A part cooling fan is something which has
become a necessity on our 3D printers since the days we started printing parts in PLA
or similar materials. But does this cool stream of air that helps
us print very detailed models maybe reduce the strength of our parts? This is exactly what I’ve tested for todays
video and we’ll also learn how different fan settings change the accuracy of our printed
parts, using a 3D scanner! Let’s find out more! Guten Tag everybody, I’m Stefan and welcome
to CNC Kitchen! This video is sponsored by Audible. When the first RepRap printers emerged in
the homes of makers the material of choice was usually ABS where cooling the printed
material down too quickly will cause layer separation and is therefore not really wanted. Only after materials like PLA have been becoming
more and more popular, fans that rapidly cool down the printed material were needed to print
fine details and overhangs. In contrast to ABS, PLA also doesn’t have
the tendency to crack between the layers or to separate from the build platform when being
cooled. Nowadays people seem to even challenge themselves
to get the most amount of cooling for their printers. But this raised the question for me if too
much cooling could actually be detrimental for the strength of our 3D printed parts because
it reduces the temperature of the previous layer and gives the new material less time
to bond together. If you have ever watched one of my filament
test videos, you might know that I usually print two of 3DMakerNoobs temperature towers
where on one I vary part cooling in steps and then chose the one with the lowest amount
of cooling but still good print quality. This is because I always assumed that less
cooling also means better layer adhesion. Though, I never tested this assumption in
a controlled manner, so this is what we’ll do in today’s video. This video is actually one part of a video
series where I want to test the influence of different settings on the strength of our
3D prints, with the goal of finding a combination of parameters for maximizing the toughness
of our parts. If you don’t want to miss that, then make
sure to subscribe to the channel and ring the bell to be notified of upcoming videos. Due to popular demand I won’t be focusing
today’s tests solely on PLA, but also included PETG. Unfortunately, this material caused me quite
some trouble as you’ll see later and I actually had to redo all the tests with a second filament
brand. This is also where today’s sponsor Audible
comes in because they don’t only entertain and educate me during travel and work, their
support also enables me to invest more time in investigations like these. You guys probably all know Audible by now. Audible has the biggest selection of audiobooks
on the planet, and it got even bigger with their new Audible Originals which is content
that was especially created for your listening experience and which you can’t find anywhere
else. I really need to tell you about an audiobook
I recently listened to while on a business trip to which you can also listen for free
with a 30 day trial by visiting or texting CNCKITCHEN, all one word, to 500500. The book is called “Skunk Works” by “Ben
Rich” in which he talks about his years at Lookheed and the development of the stealth
airplanes like the F-117 and the SR-71 blackbird. As an engineer myself I was so drawn into
his stories on all of the technical challenges that they had to overcome. He even tells interesting stories about tiny
details like the titanium airframe of the SR-71 becoming brittle because their tap water
was chlorinated during a couple of hot weeks in the Californian summer. Or about the political situation during the
Cold War and how they were still able to basically source all of the titanium used in the SR-71
from the Soviet Union without them noticing! I seriously couldn’t stop listening and
if you’re also just the slightest bit passionate about aeronautics and engineering definitely
listen to this audiobook! Visit or text CNCKITCHEN
to 500500 to get this or any other audiobook and additionally 2 Audible originals for free! No risk involved and you can even keep all
of your audiobooks forever if you, at some point decide to cancel the service. Thank you Audible for sponsoring this episode! I usually test the strength of 3D printed
materials with these dogbone tensile test specimens that I print at 100% infill in a
horizontal and vertical printing orientation. The horizontal specimens tell us how strong
the base polymer is, because we load it in printing direction where we don’t have layer
lines reduce strength. The vertical specimens on the other hand test
exactly the opposite and this is how good subsequently printed layers stick together. Ideally the strength would be similar but
the strength of the standing specimens is usually at least 50% lower. Unfortunately, these specimens gave me quite
