Community Tech Talk: Deep Dive into Auto Flow Dynamics Calibration

Announcements
1

Auto Flow Dynamics Calibration is a signature feature of Bambu Lab printers. This system calibrates both the filament and printer conditions to establish precise parameters for Pressure Advance—a critical factor in maintaining consistent extrusion during the rapid speed transitions inherent in high-speed printing.

As we continue to refine our algorithms and improve the user experience, we are seeking detailed feedback from the community. We want to know how this feature is performing in your specific real-world use cases to help us drive further innovation.

How Auto Flow Dynamics Calibration Enhances Print Quality

By enabling automatic calibration in Bambu Studio, the printer executes a specialized sequence during the print preparation phase. This process generates the most precise, real-time parameters for the Dynamic Flow Compensation (Pressure Advance) algorithm, with noticeable improvements in print quality at transitions between steep overhangs and non-overhang areas.

This calibration is essential because it allows the printer to precisely counter the internal pressure lag that occurs during acceleration and deceleration:

  • During Acceleration: The algorithm compensates for pressure lag by slightly increasing the extrusion rate, ensuring the nozzle reaches the target pressure immediately.

  • During Deceleration: The system reduces or reverses flow—acting as a localized “mini-retraction”—to bleed off excess pressure. This effectively prevents over-extrusion and oozing at the end of a line.

Enabling Auto Flow Dynamics Calibration helps achieve cleaner corners, sharper details, and more consistent extrusion, especially in complex prints with frequent speed changes.

pa_before
pa_before770×545 421 KB

pa_after
pa_after771×545 358 KB

We are seeking deep technical dives into its real-world performance. Specifically, here is what we are looking for

  • Technical Experience

How has Auto Flow Dynamics Calibration performed across your general printing workflows? Please describe your test conditions such as the types of models and filaments.

  • Comparative Analysis

Results and observations from tests conducted on the same model with Auto Flow Dynamics Calibration enabled and disabled under the same conditions.

  • Observations & Root Cause Analysis

A description of any specific artifacts or improvements observed, along with your analysis of the underlying causes.

  • Optimization & Limitations

Identification of scenarios where Auto Flow Dynamics Calibration could be further optimized, as well as any situations where its effectiveness may be limited.

  • Data & Documentation

Regardless of the angle of your in-depth analysis, we encourage you to provide detailed information in your comment, supported by photos that illustrate your conclusions.

Why Talk Shop With Us?

This marks our very first Community Tech Talk. We want this to be more than just a presentation—it’s the beginning of a continuous journey where we dive deep into the tech together with our community. Our R&D team will carefully read every comment to better understand how you’re using this feature and what you’re looking for. From there, we’ll also be inviting experienced users to join our Beta Tester Program, where you will have the opportunity to test unreleased Bambu Lab products and provide feedback directly to our R&D team.

If you enjoy digging into the details, testing features, and sharing real insights, we’d love to have you join the conversation. See you in the comments!

23 Likes
Struggling with ASA & possibly a Studio bug
2

Well, i personally think this is a great idea! I would love to comment on this feature, but unfortunately, I have a p1s, so i cant. maybe i will on a future one

1 Like
3

What a great initiative :heart:

Having started my career with rapid prototyping, I embarked on home FDM as a hobby back in 2017 using a Tevo Tarantula. I gave that up in 2022 due to … well, that printer was … even less than suboptimal.

Auto-calibration experiences, notable observations

In 2023, I read about the X1. In particular its Auto-calibration feature was a reason for me to resume the hobby. In the 1,700h that printer lasted, I relied heavily on the Auto-calibration. In particular when PETG Basic was released, it helped me greatly to finally dial it in to print my Avatar. It still took 6 weeks due to the less than ideal filament profiles at the time, but I would have not had the stamina to see it through if the LIDAR hadn’t done the auto calibration for me.

Shortly after that, my X1 started dying over a period of about 6 weeks. The LIDAR was the first thing to go. I had no chance whatsoever of achieving the same results using the manual calibration. Others here on the forum have reported great success with the manual calibration, but I never got there without the LIDAR.

With the original X1 replaced under warranty in 2024, it has now racked up 5,000h with the auto-calibration (and a few prints without when I didn’t clean the lens properly :joy: ). PLA, PETG, PA-CF.

