Taurus V5 Cooling Duct for Creality Sprite Extruder, Dual 5015 Fans, Ender 3 S1/S1 Pro/S1 Plus, CR10 Smart Pro (Stock / Spider / Ceramic)

Taurus V5 Final Release

Part of this release in the display case for Klipper WiFi connected TFT displays, please check it out: https://www.printables.com/model/624504

We're sorry this took so long, but the V5 changes were quite drastic, in fact - although it looks almost the same as V4, it is a completely different duct geometry - performs at a whole better level according to +20 independent alpha testers around the world we've worked with over the course of this project in the past months. In short, the duct has a lot more inner volume, the tube curvature is a lot smoother, the air outlets are larger and shaped differently (flat instead of oval, creating wider jets for increased air flow. In all honesty, I don't know what else to change from here to make it any better (which is the same thing I said after V4, and still found ways to make it even better ;-)

Taurus Versions 1 through 4 were designed in Autodesk Fusion 360 (with some challenging problems). Taurus V5 has been designed in SolidWorks from the ground up, eliminating ALL of the issues and is now a truly parametric model. We can now generate new custom models for any hot end offset from stock. If you need a custom duct made that is not already posted, let me know and I can generate one by configuring parameters on the model, popping out a new STL.

V5 Quality Control

Despite all the different opinions on different forums, some have posted water tests and FDM simulations, it seems to be that the Taurus series just rocks and people seem to love it, perhaps also because I always tried to keep it free. 1000's of users around the world are running this duct now (previously on Thingiverse), which is awesome. 

Initially, I did not design this duct for the masses, I designed it for myself in order to learn more about the hobby itself, get back into CAD design, the problem and a solution that would enable higher speeds and better-quality prints, while having some fun as well.

There's only so much one can test with various methods on the efficiency of a cooling duct, I learned that basing the ultimate judgement on water tests or air flow measurements is only telling a very small part of the entire story.

The result is what counts, with all these variables in play, I went for a 17 minute Benchy with V4 back then, and that worked to some extent. I upped that goal a bit to 16 minutes for Taurus V5 (I vowed to not release it before I could document that achievement).
To make that even more interesting, eliminating all the concerns about inertia and weight on the print head, here is a video of how that went with the Knomi V2 TFT display case mounted  in operation:
 

V5 Remix Culture Change

I think with all that's left from the Sorkin duct, 98% of V5 qualifies as my original design.
Between V4 and V5 I've seen a lot of people remix this duct in different ways, I didn't particularly endorse or think makes any sense. Especially when someone (you know who you are), reverse engineered the geometry, removed the branding from the sides and posted not only the STL but also the CAD model.
Another person stretched it to fit his hotend offset, which was a complete waste of time: Part of my efforts was to make it fully parametric, with an hot-end offset as the input. If you need a custom duct, just tell me what you need: It takes me less than 5 minutes to generate that particular custom model, which will then become part of the release.

Please understand that not just myself, but also many of my peers spent upwards of 1000 working hours of our time between all the versions and variants to get this thing to where it stands now, in design, printing prototypes and testing.
To see this design butchered by hobbyists and clouding the experience for potential users is not something I wanted.
With all that in mind, I'm revoking the right to remix this model as of 3/10/2024.
If you posted a remix before that date, please remove it from all platforms.
I can't force you to do this (and I will not report anyone to the platform), but please respect what I said above and think about it.

History

I wanted to further optimize the airflow and fix a few things that others have complained about, and I struggled with myself.
We are at version 5.0 now, because people around the entire globe/from many different countries started to contribute, tried this out and provided me with feedback that keeps making this thing perform better and better with every new release. Every single comment, idea and improvement request was evaluated and made it into this version in some shape or form. In a strange way, this is where organic design meets crowd·sourc·ing, resulting in a high performance duct design that has produced very consistent results and even set some new records in the 3D FDM printing world.
 
We named this duct "Taurus" because when looking at it from the top, the shape looks similar to the horns of a Bull/Taurus, paired with massive air output. After several model iterations, the only things left from the original are the mounting ears and the fan supports, the rest is of my own design, with the help a lot of contributors (please see the credits).

