Another interesting column might be long term loading to cover, for example, parts that are under constant tension. Are the stiffer parts more or less likely to fracture and fail in these scenarios?
1 Like
Yes. I believe that to be the best. Can’t wait to see if I am right, not sure if an impact test is the best test, though.
1 Like
I’ll have to defer a long term loading test to someone else. CNC kitchen, My Tech Fun, and Thomas Sanladerer on youtube do some tensile strength and impact testing if you want to take a look. However those tests are not as applicable to the mpcnc.
My Tech Fun does try to do creep tests. Load and measure deflection and elasticity over time. All plastics creep under load. Some more than others
3 Likes
I would think the tensile tests are more relevant? Been a while since I had the materials classes so I am really not sure.
1 Like
Welcome and thanks!
Well first of all, I’m not an engineer so you know much more about all of this than me. I’m an IT guy and car guy who makes a lot of things. I love testing stuff and building stuff and spend a lot of time trying to learn how stuff works and behaves, but the engineering side is all self taught.
If I understand things right, the tensile modulus (which is the slope of deformation during the tensile test) would correlate to stiffness but the tensile strength is more about how where it breaks. All the youtubers I mention above test tensile strength and impact strength (and at least My Tech Fun usually does a charpy test if I remember right) but I’m not set up for that.
FDM printed parts are weird because the orientation of the filament layers vs the layer adhesion causes it to be non-isotropic. Add in the CF fiber orientation and orientation of the layers to load is even more complicated.
Also, in the tests above, I printed the standing ones with 4 perimeters and the flat ones with 6. Both with 6 top and bottom. While printed in different orientations, they are tested flat. So in relation to the load applied, the thickness of the walls are:
| A | B | C | |
|---|---|---|---|
| Flat | 2.64mm | 2.64mm | 1.8mm |
| Standing | 1.77mm | 1.8mm | 1.77mm |
| Edge | 1.8mm | 1.77mm | 1.77mm |
In hindsight I should have made the flat ones 4 perimeters as well.
I’m considering redoing some of it. We’ll see. I love this but it is a time burner. I didn’t try to match the specimens for each orientation because i was more interested in the relative stiffness for each filament for the same orientation. But it would make it a little more consistent so we’ll see.
I hope any of this is helpful to some and not just more noise. You cannot go wrong with PLA.
4 Likes
Sorry I read the first post incorrectly. Your test is perfect! For some reason I thought you were dropping the weights on them, not just gradually loading them.
Do you have room to leave a couple of them loaded overnight or maybe even over the weekend to see if there is a significant creep difference between PETG and PLA? I figure a significant load will speed up the process.
Such a cool test. CF petg needs a hardened nozzle and is more expensive, but it does offer more heat resistance, if it offered better creep characteristics that might make it worth it. I have to take a look at the Z shrink you mentioned though, that could be a deal breaker.
I could load a few up overnight I think. PETG is going to be off the charts though compared to PLA. Even at 15# it was pretty creepy.
Just for grins, I’m printing a few samples on edge and also printing a few flat ones with 4 perimeters. I don’t know if I’ll print all the filaments yet or not. It’s a lot of printing 
We’ll see how they compare. I suspect the edge ones are going to be very strong as it has more long runs of filament that are perpendicular to the load applied. But the relative comparison between filaments will likely be the same ratios as we see in the flat ones. The CF filaments will likely be better because there are more strands oriented perpendicular to the load but the pure filaments will probably show the same ratios between them.
There is someone here that annealed 3DXtech CF PETG and was happy with how it turned out. I am not going to try that though. If it helped with the standing type of orientation, it would be really interesting as a choice. As it stands, I didn’t like how much weaker it was standing. I am in Texas so wanted the heat resistance but it was a lot weaker than even PETG in that orientation. The forces on the printed parts in the machine are hard for me to quantify though. I don’t know how much the weaker direction really hurts and how much the stronger direction contributes to offset.