some trouble in the past because if the two mounting yaws of my test machine don’t align
perfectly, the coupons will see a slight bending load instead of pure tension and therefore
fail prematurely and invalidate the results. For this reason, I designed a new set of samples
that feature half spheres on both ends and which are still nicely printable. With the suitable sleeve and holder they form
a spherical bearing and minimize bending loads on the samples. This greatly reduced the scatter in my results
and will be the new shape I’ll use for layer adhesion tests. For this series of tests, I’ve printed 3
samples for 5 different fan settings which makes 15 samples in total for each of the
materials. The fan settings I used were 0%, 20%, 40%,
60% and 100%. I printed all of the samples in a single print
job where each group of samples was finished one after another. In order to compare the strength values at
different fan settings to a baseline, I also printed 3 horizontal dog bones but these ones
only at one single fanspeed since I boldly state that there won’t be a significant
change regardless of the cooling. Please bear in mind that these set of results
are probably not 100% true for any other filament manufacturer and printer, because slight changes
in the setup can have a big impact on the results. So, use this more as a guideline and look
at the trends. I printed all of the samples on my Original
Prusa i3 MK2.5 at a layer height of 0.15mm with 2 perimeters at 100% infill. At the time of printing it was summer and
I had between 26 and 27°C in my office. I printed the samples in SpoolWorks PLA, DasFilament
PETG and PrintaMent PETG from Aprintapro. In general, the parts looked nice, you were
only able to see that the parts with no cooling did show some problems in the overhanging
regions. Let’s start with the test results of the
PLA parts that I tested one after the other on my self-built Universal Test Machine. The baseline that I printed lying, which should
represent the ideal and maximal material strength failed at 59 MPa with only very little scatter. Next we come to the layer adhesion samples
where the ones with 100% fan speed failed at 30 MPa, so only half of what the horizontal
coupons were able to bear. At 60% and 40% fan the results were only slightly
better with 31 and 33 MPa of tensile strength. At only 20% cooling fan we can see that the
layer adhesion already improves and the specimens failed on average at 38 MPa with is already
64% of our baseline. Lastly, we have the samples that were printed
with no cooling fan at all and those were able to bear an astounding 48 MPa which is
over 80% of the reference! This shows that we can increase layer adhesion
by more than 50% by decreasing the amount of cooling we use. Next I tested DasFilament PETG. The tensile strength of the horizontal specimens
was 53 MPa with basically no scattering at all. I won’t go through the rest of the results
though, because they were all over the place and something was extremely wrong with them. There was no patter recognizable and the strength
test results were more or less random. I’ve been printing a lot with this material
over the last years and was usually very happy with it, but here either the material or my
printer had quite an issue. As I already said in the beginning, I didn’t
want to leave my results like this so I took the time and printed another set of samples
in PrintaMent PET-G from Aprintapro. The sample looked really nice and very well
extruded because they still kept a kind of translucent look after printing. The baseline tensile strength was slightly
lower than the one from DasFilament with 49 MPa. Again, there was basically no scatter between
the tests. The layer adhesion strength of the PETG when
the fan was 100% on, was around 21 MPa which is only 42% of the baseline and therefor worse
than what we have seen with PLA. 60%, 40% and 20% were all in the same region with 26
MPa of ultimate strength, which is an improvement but still not too well. Only the samples printed with no fan at all
were able to bear almost 41 MPa before they failed which is only less than 20% lower than
our baseline and therefore very impressive! The results kind of confirm what my suspicions
were in the past. PETG still seems to be a bit more tricky to
print and therefore the results scatter more. Secondly, the material should be printed with
less than a full cooling fan, because this increases strength but the print quality,
as we will see in a minute, still remains very nice even at settings below 50%. For both PETG and PLA we have seen, that the
layer adhesion is always significantly lower than the base material strength. Still we can influence the adhesion with the
amount of cooling we use and less is usually better. In order to check the influence of part cooling
on print quality I have also printed several 3D Benchys at fan speeds of 0%, 20% and 100%. Ideally, I want to go as low as possible with