I'm not a computational fluid design engineer by trade (CFD), but learned a lot over the years. A good example is what water does, while flowing through pipes or down a mountain stream. As you turn the pressure up, dynamics change and it is no different than with air and cooling. Air molecules just have a lower mass, but the exact same dynamics apply - I think that every actual CFD engineer would agree with that.
Computational fluid dynamics is still a very sketchy science and no matter how you simulate it and which algorithms are chosen for the simulation, it will bite you in the ass somewhere in terms of turbulence once a sharp turn is involved. My approach was to just do what made the most sense: Zero Turbulence, zero turns, no edges in the inner volume.

This is the result of my quest to find the optimal path from the fans to the part to achieve the highest possible pressure at the cooling nozzle outlets. I actually got a lot of criticism about this approach, even from the original designer and thought many times that I might be on the wrong track, but the good results and positive feedback speak for themselves.

If you're interested in CFD science and this is nothing new, here is why I did it in this way:
The goal was to make the flow of air as "laminar" as possible, while at the same time increasing pressure by doing this:
My approach was to keep a fluid/linear reduction in cross section on the inner volume, from the top, down to the business end of it, fully organic without any angular, sharp turns involved (causing turbulence and more back pressure as a result).
The linear reduction in cross section causes the pressure to build up, turning the fan output into a high-speed jet of air, that we can now control and "shoot" directly at the center of the newly extruded filament bead. This provides high efficiency cooling at the optimal angle, while at the same time just undercutting the nozzle with the two precisely directed jets. Cooling the nozzle tip itself is obviously a bad thing to do and we want to avoid that.

I have another theory on why this works so well.
Have you ever wondered why golf balls have these dimples in them?
Look it up, or click here: https://www.youtube.com/watch?v=fcjaxC-e8oY
I believe the ripples from printing are creating this theoretical boundary layer, reducing friction and back pressure by accident, helping the laminar flow and allowing for improved transport of the air from the fan to the nozzle. Any type of turbulence by an edge in the inner volume would cause this effect to diminish.

Another major design intent was the visibility of the extruder nozzle, which this design solves nicely. Heck, we might as well put a miniature camera to ride along on it now! :-)

This mod is verified to be compatible with:

• Creality Ender 3 S1
• Creality Ender 3 S1 Pro
• Creality Ender 3 S1 Plus
• Creality CR10 Smart Pro

As of version 5.0, this duct is also compatible with (custom STL's):

• Spider Hot End Mod (3.5mm lower, use the provided STL, not the stock one)
• Volcano Hot End (10mm lower, Volcano.K1.CHCB-OT)

Please don't get too excited, this duct is not a magic bullet to fix all the 3D printing problems you may have, it addresses very specific things in the areas of cooling and visibility that I wanted to improve.
You must have a well-tuned workflow already (including slicer and mechanical) to see anything change to the better using this mod.

Enjoy, hope this will be as much fun for you as I had designing and playing with it.
Mac

Feedback

It's free, so please, please - pretty please, take the time to provide me with feedback in return. :-) You can do that by simply posting a "Make" and telling your story, how you printed it and what your experience in using it was so far.
I would love to hear back from you, so we can improve this thing even more, "together". :-)

Update History

03/10/2024 Remix Culture Change: Please see above for explanation. We no longer endorse makers to remix this model (please stop / prohibited / Verboten!). There are very good reasons for this change. You can still design around it, create add-ons or accessories, but the duct geometry itself is not to be changed from here on. This is our job and if you need a custom geometry, we will do that for you.
If you published a remix prior to 3/10/2024, please remove it.
If you have published a remix of V4 for example, you didn't do the community a service, because the V5 geometry is much more advanced and the performance is much better from what I heard so far. We don't want those old geometries clouding the landscape.