Regarding observations, there are of course a number of threads here in the forum on difficulties when using auto calibration with different nozzle sizes. I too did experience these issues, but since they are mainly related to larger diameters and (at the time) there was no significant time saving from the larger nozzles for the X1, I was able to mainly stick to the 0.2 and 0.4mm nozzles. In particular since the rare cases where I want to use a 0.6 line width, I can still do that with a 0.4mm nozzle.

In parallel to the X1, I also use a Qidi Plus 4. While that is claimed to not strictly require pressure advance (I presume that claim is linked to a narrower heat chamber, giving a more direct nozzle response), I have never achieved the same kind of results with manual calibration as my X1 does with the LIDAR. Pre-H2, it did have the advantage of allowing me to dive into PPS-CF and even PEKK-CF though.

With my recently arrived H2C, I have only started printing recently (300h). What I did notice however is that when used in conjunction with “Wipe on Retract” on PLA, it does generate some seam gaps which the X1 was free of. I have not dived deeper though as I just chucked the PrintU trials (problems with BVOH support interface) and just disabled “Wipe on Retract”. So I am not certain if that was a one-off (you state a relatively large jitter of 10% on the calibration page) or a genuine system difference.

Optimization

Although over 20 years ago, I did get my education in Active Control of Vibrations and have been wondering for a long time if it wouldn’t make sense to use the system optimization approach from feedforward active vibration control systems to determine a FDM “cost” function in order to derive the optimum print setting vector as well as studying its robustness.*

Flow Dynamics may be an interesting case of reasonable scope. Especially if we look at the H2C’s flexibility and sensorization enabling the quick generation of significant standardized data sets.

Beta Tester Program

I’d be honored to join if you decide to go ahead with that. Family and in particular work will be more demanding in 2026 than in 2025, but I do enjoy initially puzzling troubleshoting, root cause analysis and in particular scientific discussions.

:four_leaf_clover: & :crossed_fingers:

*Note: For interested readers: This is old school slang. Nowadays, we’d just say “we make an AI”. :joy:

6 Likes
4

I’m copy and pasting my previous Reddit comment on the same post:

I think first of all, Bambu should expose more technical details about how the “Flow Dynamics” actually works instead of repeating what’s already publicly available.

  1. Why Bambu insists calling it Flow Dynamics instead of Pressure Advance?
    Is there any technical difference in the details? If there is any difference, then fine we should keep this name. Otherwise we should change the translation and keep it aligned with what most of the community is using for years.

  2. I don’t know if the current Flow Dynamics algorithm can accustom to different printing speeds (Dynamic PA) . Is it Static PA or Dynamic PA? Or somewhere in between?
    For the static PA (a fixed K value measured before each print); Traditionally we have to measure the K value, based on a certain printing speed. For the dynamic PA, a realtime K value that is adjusted on-the-fly during the entire printing session, or a set of K values were measured before the print and builds up an algorithm or lut for use in printing.
    Now with the Auto Flow Calibration, is it Dynamic, or if it was Static, at what printing speed or volumetric flow rate was it targeting for?

About Static vs Dynamic PA:
For the traditional pressure advance, there is no dynamic compensation, until recent developments gave us dynamic PA.

About Realtime Measurement:
Bambu’s use of wording is quite confusing on this. In various ads we see wordings like “realtime”, does that mean the printer can continuously monitor the nozzle pressure and adjust PA / K value on the fly in the real time during the print (instead of just measured before each print)? Or it’s just doing a K value test at the beginning of each print, then use the same K value across the entire print, regardless of the lines plotted being 300mm/s or 60mm/s? It would be great if it can adjust it on the fly, but it’s also understandable if it’s not. These details are key to understanding what Flow Dynamics algorithm actually does. Just some transparency on this would be very helpful.

11 Likes
5

It would be useful to be able to do other Auto Flow Dynamics Calibration during the print, particularly the very long ones.

We could setup how many calibrations based on the lenght of filament printed and if there’s a difference, how to spread the new value.

1 Like
6

A question comes to mind, due to some bad recent experience…if you do multi colour print and the 2nd colour comes late into the arena, will there really this auto calibration? The start procedure just print the first colour in the prep process and then nothing.

6 Likes
7

yes it does. for each color.

8

Will you be sharing your conclusions/ updates with the other ie original developers of the software, or is this just something for Bambu to build higher walls with?