My favorite is the CF PC Blend. It is great stuff. But it doesn’t bridge as well and I’ve seen some slight Z shrinkage. I need to do some prints to quantify how much though. It is printed 285-295C which some can’t do, and it helps if you let it cool a lot before removing from the bed.
1 Like
Just for a preview, I tested the Sunlu PLA printed flat with 4 perimeters, and one printed on edge with 4 perimeters. That means they are very close to the same wall thickness all the way around.
They were almost exactly the same in deflection (rigidity). Both 10-15% less rigid than the flat specimens with 6 perimeters. So I’m debating on whether printing on edge is going to tell us anything. The CF filaments might show some variation due to how extreme the CF makes them super non-isotropic.
I might print a few more and make sure. If it looks like it’ll just be a lot of work for little information gain, I may not do it.
2 Likes
Version 2 of the test
I decided to take the time to test out a bunch of different filaments and compare them for rigidity. Thought I’d share the results in case anyone found them helpful.
Note that these results give a comparison of filaments within the context of the tests performed. It isn’t supposed to be the final word on what filament is the best for your machine, but it can maybe provide another data point to help you decide. Just keep in mind the following:
- These tests use a static load. Pieces may exhibit different behavior under a dynamic load
- While the tests do look at filament from different orientations, how that translates to the loads on the printed MPCNC parts is hard to fully determine
- People have printed MPCNC parts with a lot of different filaments, even some that are for sure more flexible than those listed…and most report they work fine for what they are doing
- The impact of filament rigidity on the overall machine rigidity is somewhat hard to determine, i.e. does a filament that is 30% less rigid reduce the machine rigidity by 30%…likely no
- All the tests were done with the same number of perimeters and infill for the orientation. An interesting test might be to take the most rigid, reduce the perimeters, and see how many perimeters it takes to match the rigidity with a less rigid filament. I don’t think it’s possible to make a PETG part match a PLA part, but you might get reasonably close
- Filament can vary a lot from company to company depending on their formulations. It can also vary by color to some degree. Just keep that in mind.
- Annealing was not tested as it’s not easy to do right and can really distort the pieces. It does help with most filaments though if you want to try it
I changed up the tests slightly for this Version 2 and added a few things. Here are the changes:
- Added “edge” orientation
- Changed flat orientation to 4 perimeters instead of 6 so all the walls are similar (see chart below)
- For flat and edge only went to 25# instead of 35#
- reprinted some samples with higher temps for better layer adhesion
- Added a few more filaments
- Based on the consistency of results from version 1, only printed 2 samples per orientation
Skip down if you don’t care about all the setup info and just want results. Look for the graph with the dark background and the comments after
The Setup
Samples were printed in three different orientations: horizontal flat, horizontal on edge, and vertical (standing). Two samples per orientation.
(Note: I reused the version 1 standing tests for most of the PLA where possible)
All filaments were dried in a food dehydrator, printed from a heated dry box, in a temp controlled enclosure (door open for PLA).
The Filaments Tested
I’ve given the MSRP cost for each but you can often find some of these on sale for a lot less. Because some spools are less than 1kg, I calc the cost per kg to make it easier to compare. Then how many spools you’ll need for 3kg and 4kg. I think you could print all the MPCNC with 3kg but that assumes no reprints and it’s tight.