02/22/2024 Version 5.0 Beta Release
- Switched from F360 to Solidworks, import/export was not possible, modeled a new one from scratch, using the lessons and references learned. Fully parametric now.
- The lofting features got more precise and organic, the jets changed from oval to square with fillets providing more of a wider blade instead of a pin pointed cooling approach.
- The horizontal mounting arm got eliminated to save some weight. It turned out that it wasn't really needed. The fans really needed only two vectors to be held securely in place (user suggestion).
- The print now has pre-engineered supports to print along with the of the model.
These supports can be easily removed after printing. This was to help out a wide user base during V4 that could not get an acceptable overhang transition in this area.
- The contact surface with the build plate has been enlarged, adjusting angles and geometry.
We expect this to print much better than V4, since the overhangs near the build plate have been eliminated for the most part.
- Regressed the inset mounting ear feature to use the stock screws. Countless users reported that the mounting ears were breaking off on them. This is now the way it was in V3, the stock screws are no longer supported.
- Added TFT universal display case and custom firmware (different project).

06/27/2023 Version 4.01 Minor Maintenance Release
The elongation of the mounting slots in the last release has weakened the mounting points a little, caused these to deform if a screw is tightened too much and to slip through the slot (1 report so far). Added more material around the ears to keep that from happening.

06/21/2023: Version 4.0 Official Release
This version inherits all the new features from the previous spider alpha release and some more. Here are a few highlights:

• Visibility of the nozzle is much better now, we spread the two tubes as far apart as possible.
• Relaxed the first turn of the tubes a little bit, it now sits flatter on the build plate and prints better without supports.
• We finalized a change where the original screws are meant to be used to mount this duct and rolled it into this release.
• The mounting bosses have been elongated, there is now an additional 2mm adjustment possible in both the up or down direction. While that makes no sense in stock configurations, it helps a lot of makers that run custom hot ends.
  Again, the design is such that the bottom of the two jet nozzles sits exactly 1mm above the tip of the extruder nozzle.
• Added a spacer/washer stl that you can print.
  Some have lost their original screws and used standard metric hex head screws instead, that causes the mounting bosses to deform. To counter that, we included a special spacer/washer that you can print yourself and insert three of these into the mounting bosses.
• Added a "Taurus" letter logo to both sides of the body, confirmed that is printable and paintable.
• Changed Taurus logo in the front to bold "V4" lettering.
• Added an extruder visualizer in the shape of a taurus head, printable in multiple colors in the same part. Instructions how to achive this are following in a seperate section.

06/11/2023: The spider version is ready for alpha testing. There are a lot of changes in this that will flow back to the official design, soon.
I had no way to test this, so please don't be disappointed, printing and testing this. Please provide feedback.

• It aims at a virtual spider extruder nozzle at 3.5mm lower than the first stock hot end release, has the same geometry as before, just lower.
• The tubes have been "spread apart" further, to hug the BLT some more, providing even more field of view, that looked very cool actually.
• The problem with the strange artifacts in the transition from the fan geometry to organic has been found. It's now 100% smooth.
• The attack angle of the jets has been changed from 18.5 to 18 degrees. We found that it was about 0.5mm too low for optimal cooling of filament that comes out of the extruder nozzle.
• Added a Taurus logo to the front.

05/28/2023 Roadmap: Nobody has reported this yet, but a print of this component without supports in the recommended PETG material has a tendency to "leak" air through the layers. This problem is predominant in the area where it makes a sharp turn. The problem is amplified by printing it without supports and makes it leak even more.
To combat that, the next version will have a slightly higher wall thickness for the tubular paths and I'm hoping that will seal these leaks. This may also be an opportunity to build a small built-in support gusset into the model to guide the printer better and add more adhesion during the printing process.

05/27/2023: Public Castle Preview - V3.1.1-beta

• A lot of makers have already reported that the increased cooling that was added to their printers on the initial design has enabled them to print the follow-up versions completely without supports. I did not plan for this and could not believe this at first and tested it myself. Wow! I did not expect this to make that much of a difference, but almost all maker prints had a problem near the two outlet nozzles, with lines dropping out because that is a horizontal overhang at the end of the print.
  I found a way to fix this by adding a very minute gusset at 70 degrees to the tips of the nozzles, you can hardly see it, but it stops the final extrusions from dropping out, making this completely printable without supports.
  This release includes all the features from V3.1-beta

05/23/2023: Public Castle Preview - V3.1-beta

• Corrected mistake in the documentation: "3x M3x8 screws" replaced by "3x M3x5 screws". This came from the Sorkin docs, they are too long. You can use M3x6, but this requires a washer. Let's fix this:
• Countersunk the slotted mounting points so that the three mounting screws from the original fan can be used to assemble this duct, no extra M3 screws required! :-)
• Added multiple fillets to reduce weight.
• Reduced more weight by removing material between mounting points on the right.

