Ultimate Duct Design Guide for 3D Printers | My Top 17 Secrets for Custom Cooling Ducts that WORK

This one goes out to all my homies on Discord. Hope you can read fast! This guide is, uh, information-dense, so STRAP IN.
I want to emphasize that nozzle design is sort of its own thing and an area I'm still learning. There are lots of ways to set up nozzles besides what you'll see here. This guide is mainly about getting as much airflow as possible at the point of outlet, less about the best way to use it. There probably isn't any one best way.
Duct CFD and fan adapter images used by permission of their respective owners. "Discord" chat sequences and users shown therein are fictionalized but I've been party to real instances of all these discussions on many occasions, the only difference being that the real people involved have generally been much less rude.
Trihorn cooling duct STLs shown sliced are from the VzBot github: github.com/VzBoT3D/
Tips/donations if that's your thing: ko-fi.com/jamespray
I'm also a writer! I don't care that much if you mash dat like or smash dat subscribe (I mean, don't NOT, but it's cool, we're cool), but if you like sci-fi adventure stories, check out my debut novel, JONESY FLUX AND THE GRAY LEGION: bit.ly/JonesyFluxAZ
Or visit me at:
www.jamespray.com
/ jamespray
/ jamespray.author
Timestamps:
00:00 - Introduction
00:43 - Is Duct Design Black Magic?
01:15 - This video teaches good duct design, NOT how to CAD
01:43 - What about CFD?
02:17 - SECRET 1: the Golden Rule
02:54 - SECRET 2: You Skip, You Fail
03:45 - CAD time and SECRET 3
04:34 - My duct design workflow steps
05:05 - SECRET 4
05:28 - SECRET 5: throwing shade =0
05:43 - SECRET 6
06:04 - SECRET 7
06:25 - SECRET 8
06:42 - SECRET 9
07:10 - SECRET 10
07:54 - SECRET 11
08:07 - SECRET 12: seriously this
08:29 - SECRET 13
08:44 - Secret 6 revisited in more detail
09:41 - SECRET 14: master this, master ducts
10:24 - SECRET 15
11:04 - SECRET 16: notes on aiming nozzles
11:30 - SECRET 17
11:53 - CAD design mostly done
12:28 - Outro and BONUS ADVICE
Music credits:
Track: MILANE & Greg Aven - Like You [NCS Release]
Music provided by NoCopyrightSounds.
Watch: • MILANE & Greg Aven - L...
Free Download / Stream: ncs.io/LikeYou
Track: Max Brhon - Redemption [NCS Release]
Music provided by NoCopyrightSounds.
Watch: • Max Brhon - Redemption...
Free Download / Stream: ncs.io/MB_Redemption
Keywords and stuff: CAD, CAM, CFD, OnShape, Fusion 360, TinkerCAD, OpenSCAD, Blender, AutoDesk, Solid Edge, FreeCAD, SolidWorks, sweeps, lofts, surfacing, sketch, STL, 3MF, STEP, surface modeling, manifold, extrude, offset, guideline, chamfer, export, import, custom, Voron, VZBot, Ender 3, Creality, Prusa, FLSun, BondTech, Artillery, Rook, Kossel, Delta, Cartesian, CoreXY, CroXY, Annex, K2, K3, Bambu, P1P, Revo, CHT, Bozzle, Toolhead, CPAP, Goliath, VolcoMosq, Rapido, Dragon, Revo, Pico, V6, 3D printer, cooling duct, performance, efficiency, 3D Printed Compressor, 3D Printed Turbocharger, 3D Printed Supercharger, 3D Printed Blower, 3D Printed Roots Compressor, 3D Printed Screw Compressor, 3D Printed Side Channel Compressor, 3D Printed Side Channel Blower, 3D Printed Ring Blower, Delta 3D Printer, FDM Printing, Major Hardware, Fan Showdown, CPAP cooling, CPAP blower, CPAP compressor, Printer Layer Cooling, The best print cooling, The most powerful print cooling, 5015, 5010, 4020, 4010, 4028

[@jamespray]

Fun fact: I composed the opening in a 3 A.M. Monster-fueled driving haze around hour 19 (of 23) of a roadtrip home from Florida in February If I ever do a design deep dive like this again, what's a subject you'd be interested in seeing?

[@WaffleStaffel]

Thanks, very helpful! By server fan, do you mean tubeaxial? Tubeaxial fans have more static pressure than blower fans?