| Manufacturer | Type | Color | Category | MSRP per spool | Spool Size (g) | Density g/cc | Cost 1kg | Spools for 3kg | MSRP 3kg spools | Spools for 4kg | MSRP 4kg spools |
|---|---|---|---|---|---|---|---|---|---|---|---|
| Matterhackers | Pro Series PLA (PLA+) | Translucent Clear | PLA | $52.00 | 1000 | 1.24 | $52.00 | 3 | $156.00 | 4 | $208.00 |
| Atomic | PLA | Silver Metallic | PLA | $29.99 | 1000 | 1.24 | $29.99 | 3 | $89.97 | 4 | $119.96 |
| Hatchbox | PLA | Black | PLA | $24.99 | 1000 | 1.24 | $24.99 | 3 | $74.97 | 4 | $99.96 |
| Hatchbox | Performance PLA | Blue | PLA | $28.99 | 1000 | 1.24 | $28.99 | 3 | $86.97 | 4 | $115.96 |
| Sunlu | PLA | Grey | PLA | $21.99 | 1000 | 1.24 | $21.99 | 3 | $65.97 | 4 | $87.96 |
| Printed Solid | Jessie PLA | Gun Metal Grey | PLA | $20.00 | 1000 | 1.24 | $20.00 | 3 | $60.00 | 4 | $80.00 |
| Printed Solid | Jessie PLA | Red Ice | PLA | $20.00 | 1000 | 1.24 | $20.00 | 3 | $60.00 | 4 | $80.00 |
| Overture | Eco-PLA | Black | PLA | $17.99 | 1000 | $17.99 | 3 | $53.97 | 4 | $71.96 | |
| Hatchbox | Matte PLA | Blue | PLA | $25.99 | 1000 | $25.99 | 3 | $77.97 | 4 | $103.96 | |
| Prusa | Prusament PETG | White | PETG | $29.99 | 1000 | 1.27 | $29.99 | 3 | $89.97 | 4 | $119.96 |
| Hatchbox | PETG | Chocolate | PETG | $24.99 | 1000 | 1.27 | $24.99 | 3 | $74.97 | 4 | $99.96 |
| Atomic | CF Extreme Black PETG | Black | CF PETG | $49.99 | 1000 | 1.34 | $49.99 | 3 | $149.97 | 4 | $199.96 |
| 3DXtech | CarbonX CF PETG | Black | CF PETG | $48.00 | 750 | 1.34 | $64.00 | 4 | $192.00 | 6 | $288.00 |
| Prusa | CF PC Blend | Black | CF PC Blend | $59.99 | 800 | 1.16 | $74.99 | 4 | $239.96 | 5 | $299.95 |
Print settings and temps for each filament and orientation:
| Flat | Edge | Standing | |
|---|---|---|---|
| Nozzle | 0.4 | 0.4 | 0.4 |
| Layer height | 0.3 | 0.3 | 0.3 |
| Perimeters | 4 | 4 | 4 |
| Top/bottom layers | 6 | 6 | 6 |
| Infill | 50% | 50% | 50% |
| Fill pattern | Grid | Grid | Grid |
| Extrusion width | 0.5 | 0.5 | 0.5 |
Monotonic top and bottom layers.
This gives wall thickness of:
| A | B | C | |
|---|---|---|---|
| Flat | 1.77 | 1.77 | 1.8 |
| Standing | 1.77 | 1.8 | 1.77 |
| Edge | 1.8 | 1.77 | 1.77 |
| Manufacturer | Type | Color | Category | temp flat | temp edge | temp standing |
|---|---|---|---|---|---|---|
| Matterhackers | Pro Series PLA (PLA+) | Translucent Clear | PLA | 215 | 215 | 215 |
| Atomic | PLA | Silver Metallic | PLA | 215 | 215 | 215 |
| Hatchbox | PLA | Black | PLA | 215 | 215 | 215 |
| Hatchbox | Performance PLA | Blue | PLA | 215 | 215 | 215 |
| Sunlu | PLA | Grey | PLA | 215 | 215 | 215 |
| Printed Solid | Jessie PLA | Gun Metal Grey | PLA | 215 | 215 | 215 |
| Printed Solid | Jessie PLA | Red Ice | PLA | 215 | 215 | 215 |
| Overture | Eco-PLA | Black | PLA | 215 | 215 | 215 |
| Hatchbox | Matte PLA | Blue | PLA | 225 | 225 | 225 |
| Prusa | Prusament PETG | White | PETG | 260 | 260 | 260 |
| Hatchbox | PETG | Chocolate | PETG | 260 | 260 | 260 |
| Atomic | CF Extreme Black PETG | Black | CF PETG | 265 | 265 | 265 |
| 3DXtech | CarbonX CF PETG (265deg) | Black | CF PETG | 265 | 265 | 265 |
| Prusa | CF PC Blend | Black | CF PC | 285 | 285 | 285 |
Speeds used:
The Results
For each filament at each orientation, multiple samples were tested and averaged together. The variance between those tests was within reason, only getting slightly higher for the softer filaments at higher loads.
