RE: Reverse Bowden

To understand what a reverse bowden is, first you need to understand what a regular bowden and direct drive are.

A bowden is a setup where the extruder drive motor is mounted in a fixed place on the frame of the printer. The flexible tube connecting the extruder drive motor to the hot-end and nozzle is referred to as a bowden tube. This tube constrains the length between the drive and the nozzle so when the drive pushes on the filament the force is applied to the hotend. This enforces that when length x is pushed through the drive, the same length x is pushed into the melt zone to have a controlled amount of melted filament pushed out of the nozzle. In this configuration, the only way to pull filament off the spool is by the extruder drive motor pulling on filament because it is presumed that you aren't moving the spool around during a print relative to where the frame of the printer is.

On a direct drive setup the extruder drive motor is mounted to the moving hot-end and nozzle. In these configurations the distance between the extruder drive motor and the hotend is fixed, usually around a couple inches in distance, and almost always in a rigid fixed straight path. (I don't know of any that aren't in a straight path, but I don't know every extruder out there including experimental extruders.) In this configuration, not only does the extruder drive motor pulling on the filament off the spool, but the motion of the extruder assembly can pull filament off the spool if it moves away from where the spool is. This can cause situations where the extruder assembly moves closer to the spool, essentially trying to push the filament back onto the spool. But because the spool won't rewind (unless you have a special rewinder type spool holder) this extra slack can allow a loop of filament to fall off of the spool, potentially causing tangles to happen with the loose filament on your spool holder.

The reverse bowden is one of many ways to mitigate the last issue in the previous paragraph. It is a bowden tube that is connected from the frame of the printer to the input of the extruder drive motor. This constrains the distance between the frame of the printer and the drive motor so the motion of the extruder assembly doesn't change the distance from the extruder assembly and the spool. Often if your direct drive printer doesn't have a reverse bowden then someone out there has designed a setup and you can download the bracket parts that fit on your printer from Thingiverse or similar sites to print them out yourself.

I've played with the idea of installing a reverse bowden on my i3 MK3S, but have yet to have a failure related to not having a reverse bowden, but then again I still use the spool holder that came with the printer so the spool is above the print area. This reduces the amount of potential tangle points along the filament feed path.

BTW, if you do go for the MINI, that is a bowden drive setup so there is no need to add a reverse bowden setup to it.

RE: Reverse Bowden

Wow. He never replied but I will. Great explanation. Thanks for taking the time to write all that. 

RE: Reverse Bowden

You never know who you will help with a detailed description.

That one helped me - Thanks.

RE: Reverse Bowden

Sorry, should have replied earlier, thanks for a great explanation.

RE: Reverse Bowden

One other thing that a reverse bowden will help with is minimizing the amount of tension variability in the filament under certain circumstances. As @sembazuru clearly explained: The X-Y motion of the extruder can pull on the filament at certain places on the print, and push back against the filament at other place on the print. This creates ordered oscillations in the volume of filament per extruder step that enters the melt zone. This is due to the necking (narrowing of the filament diameter) that happens as tension increases. So the higher the tension, the narrower the filament, and the lower the volume of polymer passes that by the extruder gears per increment rotation.  Likewise as the tension decreases, the volume of polymer flow per second increases as the necking stops.

This is usually only a concern if you are printing filaments such as TPU where necking is far more drastic with an increase in tension. For materials with a higher tensile strength you would need to have some really serious drag/friction/tight corner along your filament path for this to be an issue; though it can happen. Say for instance that your printer setup is in a tight space, and you want to be able to print from a dry box, but there is only one place you can put the dry box. Now suppose that the placement of the dry box puts the filament along a path to the direct drive extruder with a tight angle somewhere (either around the corner of the frame, or a tight corner to the extruder itself because the dry box happens to be below the extruder not above it. If you try to print without a reverse bowden tube under such circumstances you are going to end up with very obvious ordered periodic (over/under extrusion) distortions or internal stress warping similar to what can happen with poor first layer adhesion.

RE: Reverse Bowden

Uggh. I put a bunch more time into editing my comment to clarify things and talking about a secondary effect. When I posted the edit, I got "The time for editing has expired... blah blah", the content was gone! Hit back button: Still gone!

...Now to retype it all out again from memory since it is completely lost digitally.... 😪 

RE: Reverse Bowden

Here is the new abridged edition. 🤣  I don't have it in me to write out the full on physics lecture again.