[@jamespray]

@@WaffleStaffel Server fans are fans from (or for) servers. They tend to be much higher power as well as physically thicker than normal PC axial fans, and they can be slightly better at static pressure. There's a range of performance for each fan type, though. It helps to check the spec sheet, if you can find it.

[@WaffleStaffel]

@@jamespray Are you referring to the fans which look like normal axial fans, but are much deeper and sometimes have 2 sets of blades?

[@jamespray]

@@WaffleStaffel Yes. The ones with two sets of blades are actually just two fans bolted together, but the second one is special in that it rotates in the opposite direction (since stacking two axials that rotate the same way doesn't work very well without stators to de-swirl the airflow).

[@MeTuMaTHiCa]

Good work, especially formulas. Can u share a STL u ducted. I wanna see inside. Keep going thx

[@NathanBuildsRobots]

Related to tip 4, fan placement is super important. Really helps shape the overall duct design. If you can put the fan right up in there so the duct is only a couple cm long and doesn’t have to change the direction of the air, that has been great for me.

[@jamespray]

Very true, especially if the fans aren't huge. The simpler the duct the better, generally, and fan placement is (within your constraint envelope) often the biggest lever you can pull to make that achievable.

[@eirikrcman1511]

i remember seeing alex kenis making straight fan shroud's for axial fans with guide fins to straighten the direction of air from the fan rotation and a center cone to take away turbulence from behind the hub of the fan with really big improvements

[@Nifty-Stuff]

Wow, who knew someone could make a video about duct design so fun and entertaining? You clearly put a TON of time into this script, editing, etc. I'm subscribing!

[@jamespray]

Thanks so much! Welcome aboard

[@pusnirizda5481]

Great tips but i still just can't imagine how you manage to do all of the in sketchup. Unbelievable levels of skill.

[@airsubzero]

What I got: 0- Narrowing down a fan duct: Increase in Airflow Velocity: Faster Solidification. 1- you need to figure out how much can you narrow down the outlet before your fan can't handle it. 2- you should cut one dimension of a duct in half in a run of no less than 2.6x of that dimension. "changing the cross-sectional area too fast cause throttling" 3- to get started with a duct is to place your start and finish and then work to join them together this is a lot easier than trying to work from the start to the end. 4- you need a smooth overall run with big curves. 5- the ideal duct cross section shape is a circle the more perimeter you have for your cross-sectional area the less efficient your duct will be. 6- the shortest possible duct with hard bends is likely to underperform versus one that's a little longer but follows a smoother path with larger radius bends. 8:15 7- avoiding Sweeping Curves Along the Wide Axis: Sweeping a curve along the wider axis of a duct (which has more space) may require more area and can be less efficient in terms of space utilization. Sweep axis " sides ". 8:35 (bend it like a ribbon 8:32 8- lofting can cause problems, try having more stations and spreading them out so that you're not overlapping too many changes to the size shape and direction of your duct. 9- don't cut down to the final Outlet cross section too soon large ducks are more efficient than small ones 10:42

[@billklement2492]

James, great video! Now my understanding of flow design greatly exceeds my CAD ability!!!

[@jamespray]

[@ProDesigneHd]

Very well done. Entertaining and informative at the same time! Keep up the good work

[@jamespray]

Thanks!

[@brianadams7903]

I printed that FL-Sun SR duct! glad to "meet" the guy who designed it!

[@jamespray]

I haven't done anything for the SR, but it's probably a design space with similar constraints, so I could see such a duct looking similar to this one. The only handles you'll see my designs under (legitimately) are this one, the one in the Discord sequences here, and Kiolia.

[@jeralm]

Amazing work, I will definitely apply this knowledge the next time I design a plate in OpenSCAD.

[@jamespray]

lol! 🤣

[@avejst]

Great walk-through of the process 👍 Thanks for sharing your expirence with all of us 👍 😀

[@jamespray]

You are very welcome 😀

[@beauregardslim1914]

Great stuff. You confirmed a couple of my suppositions and I learned quite a bit. I will add your bowl-of-water-on-the-print-bed testing to my existing run-the-kitchen-tap-through-it method.

[@jamespray]

Glad it helped! Running under the tap is also a tool in my arsenal ... good-ish for visualizing exit flow :)

[@Ytbas]

Came for the technical explanation stayed for the entertainment. Hats off. You, sir won a subscriber.

[@jamespray]

Thanks, appreciate it! Welcome aboard

[@kloakovalimonada]

That's great and fun, subbed!

[@stefanguiton]

Excellent video!

