I have an H2C with 6 AMS HT’s feeding to the right Nozzle(s) through 2 x 4 to 1 PTFE Adapters. It would be really nice if the user could specify how far to back up the filament (Just to the down side of the 2 x 4 to 1 PTFE Adapters). This would save time for loading/unloading filament, not to mention the wear and tear on the ASM HTs. The only change would seem to be to the firmware.
3 Likes
I asked this too but experts claim its not that easy and stopping and starting filament inside a tube it will give issues so that trick that seems logical is not that easy. They can probably design a different ams/system that feeds things closer to the back of the printer but not just stopping/starting mid tube.
2 Likes
The new ams units can measure how much is left on the roll, based on how much filament is fed per rotation of the roll. They could use that data to do this(means the ams unitbis capable of measuring filament). The ams calibration where it figures out which ams goes to which nozzle, by sending one filament from each ams to the toolhead, could factor the same thing in. All they would really need to know is the minimum length of ptfe tube between the buffer and runout sensor. Which is a known measurement. Even if it wasnt, the ams could measure the filament thats sent after the buffer has been actuated and before the runout sensor is activated.
If any of that makes sense. Im tired. The crappy part is that this probably wont come because bambu has spread themselves thin with all the releases back to back. I mean, we are still waiting on stuff from nearly a year ago and they still have the x1c and bedslingers to refresh.
As long as the filament doesnt sit in the tube between the buffer and nozzle, it wont break while waiting. No matter what, they are limited by this because of the nature of a dual nozzle single tool printer
1 Like
A setting in bambu studio where the user can define the length to retract the filament back to the AMS-HT would be nice. Then we could set it to the exact length it would need to pull back to stop just after it passes the inlet of the 4-1 adapter. Having a setting like this doesn’t really align with Bambu’s “anyone can do it” ideology as having the wrong length setting could lead to feed failures, but they can easily hide it behind develop mode for us advanced users. This would also lighten up the load on the dev team as they wouldn’t have to make the feature automatic.
2 Likes
While I understand the desire to reduce loading/unloading time, this approach introduces significant reliability and maintenance issues that make it unsuitable for implementation.
The fundamental problem is that you’re proposing to leave filament parked in an extended PTFE tube section between the adapter and the AMS HT unit. This creates several critical failure modes:
Reliability concerns:
- Filament left stationary in tubes is prone to binding, especially hygroscopic materials that can absorb moisture and expand
- Without controlled environment storage in the AMS HT unit, the filament becomes exposed to ambient conditions in the tube
- Temperature cycling in the enclosure causes filament expansion/contraction, leading to buckling or binding in the tube
- The system loses the ability to accurately track filament position, as the sensor is in the HT unit, not at your proposed stopping point
- There’s no tension control mechanism to maintain filament position between the adapter and HT unit
Why the current full-retraction method exists:
The existing behavior of retracting filament completely back to the AMS HT units is intentional—it ensures filament is stored in a controlled environment with known geometry and proper sensor monitoring. The filament position remains deterministic and verifiable.
The correct solution:
If loading/unloading time is your primary concern, the proper engineering approach is to minimize the physical path length by rearranging your HT units closer to the printer. Using 2×4-to-1 adapters with 6 HT units feeding the right nozzle creates an unnecessarily complex topology that works against the system’s design.
Consider instead:
- Positioning HT units to minimize tube length to the nozzles being used
- Using appropriate adapter configurations that reduce cumulative path length
- Accepting that reliable material handling inherently requires complete filament retraction cycles
Firmware cannot compensate for geometry and physics. The solution you’re requesting would create more problems than it solves.
I do see your points, but there’s a few things that don’t make it seem as impossible as you’re saying.
1.) the ams lite already does this partial retraction and leaves filament in the tubes. Why would it be any different here?
2.) wouldn’t the ams ht just use the same method of tension it uses now? It would just stop early at a predefined point. There are many possible ways to determine a stopping point, it’s just a matter of Bambu developing the process of reading where the filament is. The buffer could be used, or just distance from the extruder. A manual calibration and set seems like the easiest option here. Filament loads to toolhead, screen prompts you to retract 10mm at a time until you reach desired stopping point, press save.
I do agree that leaving filament in the tubes for extended periods of time will lead to issues. I’ve experienced it before with pla. Left it in the tube for 2 days and it was all cracked and broken throughout. During a print shouldn’t be an issue though and then it’ll retract completely when done.
I don’t think it’s impossible, and it may be unnecessary for those of us who already have the shortest possible tube lengths. For those who have their ams at length though it could be quite useful in reducing overall print times on multicolor prints.
Good points—let me address them specifically:
1. AMS Lite comparison:
The AMS Lite operates fundamentally differently. It uses a direct, short, fixed-length path with the buffer mechanism designed specifically for that geometry. The key difference is that the Lite’s buffer location and tube geometry are known constants in the system design.
In your proposed setup with 2×4-to-1 adapters, you’re introducing variable-length PTFE sections with unknown geometry. The system has no way to know:
- The exact length from adapter to HT unit
- Whether the tube routing has any bends or tension points
- If there’s slack or tension in the unsupported tube section
The Lite works because its geometry is constrained and predictable. Your setup isn’t.
2. Tension and stopping point:
You’re correct that the HT unit could theoretically stop at a defined point. The problem isn’t the stopping mechanism—it’s maintaining filament position after stopping.
Consider what happens when you park filament in an extended tube:
- The HT unit’s tension sensor only measures at its location, not at the adapter
- Any slack or binding in the tube section isn’t detected
- Temperature changes cause the PTFE tube itself to expand/contract (~0.1mm per meter per 10°C)
- The filament also expands/contracts at a different rate than PTFE
- Without continuous tension monitoring at the stopping point, the system can’t verify the filament hasn’t shifted
Manual calibration wouldn’t solve this—you’d be calibrating for one specific thermal state, and that position would drift during printing as the enclosure temperature cycles.
3. “During print shouldn’t be an issue”:
This is actually where the biggest problems occur. During a multicolor print:
- The enclosure temperature rises and stabilizes
- Parked filament in tubes undergoes thermal expansion
- Hygroscopic filaments (PLA, PETG, Nylon) continue absorbing moisture from air in the tube
- Each load/unload cycle introduces potential for buckling in the unsupported tube section
- Over a long print (8+ hours), these effects compound
The issue isn’t leaving it for 2 days between prints—it’s the accumulation of thermal cycles and load/unload events during a single multicolor print job.
The real question: Why optimize for the wrong problem?
If load/unload time is genuinely impacting your workflow, you should be measuring:
- How much time does full retraction actually add per color change?
- What’s the total impact across a typical print?
- Would rearranging HT units to reduce tube length save more time?
I suspect the time savings from partial retraction would be minimal compared to the reliability issues it introduces. The engineering effort would be better spent on path optimization rather than trying to make firmware compensate for suboptimal geometry.
1 Like
I think the simplest solution would be the same kind of filament hub that is used on P1/X1 series printers that would go right next the the entry of the printer, with the addition of one filament sensor per channel and only retract filament up to that sensor. At the end of the print, all filaments get retracted in the AMSs.
Of course there will be full retraction when several different filaments are used from the same AMS but when you have multiple AMSs and you plan right (BS will do that) it can save a lot of time with small change in the firmware.
Thing is, I’m sure bambu already thought about it but if it doesn’t exist yet, might be some other issues in implementing it. It might need sort of an extruder inside to pull the filament and just not rely on the AMS to push it…
1 Like
This will be I suspect will be the job of the new device coming that can multiplex an AMS to either hotend but I have no information to back that up. But hopefully it will be able to tell the ams to stop