This chart and the following graphs show the amount of bend for each filament, averaged across the samples for each orientation. Note we are not really interested in the actual amount, but the comparison between filaments.
I decided to choose one filament to serve as the baseline to compare all the others to. The Printed Solid Jessie PLA in Gun Metal Grey was very stable in testing and serves as a good representation of a non-modified PLA. The following chart is the % deviation of the other filaments to this baseline.
The % deviation was reasonably consistent across the samples so the following averages those to a singe % and graphs them. I think this gives a good overall look at the rigidity for each filament and allows the best comparisons.
Note: Higher % over baseline = less rigid. Lower % under baseline = more rigid
Thoughts and Conclusions
PLA
There is a reason PLA is the suggested filament to print MPCNC components: it’s amazingly strong, it’s cheap, and it’s easy to print with! It’s main weakness is temperature resistance and sometimes it’s a bit brittle which can cause it to crack. Also note that there are many variations of PLA (like most filament types). Some of those variations are more obvious from the name than others.
- Printed Solid Jessie PLA, Sunlu PLA, Atomic PLA, Matterhackers Pro Series PLA - These performed very well and very similarly. I was surprised the Matterhackers PLA did so well as it’s marked PLA+ on the roll and that usually means it’s not as rigid, but it did fantastic. One note about the MH though, it was a translucent clear filament. I’d have to test a colored filament like black to make sure the results can be applied to these across the board. Also of note, the PS Jessie Red Ice did slightly worse than the gun metal grey. Just enough to note, not enough to prevent using. It just highlights that even the additives to make a color can cause variations in rigidity.
- Hatchbox PLA - This is a great PLA but it is slightly less stiff than the ones above likely just due to the formulation HB uses. I still think it’s viable, but would argue there is no point going out of your way to get it if you are buying new. If you have it already, I think it’s fine to use.
- Hatchbox Performance PLA - This is a good example of a modified PLA. Often companies label this as Tough PLA or similar and it’s simply more flexible, designed to take impacts better. I would not recommend this for MPCNC parts as you get all the negatives of PETG without any of the benefits.
- Overture Eco-PLA - Note this is not the normal Overture PLA. This is a recycled PLA and did not perform well at all. It had terrible layer adhesion and no standing samples could hold even 5#. It was nice to look at though, but just not strong. I would avoid for the MPCNC. Regular Overture PLA is probably fine but I have not tested.
- Hatchbox Matte PLA - Similar issues to the Eco-PLA. It’s just not strong enough. It’s beautiful though. Most Matte PLAs are modified with something and are much weaker than regular pure PLA. In general I’d recommend avoiding matte filaments for the MPCNC parts.
PETG
PETG is the go to second option for a lot of people as they are trying combat temperature issues found in the location of the MPCNC, or they have broken PLA parts and want something more resilient. PETG handles both issues very well, and there are several machines running PETG successfully. However, it comes at a cost. PETG feels almost rubbery compared to PLA and flexes a lot more. I’m not sure you could give it enough perimeters to really offset this characteristic of the base material, but perhaps someone can test that.
- Prusa Prusament PETG - Note this is a white PETG which sometimes acts a little different than other colors. It resulted measurably better than the Hatchbox. However, PETG is still 80+% more flexible than PLA. I can’t overstate how it is just not on the same planet as PLA for stiffness.