[@jamespray]

Thanks as always

[@willl84]

Watching you do this in SU makes be have PTSD from when I first started learning CAD for 3D printing with SU. I started working at a place that uses Inventor (this was before Fusion 360 was a thing) and I was blown away at how much easier it was to do basic things. Plus having a model history! I don't see how people still use SU honestly lol

[@jamespray]

Yeah, this is almost literally the worst sort of use case for SU (compound curves) but honestly this is the only important area I feel like the tool is in my way for the things I do. I wouldn't recommend it for 3D printing design stuff to someone who didn't already use it, but I think it's more powerful than it gets credit for and I have built some extraordinarily complex designs in it without major issues. It's just up to you to structure projects correctly ... it doesn't hold your hand in that regard at all.

[@reubertt]

I certainly had a bit of a scare when I saw that it was being done in SketchUp (team Blender here), but I was surprised by the quality of the tips and content. KZbin recommended it to me because I'm thinking of buying a Sovol SV06 and I've already researched its cooling problem. Because of this video, I think I'll risk developing a better system for the Sovol instead of using the ones that are already available on the internet. Congrats man, really good.

[@jamespray]

Thanks very much for the feedback! I'm glad you gave it a shot and found it helpful. That's really just the sort of case I was hoping to support with this. There are a lot of bad duct designs out there, but that could be changed! Good luck with yours, and cheers to a fellow mesh editor ✋

[@HarryPorpise]

Holy shit this just came at the perfect time, I’ve been putting off modeling the ducts for my custom printer for a while haha

[@jamespray]

Awesome, hope they turn out great!

[@PureRushXevus]

Found this video linked on a reddit post when I googled "blender fan duct narrow" and didn't expect to watch the entire thing right away at all, very good job on editing and keeping it informative without padding out the video to 40 minutes lol also thanks for linking the music in the intro, also didn't expect to find a banger in a video about ducts xD

[@jamespray]

You're welcome! It's the worst when you hear a good song in a video and there's no dang credits. Like the one midway through the Dakar 2018 Cars video. Such a banger and I'll never know where to find it

[@PureRushXevus]

@@jamespray you talking about "Best Of Car - Dakar 2018" ? I just tried using shazam for like 10 minutes, and got 4 different songs that were wrong :(

[@jamespray]

@@PureRushXevus Yeah, that's the video. The song that cuts in at 2:30. Gosh I want that track!

[@RaphaelRema]

Great bunch of info. Thanks 👍

[@jamespray]

You're welcome!

[@paulngopal]

Dood! I need a duct - 40mm going to 20mm with a 90 degree twist and a 120 degree bend and I need it yesterday! I am sure you can knock that out in a couple minutes. (Great video - it took me a week to design mine and I am still "tweaking" it. ) Thanks for the info. Great job deciphering "11 degrees" edge to edge - That's like plank's constant of Max Duct "Narrowing" Angle. Nice!

[@jamespray]

Haha, glad it was helpful! Duct tweaking is a fact of life, but when it's your own design it's a lot easier, and they tend to be quick prints, at least!

[@tron11751]

been using sketchup since i was 12 its amazing software, i believe i just reached its limits and plasticity 3d filed the gap well. amazing video and great explinations keep at it!

[@jamespray]

[@fish3dp372]

Good vid, subbed!!

[@jamespray]

Thank you, and welcome aboard!

[@Macgyver1307]

Great video 🎉

[@jamespray]

Cheers friend!

[@C1Rob1990]

Great video! :D

[@jamespray]

Thank you!

[@tinchodias]

Impressive, thanks!

[@jamespray]

You're welcome, hope it helps!

[@NathanBuildsRobots]

Nice work. I have designed a fair few ducks, would love to get some feedback/advice on them. I try not to change directions or reduce area much more than 50% of the blower outlet area.

[@jamespray]

Sounds like you're already ahead of the curve! The CAD channel in Rolohaun's discord is a great place to get duck feedback 😆

[@mitsubishimakes]

So good

[@NathanBuildsRobots]

Last question, what do you think of the bambu lab part cooling solutions? The ducts are asymmetrical, they don’t collide in the middle, rather they are offset and create a vortex. The consequence of this IMO is that you sacrifice part cooling performance in the exact center of the nozzle, but get improved cooling in an annular area around the center. I can provide pictures if needed. Also, side mounted fans m, what u think, good idea or no?