05/22/2023: Released V3 - Initial Public Release

Enjoy and please provide feedback, post a make!

Print Settings

Printer:

Any

Rafts: Doesn't Matter

Supports: Yes

Resolution: 0.2 Layer Height, 0.4 Line Width, but choose your own.

Infill: 20% (Doesn't really matter, it's a "skinny" model as reducing weight was/is a design intent.

Filament brand: Any

Filament color: Printing all mods in blue has been fun, choose your own.

Filament material: PETG (recommended)

Notes:

WARNING: PLA doesn't work very well for this use case as the tips of the duct may melt near the hot end during normal operation.

RECOMMENDED: PETG works great for this application, but you can use any filament that can take the heat.

TIPS: You can print this any which way you want, but it is designed to print flat side down (the front in the pics, I oriented the STL so that it should land in your slicer like that by default). Just give it a little support and take your time. In Cura, the default tree support works almost perfectly.

I'm not saying this is easy to print, but totally possible with a little support in the critical areas - I did try to optimize the flat faces as much as possible to not create any crazy overhangs or bridging situations.

Post-Printing

Assembly

What to purchase:
2x 5015 24v fans
The 5020s should work on this interface (not tested), are not as efficient, but also not as noisy. I went with these 5015s and really like their performance so far (noise is bearable):
https://www.amazon.com/gp/product/B07DB7DLMM

3x M3x5 screws
2x M4x20 screws

The M4 screws go through the top holes to mount the fans. Screw them in while applying a little pressure, they cut their own threads and should end up moderately snug.

• Wire the two fans in parallel, you can cut the end off your original fan so you don't have to purchase the little plug (there's a picture included to show what the splicing should look like using heat shrink tubes).
  If you want to keep your original fan intact, you can purchase these: https://www.amazon.com/gp/product/B013JRWCBU/

BEWARE: The colors (red/black) on this product were reversed for me, so make sure you compare the original fan plug to what you get and wire accordingly. The included picture shows the aftermarket plug (black wire on top, reversed). If using the end of the stock fan, the red wire should be on top. If still not sure, measure it - using a voltmeter.

Adjustments

As with the original K3D duct, this version has slotted mounts so you can move it up or down by a little. If you look at the renderings, it was designed to sit about 3/4 up in the mounting slots and that is where it should be.
The design is such, that the bottom of the cooling nozzle tips should be ~1mm higher than bottom of the hot-end extruder nozzle, so that even on "dirty" prints - it will hopefully not scrape across the top of your print, worst case: Catch and knock over your part.
You can eyeball this, there isn't really a right or wrong way to adjust it, just go with what you feel is best for your build.

Update here (06/21/2023): Some are running custom hot ends, which asks for more adjustment space, so we elongated the mounting bosses by 2mm up and down for more flexibility. A custom spacer/washer is also provided in case you lost the original hardware.

V5 Alignment Tool

As mentioned above, the efficiency of this duct depends entirely on how high or low you mount it in reference to your extruder head/the nozzle.
The design is such that the jets that come out of the duct blow exactly at the spot where the filament comes out of the nozzle.
If you set this too high, you'll be cooling the nozzle - which will have an adverse effect on everything else. >>> Not good.
If you set this too low, you'll still be cooling, but not as intended.
The design is such, that the bottom of the cooling nozzle tips should be 1mm higher than bottom of the hot-end extruder nozzle.
Up until now, we assumed that the user would figure this out, but the geometry of the duct is very deceiving.
To make that easier for you, we included a new alignment tool (sparked by David Ericsson) with V5 and up that you can rest your assembly on or even adjust it on the printer. It's up to you how you do that 1mm gap between the nozzle tip and the bottom of the nozzle outlets.
In that setting, the duct will perform best and at peak efficiency, cooling exactly what it was designed for: The freshly extruded filament bead only and nothing else.