cooling but at some point the prints really start to look horrible. Both PLA and PETG prints without any cooling
looked like they were melted in the sun and I did already increase the layer time to over
20 seconds. That’s actually a setting which can help
you print nicer at lower fan settings because it will slow down your prints when a layer
takes less than 20 seconds, increasing the natural cooldown time. Chuck from Filament Friday even stated that
you can print PLA without any fan at all but he was using an Ender 3 for his tests. The thing is that all of these CR-10 variants
have a horrible fan shroud design where the fan that cools the heatsink also directly
cools your hotend and the print. So 0% fan on such a machine is still some
cooling and will decrease layer adhesion. The Prusa on the other hand has two totally
separate cooling paths where no cooling air from the heatsink touches the print and therefor
0% part cooling fan also means not cooling at all. That’s also quite important for higher technical
materials like Nylon, PC or even ABS. Just keep that in mind. I currently have a professional 3D scanner
from the company GOM at my disposal that I can use for a couple of weeks for my research. For this reason, I also scanned all of the
3DBenchys to find how they really deviate from the ideal geometry. I had to coat the translucent PETG prints
with some chalk spray to make them scannable. For the comparison of scan to reference geometry
I used GOM Inspect which can actually be downloaded and used by anyone for free. I will show you a couple of deviation plots
in a second. The way you read them is pretty simple. The color is a measure of the geometrical
deviation of the real scanned part to the nominal STL file. Blue is negative and this means that the real
part is smaller or inside the reference. Red means a positive deviation, so the scan
is outside of the reference, so basically more material at that location. Now to the results. I guess for the prints without any cooling
fan we don’t really need to talk about deviations, because especially the PLA one just looked
horrible. The thing that was interesting to see though
was, that you can clearly see the sagging of the horizontal surfaces on the PETG part. The parts with 20% cooling already looked
way better and the PETG one was basically looking perfect and you almost can’t see
any quality problems even at this low level of cooling. The PLA Benchy at 20% cooling is even more
interesting because it looks okay from one side but the other side, which is the side
facing away from the part cooling fan is totally deformed. So it seems as if 20% cooling with Prusa is
still on the low side and it might be an interesting project to optimize the fan shroud so that
it’s more even in all directions. If you guys are interested in that topic there
is actually a cool blogpost from Nophead, the father of the Mendel90, on the uniformity
of different fan shroud designs. I’ll leave a link down below! Both of the 100% prints look perfect on first
glance, only the 3D scans reveals that there was some deformation in the highly overhanging
areas. So nicely printable doesn’t always mean
geometrically accurate. If you by the way have other project ideas
I could do with this 3D scanner from GOM then leave a comment down below! In summary, what have we learned? So I think we have nicely shown that the layer
adhesion of FDM parts is significantly influenced by the amount of cooling air that is used. Less cooling usually means better layer strength. Unfortunately, at some point the looks of
our prints is influenced and at low cooling values print quality suffers. PETG is better in this regard than PLA. If you want to improve your part-strength
than play around with the amount of cooling and find the lowest setting possible with
still sufficient part quality. Slowing down your print and using minimum
layer time will help you with print quality. Also, the experience with the first PETG samples
showed me again how unpredictable 3D prints can actually be. Without proper quality control measures our
3D prints shouldn’t be used in safety critical applications. But what do you think? Was this something new for you and how are
you handling layer adhesion problems? Also let me know what other print setting
you’d like to see me test in the future! Leave a comment down below! Thanks for watching everyone. I hope you’ve learned something new today. If you did, then leave a like and make sure
that you’re subscribed for upcoming videos. If you want to support my work then head over
to Patreon or help me out in other ways. You can also support the channel and get a
free audiobook if you visit or text CNCKITCHEN to 500500 and sign up for
a 30 day free trial. By the way, also check out the other videos
in my library! There are more you might like. Auf wiedersehen and until next time!