[@jamespray]

I am not familiar with the BL part cooling design specifically but I think some asymmetry in the nozzle aim is good. If the airflow collides head-on right at the nozzle, I believe that's actually going to make the nozzle the area with the lowest airflow because the air has to slow down and change directions 90 degrees to exit the area. It's important for cooling air to have an easy exit path. Most of my designs place the nozzles at 120 degrees from each other so the airflow collides at the nozzle but each air stream only has to turn a 60 degree corner to flow away. It is also good to have a nice "splash zone" around the nozzle, because the molten filament is always going to be moving away from the nozzle in some direction and the larger the area being cooled, the longer that filament spends being cooled before the print head moves away.

[@jamieclarke321]

Watching this be done in SketchUp is soo painful

[@jamespray]

It's cool, next time I'll break the mech pencil and paper

[@SantiagoLuz]

Looks like those skilled artists making nice art in mspaint 😅

[@cybyrd9615]

​@@jamesprayI wanna do this is VR

[@jamespray]

@@cybyrd9615 I kind of do too, but I think I'd get seriously ill

[@cybyrd9615]

@@jamespray well that problem stopped for me with the Varjo Aero also unless you're moving it's not a ptoblem

[@salty_systems]

Hey James this is a great Video 😊 There are some new things i haven't seen yet 😊 What do you think about creating a Vortex or turbulences "at" the nozzle-tip?

[@jamespray]

Thank you! I am not an expert on nozzle design, but I'm pretty sure you get turbulence whether you like it or not! I do think it's better to aim airstreams so that they can easily exit the nozzle area, versus aiming them right at each other, but there are many valid approaches and only testing really shows what's best.

[@MrHeHim]

Awesome video!! i personally treat it like a game and don't look for advice and just figure it out myself, then i go look for cheat codes after i "beat the game" 😅 TONS of great cheat codes here. The 11 degree grade is something i wasn't ware of, but i did "naturally" find certain things on my own, like the curves are less restrictive when C shaped and not cupped and a little backpressure/resistance in the ducting helps even the air flow to multiple exits. But im sure there's better ways, just that the blowers typically have cupped in blades to accelerate the air flow that tends to push the air to one side. My guess if the fins where straight then the air flow would be even coming out.

[@jamespray]

Thanks, I'm glad you got something out of it! The uneven airflow coming out of blowers is a result of the air trying to flow outward all the way around the impeller and getting accelerated around the housing wall as a result. Blowers can actually have forward swept, straight, or backward swept vanes, without changing that aspect; the vane direction has other implications like the balance of static pressure vs CFM (straight vanes also accumulate less dust over time).

[@hal_wayland]

I love the theory tips but I can't help but think this would be a great case for using subdivision modeling in Blender instead of using CAD.

[@jamespray]

I would tend to agree. Most CAD suites aren't *as* adept at this sort of surface modelling; I think a mesh editor is a better starting point. But most tools can handle it if you know how to go about it. (Also, I'm obviously biased toward mesh editing.)

[@NathanBuildsRobots]

Last last question, what about silent or low noise duct designs? Axial fans kind of suck for part cooling applications. But can produce good results with stator blades.

[@jamespray]

So far as I know, most duct noise comes from turbulence. Efficient ducts are efficient because they make less turbulence, so the more optimized the duct, the quieter it should be! Stators can be noisy if the vanes are very close to the fan blades; I know that can be reduced a little by increasing the gap, but in general I would not use an axial fan for forced-air duct applications unless I absolutely had to.

[@siriusscepter]

Glad I watched till the end, was about to ask you to make me a duct 😂

[@jamespray]

[@hernigti]

Can you give some tips, where to target the center of the tip of the duct respect to the hot end nozzle? Why some have cuts to the end? I dont want to cool the nozzle hot end. Thanks for sharing your knowledge. I m doing for 2 5015 blower fan setup with e3dv6 for argentinian diy printer ( makerparts)

[@jamespray]

There is lots of room for experimentation with nozzle placement, and I am not a master. What has worked for me is aiming the nozzles as close as possible to the very tip of the nozzle without actually blowing directly on the heater block/sock. I also think it is good to avoid aiming nozzles straight at each other -- offsetting to either side or angling slightly apart seems to give better results. But every setup and hot end is different, so try different things and see what happens!

[@NathanBuildsRobots]

Couple questions, what do you think about guide vanes inside of the duct? Personally I hate them, while that can be used to even out airflow, it’s a poor solution as it adds unnecessary resistance

[@jamespray]

I wouldn't say I hate them, but I view them as highly situational for the reason you mention -- more perimeter creates more resistance to airflow. In ducts that must make tight, awkward turns due to other hard constraints, they are occasionally the lesser of two evils (since they can reduce turbulence around tighter bends).