Printing the Extruder Indicator (Advanced)

This will be a little bit of a challenge for you!
But perhaps you will also learn a new technique in the process (at least I did).
I suggest you practice this first (two color layers) before you print the actual indicator with the stem.
The detail in the Taurus head logo is so high that is should be printed with a finer nozzle than the standard 0.4mm.

Here are the few things you need to get started

• Red, White and Black filament (PLA, PETG, you can use whatever you have, but obviously it should be the same filament type in all three colors).
• The smallest nozzle you can find (I had a 0.2mm nozzle laying around, that worked quite well). A 0.25mm nozzle might work too, up to you.
• Cleaning solution (I recommend Isopropanol). The details are very small and we need the best possible adhesion so things don't come off while you're printing this.

Parameters

• We'll be printing this at 0.12mm layer height and 0.2mm line width.
• In terms of speed, go very slow: 10-20mm/s perhaps 30, but not any faster than this.

The Problem at Hand:
We're printing this in three steps, first white, then red and then the black on top.
If you haven't messed too much with the standard startup g-code in your slicer, in between these three prints your printer will do a homing routine and automatic bed leveling (ABL), while what you already printed is still on the bed (Collision Warning!)
To get around the z-axis homing is super easy: As long as the print is not located in the very center of your bed, the BL touch will not land on top of it and throw your z-height off during homing.
To avoid the ABL touch points is also easy:
I've seen videos where people turn their ABL off by modifying the startup g-code, then manually loading the mesh and other such trickery in order to support this technique. I wanted something much easier, and it really is once you understand it: Think of the standard bed mesh as tick-tack-toe grid (4 boxes wide and 4 boxes high). Where the lines intersect is where the BL touch probe contacts the bed to get a reading. As long as we locate the print in one of those boxes, we're fine and the probe will never hit our stuff!
You can even look at these boxes if you have a sonic pad, go to "Tune", then chose "Probed Matrix" and "Wireframe".

Slicing the models

• Load the white model into your slicer (Extruder Visualizer White 0.2mm.stl).
• Locate that model 10 mm to the right of the center of the bed. Then you can either move it up or down by 10mm, I moved mine up. This will ensure that the BL touch probe doesn't hit it during homing or the ABL routine (we're in one of the tic-tac-toe boxes now).
• Load the red model into the same slicer session (Extruder Visualizer Red 0.2mm.stl)
• Do the same as with the white model (10mm over and up or down by 10mm)
• If you're lucky it will fall in place, if not - zoom in really close and move the red model so it fits into the white model like a puzzle piece. This might be a bit finicky, so take your time to get it right, while zoomed all the way in.
• Finally, load the black model into the same slicer session (Extruder Visualizer Black No Stem 0.2mm.stl)
• Do the same as with the red model, until it fits over the two other models like a puzzle pice. Just take your time to get it right.
• Don't forget to change your profile to 0.12mm layer height and 0.2mm line width and crank your speed down to 10-30mm/s, the slower the better in this case.
• Save the file, this is now our base-file with all three models in it.

Prep your Printer

• Change to the smaller nozzle.
• Print some first layer tests, adjust your z-offset if needed and make sure you have very good squish and adhesion, along with a nice first layer.
• Clean your bed really good with cleaning solution where the print is going to be located, I recommend Isopropanol, we need maximum adhesion for this technique.

Test Printing

Step 1:

• Load white filament into your printer
• Load the base-file, delete the red and the black models
• Print and leave it on the bed.

Step 2:

• Load red filament into your printer
• Load the base-file, delete the white and the black models
• Print and leave it on the bed.

Step 3:

• Load black filament into your printer
• Load the base-file, delete the white and the red models
• Print and inspect the result.

Print the actual part

Once satisfied and you have it working well, it is time for printing the final part.
Simply switch out the black model with the black indicator base (Extruder Visualizer Black With Stem 0.2mm.stl, you may have to rotate that).
At least in Cura, I couldn't locate it looking at it from the top because the red and white models are now covered up, simply pan your view so that you're looking at the build plate from underneath.
Then simply repeat the process outlined above to print the part.