100 thoughts on “MUCH STRONGER 3D prints due to LESS COOLING! Testing PLA & PETG”

  • With cura I print petg with either 0% or 20% and bridging and layer over support at 100%. It would be nice to do that also for outer walls and bottom layers (except for 1st) if possible (or specifically any overhang), so I would get the best of both worlds. Though that's not possible in cura (which I prefer because of the many settings, like I slow down print and acceleration on outer walls to have a good look and good overall speed)

  • It would be interesting if you can redo the test, but with a enclosure, so the cooling air temperature will be higher, closer to the glass transaction temperature.

  • Hello, It's great video!!! Can you test PLA from anycubic for example, in my experience it is very strong. Or maybe transparent filaments will be stronger than colored?

  • I usually increase print temp to increase layer adhesion. It would be interesting to test how temp affects it and if increased temp can overcome the negative effect part cooling has.

  • Thank you very much for this illusive test!
    As 3D printing is a field of mostly hobbyists (only those share information), many sources are very contradicting. Your (and others ofc) rigor really drives good quality information spreading and clears very important questions. Much needed answers should be found for strength, since quality vision based is really well explored, but repeatable strength tests are VERY rare.
    Since I build mostly usable parts this is of high interest to me – Thank you very much! Keep it going!

  • Your should include the actual ambient air temperature in your results which is being used by the cooling fan. I usually print petg in closed enclosure resulting in 30-35C cooling air. It makes a huge difference if you blow with 15C or 30C.

  • Well thought and professionally executed, with useable results, as always, thank you!
    Of course your results raise additional questions:

    1. How much would the temperature of the heated build chamber affect the layer bounding of ABS/ASA prints? (I think ABS still is very important, because I consider the properties of ABS as suitable for everyday use objects, instead of just vanity objects).
    2. Could PLA prints get even stronger by printing those in a heated chamber?
    3. How far would the results from this tests change, after annealing the 0%, 20%, 40%, 60% and 100% PLA prints? With focus on the question: Could annealing compensate for the strength loss by printing with the fan on?

  • Very nice Stefan! You must try cpe from filamentum. It is the filament with the best layer adhesion i have tested and I use it for all mechanical parts on my 3d printers. Also petg type must be dry!

  • Well done video on a very important topic! I have suspected that part fan speed should be high only during overhangs -now you have proven it!
    Now I wonder if post cured (what they call annealed) PLA parts can get to 'no-fan' strength but use a bit of fan to help overhangs.
    Post cure of PLA functional parts is becoming my standard because heat deformation and creep are unacceptable without it.

  • 0:51 – busted x axis pulley bearing. I've had this too, and, quite surprisingly, the prints were somewhat ok (I'd expect much worse). However, replacing the pulley with a new one made a world of difference in print quality. Also, assessing the situation, I concluded the bearing shatterred due to too high a belt tension, and I've lowered the tension by one or two teeth.

  • FYI, others blamed my results on the Creality leaking fan shroud so I did follow up tests with the bottom of the fan shroud blocked on my Ender 3 and got the same great results with no cooling fan. So my profile does work. Watch at 3:05 here to see what I did.

  • You're definitely right about the Ender3/CR-10 heatbreak fan. I upgraded a CR-10 mini to an E3D v6 and when I printed a benchy when I hadn't printed the cooling fan mounting bracket yet, it was the most horrible result I ever had.
    I'm also thinking this may be related to the watertightness of 3D prints. I used to get some good results with small boxes, but a large-ish boat I just printed is leaking. I'm going to try cranking the temperature up and disabling the cooling fan. It will still be cooled a little bit due to creality's design.
    I really also want to get my own universal testing machine running. So much to test, like creep and fatigue life. My machine is based on a raspberry pi though so I still have some programming to do.

  • I can’t believe my first comment is about the fantastic book you mentioned in your ad! I’m a pilot, and, while not really an engineer, can really appreciate the skills. That is a fantastic book! I’m also humbled to say that I know people who participated in the early days of the Skunkworks.

  • I appreciate this testing. I’ve done quite a bit of fan testing on a few different printers. I find that with PLA, I need 80% (at least) to get a nice 3D Benchy at 60mm/sec on the Prusa MK3 with PLA. With PETG, I can lower the fan speed to 40% and get good prints.

  • I wonder if there is more time for crystal formation when cooling is slower… Or does cooling need to be much slower for that?

  • Great work Stefan, will you also do the stress test after an annealing process? Perhaps that will counter act some of the negative effects due to cooling during the print process.

  • Greight video, thanks. I find 0% fan is best for most prints, but I have it on 20% for areas that are unsupported (bridging), support interfaces, and for very small features where the layer time would otherwise be too short. But I agree, no cooling is usually better!
    Edit: This also applies to TPU!