[@ThePrintHouse]

Funny. Just yesterday I modeled my duct work. absolutely no engineering involved. Just made it work. But now the algorithm is suggesting this stuff to me. Funny to see this uploaded and its only 12 days old lol.

[@jamespray]

It's kind of like when you buy a blender and then all your ads are blenders for while, lol. Thanks, I only needed one!

[@ThePrintHouse]

@@jamespray exactly. I mean its suggesting this stuff to me because I looked up 1 video about it, so I'm not really surprised, but just funny I'm now getting suggested a very recently posted video :)

[@matthiasmartin1975]

That was funny and informative. Speaking of inappropriate tools: I created a series of ducts using Blender - a combination of low poly modeling and a stack of modifiers produced nicely organic looking ducts. Later I also figured out lofts in Freecad...

[@jamespray]

I keep thinking about trying Blender again ... from the looks of it, if you can make lofts work in FreeCAD, you can probably do anything 🤣

[@ballbous]

@@jamespray Freecad Lofts aren't so bad, just stay inside part design workbench and it's just as easy as fusion 360 imo. FreeCAD gets way more complicated(and powerful) when you start mixing workbenches.

[@oxorade979]

Hi James, for secret 2, when you talk about axial fans only tolerating high ratios, are you referring to the outlet to inlet ratio ?

[@jamespray]

Correct! In fact, I have done additional testing since making this, and I found that axial fans don't tolerate any reduction -- the outlet / inlet ratio times the rated CFM is the best-case limit for how much airflow you'll get out the end (e.g., if the outlet is half the inlet, an axial fan will likely deliver no more than half its rated CFM).

[@davidbrennan7422]

Yes, I think more so than radial fans, axial fans rely on entrainment to increase airflow. Some of the air on the inlet side is traveling radially towards the fan disc and any obstructions of the inlet obstructs this flow, producing the overall effectiveness of the fan.

[@dennisbuns]

for some reason I remember the actual discord chat lol, great video tho !

[@jamespray]

I have seen that conversation close to word for word multiple times, with different people involved each time. It's just one of those things

[@alexnguyen7613]

Hi mate great video! What CFD software do you use?

[@jamespray]

Thanks! I don't use CFD software at the moment, just empirical testing, but I am considering learning SimFlow...

[@cali-co6428]

god damn i love this man

[@jamespray]

Haha, thanks! Glad you enjoyed!

[@Embassy_of_Jupiter]

Have you ever seen the solid state fans by Frore Systems? I wonder if they could be useful in 3D printing

[@jamespray]

I have seen those! The form factor might make it tricky, but maybe...

[@jeremiahwennermark5926]

Does anyone happen to have a link or info regarding the hoses shown with the CPAP fan setup at timestamp 2:53???

[@jamespray]

You can get CPAP hoses on Amazon. Search for "CPAP hose"! You can even get them in black, like I use, or other colors.

[@shawnabtey8315]

Great video but a few questions. 1. When you say reduction % wouldn't it be 1-outlet/inlet*100? Because that would be the percentage then that it was actually reduced vs without the 1- you would have how much is is left of the original? 2. When testing different reduction values do you take and do that with a straight design or do you have it instead follow the flow that you would actually have. I would imagine it is the later since then it would more in tune with your actual results you get with your final but wondered 3. How would you manage reductions down when there is very little distance between inlet and the outlet of the duct? Do you use space in all directions to kind of length the distance you have to reduce? Or would you increase your distance by potentially moving where the fan is at to give more distance?

[@jamespray]

Great questions! 1: it doesn't matter which you use as long as you use the same math for testing and for modelling your duct. The equation shown is what I use, but either works. 2: I test with a straight duct with at least golden rule reduction rate to get an idea of "best case" performance numbers. Then I tune the duct design with nozzle size variations afterward. It's really hard to know for sure until you start running overhang tests on your actual printer. 3: It's always a balancing act, and sometimes you just don't have much room to maneuver, but moving the fan is often a good way to get some room for smoother ducts and slower reductions. I had to do this for the hot end cooling duct on the design shown in the video -- as originally placed, there just wasn't room for a smooth bend, so I shifted it out 3-4mm and it made a huge difference.

[@kornephoros]

I was wondering: knowing the parameters of the fan that you're gonna use, do you know if there's a way to simulate the amount of chocking that can be applied to the duct (in order to skip the empirical test and save material)

[@jamespray]

If you don't want to test empirically, the best option is CFD, but then you do have to learn CFD. As a rule of thumb, though, I've discovered since making this video that at least for axial fans, you shouldn't choke them at all. Whatever % reduction you use is roughly the % of the unchoked CFM you will get out the end -- in the best case. Yikes!