1. Ordered oscillations to the filament flow rate:

   1. Caused by the decrease in filament diameter (necking) that occurs in elastic materials as the filament tension increases.
   2. Is worse for more elastic materials like TPU.
   3. Not much of an issue for the more rigid filaments.

2. Ordered oscillations to the XY displacement:

   1. Caused by the unbalanced force vectors applied to the extruder by the filament tension and the way the gantry applies forces to the extruder in response. This translates into a torque vector being applied to the extruder, and causing unintended displacements in XY.
   2. This is a greater problem for stiffer filaments that will reach much higher tensions and therefore much higher torques.
   3. Not as much of an issue for flexible filaments that will just reduce the tension by necking (stretching a spring). Think about the way springs buffer dynamic transfers of kinetic energy by converting some of the applied external energy into internal potential energy, and then releasing it back as kinetic energy later after the collision.

3. A reverse Bowden tube solves BOTH of these problems simultaneously...

   ...by preventing excessive tension variation in the filament as it gives a nice gentle curved path right to dry box itself! Or to a very versatile/adaptable/relocatable frame mount that can give a good configurable fairly straight filament path from the spool location to the entrance to the Bowden tube. With the reverse Bowden, the only substantial forces being applied remained contained within the filament itself. There is a huge difference in force loading between static forces and dynamic forces. With the reverse Bowden the filament force loading is thus:

   1. At a given constant extrusion speed the direct drive extruder is applying a constant force (tension) to the filament that acts to "pull" the filament into the extruder. This is a static load.
   2. Frictional forces. These are static force loads only as long as there is no motion. The moment friction is no longer enough to stop motion, friction becomes a dynamic load. (Static friction vs Dynamic friction)
      1. Frictional forces in the reverse Bowden tube act to "pull" the filament back towards the spool end of the tube.  PTFE tubing is specifically chosen because PTFE has a very low coefficient of friction, which will minimize this load.
      2. Rotational frictional forces applied to resist the rotation of the spool translate into a linear "pull" of the filament back onto the spool. This can be minimized by using a spool holder that uses really good bearings. That being said, generally a solid fixed rod is more than enough to let the spool spin freely enough.
   3. Dynamic loading from the extruder during changes in extrusion rate and retractions.

4. Further discussion on Dynamic Loading:

   1. As long as the loading is static dominate, the extrusion flow rate should be fairly consistent. Strong dynamic loading on the filament will result in swings in tension, and therefore variation in filament diameter (necking or bulging), and therefore filament flow rate. The only way to apply strong dynamic filament loads in a reverse Bowden setup is with the extruder drive itself; so completely under the control of the slicer. In fact I think I remember seeing a new feature in PrusaSlicer to smooth out the print speed transitions near sharp corners, which would also smooth out dynamic filament loading as well, as extrusion rate is a function of XY print speed. If your dynamic filament loading looks more like an impulse (sudden change in the application of a force) then you have GCODE that asked for it!
   2. I am pretty sure that the calibration routine in the Klipper firmware is looking at dynamic inertial loading of the XYZ motion. The math required is Ordinary and Partial Differential Equations. So all of this can be accounted for in the firmware if you have a way of importing calibration data regarding the dynamic loading frequency modes in your linear motion components. Basically attach accelerators to the bed and hot end, then cause a series of dynamic impulse loads at various frequencies by rapidly switching the direction of various motion components and recording the data from the accelerators. The data can be analyzed and the resonant modes of your printer discovered. As long as your dynamic loading doesn't excite resonate modes, it should have minimal impact on part quality. Again, I suspect this is something klipper already does, but I haven't read the source code; so I can't confirm. Ultimately what you really need to maintain good part quality is to avoid is the resonate frequency modes of your specific 3D Printer, so slowing down isn't always the right answer; sometimes going faster is better. So careful with DE solvers, as they will hand you back a solution set, but it could just be a local minimum or local maximum. Sometimes there is a better solution elsewhere, so make sure to feed the solve quite a few different initial conditions to check for hidden global max/min solutions.
   3. A full dynamic loading calibration model integrated at the firmware level with real-time access to accelerometer data would be super sweet! The firmware could detect resonance issues and adaptively alter the GCODE being used on the fly in order to avoid resonate modes rather than just relying on calibration data.

RE:

Thanks. That was useful to me.