  • According to my experience it is not only the cooling settings that have an impact on the layer adhesion, it is also the printing temperature, the printing speed, and the geometry of the print. Printing a small and compact geometry like the test samples is a totally different thing from printing large and thin parts (for example only 0,4mm thickness for a wing of a model airplane). This is because at the test samples the printhead keeps the temperature of the actually printed layer quite high because it is always close to it and maybe heats up the already printed material when it prints the neighbouring material. But while printing a large and thin structure all heat energy is able to dissipate before the next layer is printed, so the layer adhesion is much worse. I am sure that you would get totally different results if you printed for example a 10cm tube with only 0,4mm thickness and tested that.

  • I’m quite new at this but have already figured this out. I mostly print ABS because it’s easy to get a nice finish and further refine it. But I’ve got some excellent PLA prints out with much higher temperatures than usually recommended and practically no cooling. Empirically, these are substantially stronger than cooled versions and look much better too.

    Good information here! I like your methodical approach.

  • Great work like always! It would be interesting to know if an enclosure makes a difference here. Even a simple cardboard box around a printer will raise the temperature to 35-40C which is a lot closer to PLA's Tg thus it may improve layer adhesion significantly. It does for ABS so ….

  • i was considering buying a spool of das filament petg for my first petg spool…. not so much now, that layer adhesion was fucked up, any idea why?

  • Maybe we should only use part cooling for the outer layer? Best of both worlds, nice outer layer and good layer adhesion inside.

  • As you already have the installation for annealing, it could be very interesting to see if you can cancel the impact of cooling with annealing. It could be a good way to have both of gemometric quality and strengh performances.

  • This more or less validates everything I've observed so far when printing, but I think there's a few things worth mentioning for newer users:

    – More emphasis on slowing down prints/layers and tuning that. I usually slow down the print for layers below 20-30 seconds, and kick up the fan for layers below 10-15 seconds (depending on filament of course).

    – Super-important: Separate bridging cooling settings! Slic3r, Cura, S3D all have this (not sure why it's missing from Ideamaker, last time I checked), and I set it to full blast when bridging, which is critical on a 0.6mm nozzle! I still sometimes get artifacts as the fan takes time to spin up, but most of the time it works remarkably cleanly.

    – In addition, I think some slicers might offer separate overhanging cooling settings. I don't think there's an ideal implementation of it, where it would adjust the fan speed based on the overhang angle and strictly limit it to just the overhanging region, but I'm pretty sure Cura had an option for it at some point, and it's better than nothing. Overhangs tend to be where I have some trouble, even when following the 45 degree rule, simply since I can't have the slicer crank up the fan for those regions.

    Lastly, having visited your videos on how print temperature affects part strength, I think it would be interesting to visit a combination of the two factors and how to tune them together for different filaments, at least if there's anything new that you find.

    Thanks for covering the topics of 3D printed part strength. There are very few resources on it online, I had to learn most of it on my own, and it's nice to throw some proof at fellow 3D printing hobbyists that there are better ways to do things lol.

  • Have you considered a larger fan outside print area, but close to the printer? A regular tabletop fan or even just a large PC fan to create a more even cooling effect as it seems like as little cooling as possible without compromising print quality is the way to go. Clearly the sweet spot depends a lot on material you are printing

  • Hi
    can you do an impact testing a 3D printed filaments and then scan it with your 3d scanner to see an indentation/ damage caused and plot it.
    Thank you

  • I tried to get a job as a mechanical engineer at Skunk Works. By the time they got back to me (almost a year later), I had already accepted another job. This other job with General Physics eventually sent me back to the rocket site at Edwards very close by. That was about 20 years ago. I worked there for a couple of years. The desert there is miserably hot and windy. I love your informative videos.
    I make parts for function and don't care how ugly they are. I will try to print with less cooling to see if I can get stronger parts.

  • As if wanting to buy a 3D-printer for over a decade now wasn't enough, now I see a 3D-printed SR-71 Blackbird in black… #wants

  • I guess, you have a dual extruder printer, Stefan. Could you try to make a test with the inside printed without the cooling fan, for strength, but print the outer layer with the second extruder and the cooling fan activated to make the surface look good. It would be great if you could try this, because I haven't got a way to accurately test the strength of a part, but would like to find out if it is possible to get good layer adhesion and good looks.