[@kornephoros]

@@jamespray yeah, axial fans are no good for part cooling fans most of the times... Rn I was thinking to install a remote 7530 blower fan via a Ø22mm cPap tube... Is probably gonna be a little too overkill, but we'll see. And I guess I'll have to learn a bit of CFD🥲

[@jamespray]

@@kornephoros My view is, as long as powering/controlling a big remote cooler isn't a problem, there's no reason not to have more than you need. Then you can underrun it and it'll be quieter, probably!

[@shamancredible8632]

Normally I say nothing, but this is like the 50th video where someone is trying to explain something I want to learn, and their microphone is so quiet I need to turn my massive sound system TO MAX just to barely hear them. It's current year, you own a computer with video editing software that shows volume levels. No excuses.

[@jamespray]

Yep, it's bad. It was also my first foray into cutting videos with separate voiceover. Didn't use a good mic, didn't understand levels. I learned from it and if you care to check it out, I believe you'll find improvement in my content since. Cheers!

[@3lemental]

Ive watched this a couple of time, but still don't get it. I have the fan where I want. But mind is drawing blanks on how to make the duct. Specifically the mid sections and the guide rails for the loft

[@jamespray]

To be honest, that is one of the tougher parts, especially if you're new to it, so don't be bummed if it takes some time. Working from the ends to some midpoint usually helps me find the right line(s), breaking the duct into multiple parts and solving each area on its own. It also helps sometimes to have the print head (or whatever) in front of you for reference, so you can pick it up and turn it around instead of only viewing it on the screen.

[@DarkestBunny]

Hi! Where discord? Also, would you mind sharing your duct design resources? I think i got the basics out of a HVAC manual but i feel i might be lacking.

[@jamespray]

Hi! Re: Discord, seek (in the places one might expect) and ye shall probably find! And I'm afraid I didn't save my reference links, but I got all the mileage I needed out of googling things like "HVAC design rules of thumb" and "... guidance" and similar. Hope that helps!

[@DarkestBunny]

It's not on your website, it's not on your youtube channel, please give link! Accidentally already joined a singer's discord by mistake! Also, from what i understood you are still learning about it, so that means you are still have current sources, right?

[@martonlerant5672]

...as far as lofting goes, if you are using something like blender, you can do a more precisely narrowing organic shaped duct, by sweeping along a curve.

[@PureRushXevus]

thanks for the tip, I was trying to get a good result with a screw modifier and adjusting the origin but wasn't getting anywhere. using a bezier curve makes a lot of sense

[@mikhaildavidenko3841]

I wanted to give you more than 1 like, so here is my comment with my extra like!

[@jamespray]

I extra like it!

[@SianaGearz]

I heard a rumour that a little obstruction isn't always bad. That vanes, although they cause friction and can redue airflow, can help preserve or accumulate pressure by suppressing turbulent flows which can block the airflow. Is that about right, and in which situations is it relevant? Also can you incorporate venturi effect into the nozzle tip? Which are the best cheap ways to test your duct, nozzle or mockup?

[@jamespray]

Thanks for the Qs! Some As: Obstruction (from vanes) isn't good, but sometimes it is better than the alternative (a duct with corners or sharp bends that cause turbulence). Vanes are worth trying if you must run a duct around a corner that creates an inner radius much smaller than the duct width, because vanes create smaller-width "sub-ducts" that effectively get a larger bend radius in relation to their size. (Large bend radius = smoother airflow.) Ideally you design ducts that don't need vanes to flow smoothly, but toolheads are rarely ideal places! Yes, you can definitely entrain air with a nozzle. You actually can't really avoid it, but you can encourage it by designing a nozzle that allows for smooth airflow coming from behind itself. Of course, entrained air isn't moving as fast and won't be as effective for cooling, but it's not hurting cooling either. And lastly, I think the cheapest way to test ducts is just to print them and see how they do. They tend to be small parts, so it doesn't set you back much filament to try and iterate. I have boxes full of them

[@GabrielAlejandroZorrilla]

Isn't there a formula or something to know the optimal outlet cross sectional area? Avesome video!

[@ulamss5]

this video induces motion sickness lol

[@jamespray]

You and me both

[@lukewoolfson]

Nice guide. How well did it perform though?

[@jamespray]

Thanks! I am still waiting on the hot end to go with this duct, so I haven't been able to run test prints yet. This was a design I happened to work up well in advance of the hardware availability. Water tests with the prototypes look good and the duct itself is quiet with a wide-open nozzle variant, which are good signs. Still, I can't render final judgment until I can start running test prints and A/B results against prior designs with various nozzle geometry.

