, but the other will not.

To be clear though, this isn’t after the material has been sitting idle in the second nozzle for any period of time, it goes directly from purge and wipe to printing, but has a retraction or filament gap that stops it from laying lines until it overcomes it. The prime tower just hides that, but it looks like it doesn’t print correctly either.

I think the simplest repro is to create two flat shapes side-by-side with two different filaments, one on each nozzle, and then disable the prime tower. Watch the first layer of each print. Both will purge and wipe, but one will fail to extrude plastic at first, the other seems fine.

[Keybaudio]

This is what the minimum repro looks like. Two cubes, resized to only twice the layer height of a 0.6mm nozzle (two layers of a 0.15mm profile = 0.3mm tall). One printing in black PETG on the first nozzle, transparent PETG on the second nozzle. The black prints the first layer without issues, the first layer of the transparent cube misses the corner. The second layer of transparent prints fine (because it’s a continuation of the first layer without a swap), but the second layer of black misses a corner after the swap.

I can see how the behavior makes sense if the pressure is backed off after the purge, and between nozzle swaps… but it does seem that without the prime tower, there isn’t enough of an attempt to do anything about that with default options. First layer of black is fine because the lead line at the front of the plate will effectively normalize the pressure.

So in conclusion, I don’t think it’s wrong as such, but it could be better and perhaps there are options that I can noodle to improve this edge case. If you are printing something with supports or infill, then priming into those might solve this too… my particular reason for disabling the prime tower doesn’t really allow for that (transparent, thin film, solid prints).

[Keybaudio]

Well, below is my solution, but your mileage may vary and use only at your own risk. From what I can tell, there is 2mm of pressure missing post nozzle swap. There are no good options to resolve this based on my diffs of gcode produced from various tweaks and the machine material change retraction settings don’t fix it either. What did work was to modify the filament start gcode (for all filaments used) to insert the missing pressure. Since this isn’t required for the very first filament after the lead line, I also inserted a retraction to cancel it out at the end of the machine start gcode… the two fragments look like this:

; filament start gcode - preprimed for no tower
{if (!enable_prime_tower)}
    G90 ; absolute coords
    G21 ; mm units
    M83 ; relative for extrusion
    G1 E2 F1800 ; 2mm of extrusion/pressure
{endif}

; VERY END of machine START gcode - retracted to cancel first filament extrusion
{if (!enable_prime_tower)}
    G90 ; absolute coords
    G21 ; mm units
    M83 ; relative for extrusion
    G1 E-2 F2400 ; 2mm retraction
{endif}

I’d appreciate @SupportAssistant weighing in on this, because the 2mm feels awfully suspicious given default options for 2mm elsewhere. As you can see from my pictures, it is a dramatic improvement, but it doesn’t fix uneven pressure issues at the start of the new material - that DOES require a prime tower and I don’t think there is any escaping that.

[hotellonely]

@Keybaudio

it looks like it may retract, and then seems to make no effort to de-retract?

You would get a blob or under extrusions if you do a deretract. The actual deretraction length is incredibly hard to fix down for different filaments, so whatever we do, a prime tower is needed.

[Keybaudio]

Tested a few materials, and in my situation at least, there is always 2mm of missing pressure post swap (on 0.6mm nozzles at least). There is no attempt to resolve this from what I can tell, besides printing the prime tower - but even that misses 2mm worth of plastic printing so my prime towers are pretty spotty without this change too. This change didn’t lead to more blobs or issues so far, it just closed the gap of missing plastic between material changes. I’m doing a six color print now to really stress test this.

Edit/Update: The sweet spot does vary by material, some of the high flow materials require a little less by the looks of it… but I think if you want to put the time in to calibrate, this can avoid a prime tower under certain conditions. At least, it works for this awkward project. Ideally, it seems like it would be nice to be able to purge into two towers that are in the unshared space to the sides of the plate. Losing the overlapped space to a large prime tower is unfortunate.

[hotellonely]

I would suggest you to submit a issue on Github as well, and see what they say about it: GitHub · Where software is built

From my understanding the 2mm retraction is necessary for prevent dripping/oozing from long travel moves. After a long travel deretraction of 2mm might be too much. Without a prime tower you would not want such a long retraction, I think they need to turn it down a bit. With a prime tower, if you add a deretraction here, chances are for some filaments you would have lots of build up:

These blobs and build ups might cause the prime tower to fail and get knocked over.

https://wiki.bambulab.com/en/software/bambu-studio/parameter/prime-tower

Better have a detailed read here, it explains a lot of concepts behind the current design of prime tower.