  • Isn't there also another variable to consider: the speed of the air coming channelled from the blower, not only the rpm. A small air tube design for air flow at 20% has for sure a sharper airflow and therefore higher cooling effect then a bigger air tube at the same 20% setting. If this is off the path thinking, please correct me 🙂

  • any suggestions on maybe temp tower test but changing the cooling instead of the temp? or a combination of both?

  • Doesn't the heatbed change the temp of the part cooling air?
    I put a couple pieces of paper towel over the areas of the bed I'm not using and tape them down.
    The bed temp doesn't change as much and I use less electricity, along with cooler air above the bed itself.

  • I hope you corrected your comment about the ender and cr10 printer fans being bas as there is only 1 I own an ender 3 and mate a cr10 and they have 2 fans one pointed forward aimed at the hotend and one on side which has the vent to cool the print surly you should know that being a 3d printing channel… not trying to sound aggressive it just made me cringe hearing that you obviously didn't do your research

  • could you test the strength values of prints with Ironing throughout and those with Ironing on just the top levels against prints without ironing? I've just started using this feature as the end product looks allot nicer but i wonder if it has any effect on the strength or adhesion of the print.

  • So 3D pens have a diferent way of heating the hot end compared to standard 3D printers. I't looks to me like the heater itself is the hot end so does this grant better control over temprature. If we put this on a printer would we still require hot end cooling? I'm not sure but I think this would be a fun topic.

  • Hi, you could use packing to adjust for any offset in the jaws.
    Please would you consider some temp and cooling tests with TPU 👍

  • The problems of 3D printing appear at all times, this video is amazing, I have learned a lot, also in these links there is valuable information.

  • I dont know if you have done it yet but it would be really cool to see you scan parts with varying degrees of mechanical tolerance and then convert those scans to .stl files and print them and then compare. I have used my printer to replicate many parts over the last year but i often struggle with bad tolerances and id love to see how advanced the scanning software is anyways 🙂 Thanks for the video

  • Hey. My name is Vlad. I want to work with you. I will translate and voice your videos on your terms. It will bring you more people from Russia. Since not every Russian can watch your videos, because they are not translated. Your videos are very interesting, so I’m sure that you will have more subscribers from Russia. If this interests you, please contact me in response to this message.

  • Should test to see if you can compensate with temperature. I do, but I don't know how much of a difference it makes. If I'm running 100% cooling I usually crank it up 10*c.

  • Maybe printing with 100% cooling, then reheating the whole finished part afterwards to allow for better bonding, without spoiling the quality of print.

  • what about higher temp+more cooling vs lower temp+less cooling? for example PLA 190degrees+20%fan vs PLA 220degrees+100%fan

  • Thank you Stefan for your professional researches ! As a beginner in 3d printing field I need the information you provide on your channel.

  • Interesting, but if I need a lot of strength for such a print then I could use annealing, or not? Putting it into an oven for a while, and there are no layers.

  • I wonder if this actually affects the strength of bigger parts because of the increased layer time.
    Probably the previous layer will already have cooled down so much that Fan speed doesn't make a difference.

    Great video👌

  • Would be interesting to reprogram the fan control to cool only when printing overhangs and outside perimeters and see how that would affect strength… on one hand, it should not compromise strength on the infill while keeping the outer surface tidy but on the other hand, cracks usually originate on stress points on the surface which this approach would not change from the normal cooling approach… so it would be very interesting to test this out.

  • So, I'm still relatively new to this. But I'd like to ask you and your fans; Do you notice that your settings need to be set differently for different colors? Let's say between natural, red, black and green, and from same manufacturer? ABS as well as PLA. . Also, does it look like there's a happy medium between high and low fan speeds? I call it goldilocks zone.

  • 1) printing in batches of 3 pylons affects the cooling of the material
    2) you can use 0 fan speed but you need to lower the speed

    I have a CNC screw driven repurposed for printing I do not have a fan, but my max speed it 9mm/s. printing in 2mm I have no quality issues. find the speed where you can print with 0 fan speed properly and you can then enjoy the strength of the material

Leave a Reply

Your email address will not be published. Required fields are marked *