[@jamespray]

I did post a short (the "cursed" duct) vid recently that features this design, but it's not really a showcase of how well or poorly the duct itself performs, since it's using a high-restriction nozzle intended for a high-pressure air source, where smooth duct airflow is more of a secondary consideration.

[@JohanFasth]

If I don't remember wrong from the book "Gas turbine engine" the 11 degrees is maximum for divergent flow without the air break away from the surface. Interesting that it's useful for convergent channels to. Or do I remember wrong, it was 30 years since I read it. :)

[@jamespray]

It is also 11 deg for divergent channels, so you remember right! I have found that it is definitely just for 11 deg in one dimension, though. If you want a really efficient cone reduction, for example,. where it's converging from all angles, you need to go much shallower than 11.

[@JohanFasth]

@@jamespray Ahh, good. And thanks for letting me know I'm not to old to remember things! 😁

[@TheMastaRob]

Great video, even if watching you use sketchup made my eyes bleed lmoa.... question though, have you tested two outlets vs one? I see you talking in comments about offsetting outlet angles 120deg, so the flow doesnt cancel, but I cannot find anyone who has actually tested this vs a single outlet. Well apart from one dude who wrote a quick reply on the Duet forums saying of all his tests, simple single outlet is better. Any thoughts yourself? Aside, I use a laptop CPU fan (Lenovo B450) and then design the scroll housing in with the ducts. More powerful, quieter, etc. Just upgrading my extruder and dithering over 1 vs 2 exits now!!

[@jamespray]

With the caveat that "It depends!" based on many factors and I wouldn't go so far as to say any one approach is best for ALL situations, I have found in testing that two well-designed and well-placed ducts will give better cooling results than a single one, all else being equal. (And yes, I did test this!) The main upshots of a single duct are simplicity and minimized duct perimeter (theoretically less losses) vs multiple ducts, but the price you pay is more directional (or you could say anisotropic) cooling, where features favorably aligned with the airflow are cooled more than those facing the other direction. A good overhang test with steep features facing multiple directions will show this effect clearly. Multiple ducts, meanwhile, require more space on the toolhead and probably more work to design, but the reward is more even distribution of cooling and less variation based on the geometry being printed. My 120 degree design is a compromise, but so is everything else -- there's no avoiding them, only picking the ones that best meet your use case. I hope that helps!

[@TheMastaRob]

@@jamespray Awesome, thanks for the detailed reply!! Really appreciate it. Yes, I have seen this effect with just one outlet, overhangs on the opposite side of the nozzle do suffer. Sigh, guess I'll design in two then. At least it looks cool!!

[@TheMastaRob]

​@@jamespray Been having some annoying stringing problems so I tried different duct configs to see if that made any difference. It didnt, but I was able to confirm two exits at 120deg apart gave much better overhang performance than just one!

[@jamespray]

@@TheMastaRob Cool! (lol)

[@SirRobinII]

Great video, but I will probably skip half of the secrets and try to use a loft. Since its a pain in the ass to make 3d sketches with this program.

[@jamespray]

The secrets are totally compatible with lofts, especially if you don't try to make the entire duct out of a single loft but use several to manage the shape more closely. And either way, I don't think you could apply them *all* on any one thing without going out of your way to be a tryhard. Have fun and good luck!

[@Embassy_of_Jupiter]

if it walks like a duct, if it quacks like a duct...

[@nathansmith9820]

I have made several mistakes you identify here. I asked around on forums and nobody could identify the problem. This video contains precious practical knowledge (distilled down from arcane formulas or passed down as tribal knowledge?) i have a question : isn't "change the cross section of the duct as gradually as possible" and "keep the duct wide until the end" conflicting commandments?

[@jamespray]

Thanks! That's exactly why I made it. As far as those two commandments, it really depends on how much duct you have to work with. The second one about maintaining a wide duct as long as possible really kicks in only when the duct is very long, which isn't common on toolheads except sometimes in CPAP situations. All of the guidance really applies as a matter of balance and practicality -- you'll rarely have all the space and freedom you need to design a perfect duct, so it's all about knowing what will get better results within the constraints you have :)

[@mitsubishimakes]

Can I please cameo in the "discord argument" section of your next video?

[@jamespray]

haha, sure!

[@Parkinsr19]

Made a mistake and didnt read the title. I heard ducks.....