Why this apparently works:
I learned while developing this that printing just one layer of color at a 0.12 layer height results in a paper thin layer of filament. With the black right behind it, the white looked more like a gray and the red like a smudgy dark redish "something". I had to add a 2nd layer to keep the colors somewhat true.
So what we're printing is actually two layers of color and then the black model fills in the rest as the background.

Good luck, I hope you enjoyed this little challenge and learned something new! :-)

How to use this or any other high efficiency cooling duct

1st layer and adhesion

You should never have the fans turned on for your first layer, that's just bad practice and the #1 cause for adhesion failures. If your slicer supports it, keep the fans off and then gradually go to your final percentage by the 3rd or 4th layer.
This transition is important because if you don't do it gradually, then the first few layers might warp and pop the print off the build plate on their own.
Using a duct like this vs stock amplifies this problem but has many benefits in other areas, if you can find a way to control this effect.

Benefits

What this duct is trying to solve are several problems. We can all agree that Creality printers like the E3S1Pro have largely insufficient cooling to support higher speeds. It's fine if you print at 60mm/s but as you crank it up with something like a SonicPad, the still relatively low-cost consumer package starts to fall apart and the print quality drops through the floor.

Next, we can't see anything from the front of the printer. It's ok'ish with the stock cooler, but almost all the mods that exist currently totally block the view to the nozzle from the front.
A major design intent of this duct was to solve that issue and provide a nice view of what is going on around the nozzle from the front of the printer. While that is still a bit restricted, you will hopefully find this better than any other solution currently available.

Extreme bridging is something I also wanted to address with this duct, this requires a high flow of air to be able to suspend extruded filament basically in mid-air with no support. The nozzle provides the tension to bring the filament to the other side of the bridge.

High efficiency cooling on small features is a benefit also.
The chimney on the Benchy is such an area. Normally the slicers reserve a minimum cooldown time to for the layer below the currently printed layer. You can set that to zero, but if you do - the layers below are still gooey and unstable. This duct has such a high air volume output that it will almost "freeze" the lower layers in place.

Weights

Stock fan alone: 26g
This duct with both fans alone: 87g
Sprite pro print head with this duct and both fans, total: 398g
Sprite pro print head with stock fan, total: 337g

Delta: +61g

Credits

• Thanks to Sorkin for the original K3D design.
• Thanks to Kingsidorak for modeling the Sprite extruder.
• Thanks to Jimmy Phan for overhang testing.
• Thanks to Jake Loughran for the idea with the extruder indicator, that sparked what I did with the Taurus logo as a fun and at the same time challenging extruder indicator. He also tested the Sphere on a stick challenge successfully and allowed me to post his pictures as part of the V4 release.
  Also thanks to Jake for general moral support, advice and always trying to help/test print new models during V5/Knomi.
• Thanks to Brad Brooks for validating the Knomi case during V5 as well as testing the engineered supports and finding a problem with those. Your help is very much appreciated.
• Special thanks to David Ericsson for the idea for the alignment tool.
• Thanks to Chance Salway for testing and validating the V5 release.
• Thanks to Guga Machado de Carvalho for extreme bridge testing and moral support.
  Guga was also responsible to point out several problems with the attack angle.
  Guga also helped me develop the Spider version, during many design iterations and test prints. Guga was also instrumental in getting the "Taurus" lettering for the V4 version and upwards scaled and situated in the optimal spots.
• Thanks to the members of the Creality Sonic Pad Official Community for suggestions comments and ideas during the design stage.
• Also thanks to everyone else who chimed in, your words of encouragement that kept me going.

Licensing

05/27/2023: Sorkin has agreed that there are no restrictions from his point of view. He doesn't agree with the efficiency of the design, but that is my problem now, taking the candle forward.

With that out of the way, the exclusive and specified non-commercial license goes into effect as of 05/27/2023.

06/03/2023: Some have already asked. Licensing Clarification is necessary and asked for. Under the current licensing agreement (CC Non-Commercial), you can print one item or more for personal use, but you cannot print this item and sell it on Aliexpress, Amazon, Etsy, etc. That would result in direct intellectual property infringement towards my entity, the design IP itself and I don't want to have to deal with that mess if possible.

Offer: If you are interested in producing this item for profit in quantities or to make a kit, please contact me directly - I'm sure we can work something out.

V5 Update