2 Likes
9

"It’s just a suggestion, but I think a dynamic pressure advance sensor would have one huge advantage. Right now, as I understand it, you get just one median value that is applied to every speed. But what if you could implement dynamic pressure advance, similar to what OrcaSlicer has: adaptive pressure advance calib · OrcaSlicer/OrcaSlicer Wiki · GitHub

The idea is to calibrate a wide range of acceleration and deceleration from the print itself to get PA values for each speed profile—for example, one value for the outer wall, a second for the inner wall, and others for slow bridges or top surfaces. This would give us much more tailored and higher-quality prints. It would also provide an advantage over competitors who have already copied the basic automatic PA feature."

2 Likes
10

Would we not see different filament strips at the front of the printer if that was the case?

1 Like
11

I don’t have any real specifics to offer, but I’ve observed my H2C isn’t quite doing as well as I’d have expected. Flow always seems to calibrate beautifully, but PA tends to be a little off. Haven’t done any kind of experimentation to understand it better, just an observation at this point.

12

Manuale says filaments are less likely to go past 0.02 PA.

Auto calibration gives 0.025 up to 0.035 not less. :man_facepalming:t2:

13

I’d like to suggest an improvement for Dynamic Flow Calibration.

Currently, dynamic flow calibration is performed the first time a filament is used during the print itself.
On H2D / H2C printers with fast left/right nozzle switching (and also on H2C with relatively fast hotend changes), this causes a significant increase in layer time on the layer where the hotend/nozzle change happens.

Because of this delay, the printed layer cools down more than usual, which can lead to plastic shrinkage and visible artifacts on that layer.

I propose adding an option to run dynamic flow calibration for all filaments used in a print job before the actual print starts.
This would avoid long pauses during printing and help keep layer times consistent.

For single-nozzle printers, this feature is probably not very important, since color or filament changes already take a long time, and increased layer time is expected there.

8 Likes
14

When are you going to increase number of PA profiles? for (at least) A1/A1 mini, as it is limited to 16 profiles.

I have many filament from different brands many of them unique and Bambu does not offer alternatives but I can’t store calibration result due to the very low limit of 16 profiles.

Orca slicer by the way does not have such limit, but I can’t use Orca slicer anymore due to know reason.

15

Community tech talk where Bambu posts something that everyone can read in the manual or on the wiki and then proceeds to ignore questions from the community? lol.

I like the idea of this, but you need to engage with the actual community. I have a lot of questions as to how the “flow dynamics” aka PA (why not just call it that, why be a snowflake?) works on the H2C, but since no one from Bambu is here it seems pointless to ask because it will just be the community guessing, which isn’t helpful.

2 Likes
16

Hello, first of all, I would like to say that I really appreciate how you are gathering ideas and feedback from the community regarding flow dynamics. Thank you for valuing user input and involving customers in this process.

For flow dynamics calibration, a single K-factor value is either determined automatically or entered manually after testing. However, our printers operate at different speeds, and in practice they require different K-factor values depending on print speed. For this reason, I’m wondering whether there are any plans to extend these calibration tests to cover multiple speed ranges. I’m not sure if you are already working on something like this, but it would be a great option to improve overall print consistency.

Additionally, on my A1 printer, I see almost a twofold difference between the results obtained from the flow sensor-based calibration and the values I determine through manual testing. I’m curious whether this is a known or general issue.

Unrelated to the above topic, but worth mentioning: on my P2S and H2 series printers, when the bed temperature is changed during a print, the hotend flow changes as well. While the bed temperature is actively increasing, the printer experiences under-extrusion. It seems that the heated bed switching on and off may be generating EMI or electrical noise that affects the operation dynamics of the extruder servo motor.

17

How? We don’t see anything on a different color

18

The first purge after the colour swap is.

19

For what I understand from the interview with Dr. Tao, is that the servo motor in the extruder is capable of making numerous measurements while running, and this is used for the Flow calibration at start, but I assume that they can measure the pressure in the nozzle on the fly also, and compensate it dynamically.

If that is already implemented, I really don’t know, but in theory it’s possible.

It would definitely be interesting if @BambuLab could elaborate a bit more on the workings of the “Automatic Flow Calibration”.

3 Likes
20

yeah. there are lots of things that in theory would work, but probably not really working.