[@jamespray]

Ducks are an underrated 3D printer component, for sure 🦆

[@madvelila]

I wish I knew how to use cad. 😢

[@jamespray]

No time like the present! Or tomorrow! Or whenever!

[@madvelila]

@@jamespray yes, it is really the lack of motivation or priorisation.

[@jamespray]

@@madvelila I feel you! I usually wait until I have a clear use case for a skill I don't have. Something I really want that I can't get without learning it. Then the motivation/priority takes care of itself.

[@madvelila]

@@jamespray same for me uaually. I have tried to start learning something that I thought would be nice thing to learn. But with out real motivation from real need I have quickly noticed that I have to push my self to learn that thing and then quited after awhile

[@kingoftennis94]

this is the most complicated video I have seen

[@jamespray]

you were warned

[@RyanSeacroft]

hmm discord light mode.....

[@jamespray]

[@Jhelm]

What kind of monstrosity is this cad program. Hobby solidworks is 10 bucks a month bro. A loft with guide curves would be all u need.

[@Jhelm]

This is like creating an animation using powerpoint slideshow 😂 something i did when i was 13.

[@jamespray]

@@Jhelm If you think this is bad, you should see my flip phone 🤣

[@black99rt]

Yeah this is downright masochism. If you are this good with SU you would be a beast with SW.

[@AndrewBoraas]

"guide" *is a absolute sketch up wizard*... guide in fusion?

[@jamespray]

There are great Fusion modelling tutorials out there from people who actually know how to use it (i.e., not me). I can't teach you to model a hollow noodle in Fusion, just how to make your hollow noodle the best it can be 😎

[@BennXdesign]

Sketchup?? really?

[@jamespray]

Free, local / non-cloud, and I have 13+ years of experience working in it, so yep, really. The best tool presents the lowest barrier between brain and output, and for me that's Sketchup. Don't be scared! It's just a mesh editor. And I'm not telling you to use it.

[@BennXdesign]

@@jamespray I understand that, I am a heavy user of SketchUp myself...but for building my wood decks. I mean, I understand you are proficient with the soft, but sometimes you have to take the right tool to do the job. you don't paint on a A4 size canvas with a paint roller... well, actually you could (by using only the edges of the roller??) but that would be incredibly inefficient, and it will give you very little control. I had a few hours to kill today, and I tried to model a fully parametric duct for my printer on rhinoceros and grasshopper. With this, I can fine tune pretty much all the points you went across in your (still very good and pertinent) video, and modify every dimensions, every angles, in real time. that's the thing, there is a immense gap between the advices given in your video (all very good), illustrated by an incredibly inefficient working process. it is painful to watch, to be honest.

[@jamespray]

@@BennXdesign It is true that duct design (complex curves generally) is one of the worst possible things to do in Sketchup, and I won't contradict your point that it's inefficient to do it the way I'm doing it. The ease with which I can do almost everything else balances out this weakness for me, but only because ducts aren't something I'm designing all the time. If ducting or organic modelling comprised more of my work, it would make it more worthwhile to expand my toolset to cover that need. As-is, the ROI for picking up another CAD or modelling program isn't quite there at the moment for me. In the future that may change 🙂

[@BennXdesign]

@@jamespray If its a one off, I can get behind this, but still, painfull to watch 🤣 It remembers me the japanese (then worldwide) trend of oekaki : drawing incredibly good paintings with... microsoft paint. they are doing this just for the performance of doing something amazing with a crappy and limited tool. As an artistic performance, I am all in favor of it. with the level of technical achievement you are doing in this video, it is pretty much the same (I would even argue that it has some artistic feeling, but that's only me ^^). If you like polygon modelling, you may give blender a shot. super quick to pickup, and free. still not quite the right tool for measurement sensitive CAD, but yet perfectly doable. I am not super super fluent with blender, but still I would design a duct pretty efficiently with it, I am quite confident on this, and if you use a spline for the airway, you may even make it editable in a parametric fashion. You may even be able to test it with a fluid simulation! (IDK if the results would be physically correct though...)

[@protator]

Why are you torturing yourself with Sketchup? Two whole hours to loft two rectangular shapes along a guide curve with a bit of rotation here and there? That's insane, you need to value your time more. This is a ten to thirty minute task in a proper software suite.

[@jamespray]

I value my time very highly (it's an often-missed factor, though). SU does save me a lot of time outside of this particular worst-case use case, but only because I have a lot of practice and good workflows. Many of my reasons for sticking with it revolve around the deficiencies of other products in the same space, my comfort level with it, and the fact that I'm a curmudgeon I'm not recommending it to anybody else!