- Hatchbox PETG - Slightly more flexible than the Prusa. Very flexible relative to PLA.
Carbon Fiber Filaments
I’ll just discuss these on their own as they really do act a lot different than the pure filaments.
The addition of carbon fiber to the base resin really does stiffen the sample…well sort of. These tests are measuring the flexural strength of the material. That means it’s bending a beam-like structure causing compression on the top side and tension on the bottom side. The carbon fibers travel along, and are oriented with, the direction of the layer path. As long as the forces are traveling along those paths, carbon fiber can help out. Same as in all composites. Since no strands of carbon fiber orient between the layers (the Z direction), objects like our standing samples do not benefit at all from CF, and may even be weaker due to the often weaker layer adhesion CF filament seems to produce. The edge samples have the most laters running perpendicular to the force, so they do the best. With CF filament, more perimeters make an even bigger impact than they do in just pure filament.
- Atomic CF PETG - I’ve use this stuff a long time and I love printing with it. It’s marginally stiffer than regular PETG, but not much. Compared to most CF filament which prints in a matte finish, it maintains more of the PETG shininess. I would say it’s overall better than pure PETG for MPCNC parts, but not much.
- 3DXTech CarbonX CF PETG - So this stuff is interesting. It’s almost as stiff as PLA in the flat samples, and way stiffer in the edge samples (where the CF can help more). However, it’s about the same or worse as pure PETG in the standing samples. In fact, layer adhesion is so bad, and it flexes so much in this orientation, I could only get 5# on the sample before it broke.This goes back to the CF not being able to help at all in the standing samples. So, the effect of orientation of the layers relative to the force is really amplified here. I am really impressed with this filament, and other than the cost, it’s probably what I would use if I had to use PETG for really strong parts, having taken the layer orientation and forces into account of course. Is it ideal for the MPCNC? Well I don’t know. Some parts are probably going to have most of the forces going in the right direction, and some may not. So it’s probably ideal for some of the MPCNC, not ideal for some, and good enough for others.
- Prusa CF Polycarbonate - I’ve printed a few rolls of this stuff over time and it’s pretty amazing. PC has ridiculous strength in tension and the addition of the CF makes it relatively easy to print. It’s as good as PLA flat, much stronger on edge, and not too bad standing. It has the best temp resistance of all these as well. Seems like the perfect material except: it’s expensive, requires a full metal hot end that can print 285c-295c, a bed that can reach 110c, it doesn’t bridge as well (but may can be tuned to acceptable levels), and I’ve seen a slight Z shrinkage on some parts (but I need to do more testing to validate and quantify). Should you use it? Well if you aren’t a 3D printing newbie, have extra money, are fine if some of the bridging isn’t perfect and you have to post process the parts more, want something that can handle any temp any garage ever could get to on planet earth, and want a rigid machine too? Well then yes, it’s a great choice.
I hope everyone finds this helpful. Testing filaments can go on forever and I’ve used about all the spare time I could muster on this. I’ll point you to Youtubers CNC Kitchen, My Tech Fun, and Thomas Sanladerer for more info on various filaments. One interesting test is for creep (how much a filament deflects over time with a load applied). My Tech Fun for sure tests that in an interesting way. These Youtubers also do tests on strength vs infill, strength vs perimeters, strength vs temperature, and a lot more.
Merry Christmas 2022 and Happy New Year 2023!
Link to spreadsheet for Version2: Filament Shootout V2 - Google Sheets
11 Likes
WOW, that is a ton of data.
It looks like the CF filled stuff has poor interlayer adhesion for one reason or another. While it is possible to have amazing properties, standard PLA seems to be the all around champ.
2 Likes
Definitely lower interlayer adhesion for CF materials in general. Then you have the factor the CF doesn’t contribute anything layer to layer and it becomes super anisotropic.
I think PLA (pure PLA, not the modified ones) is still king, and by a fair margin. Price is a big factor too. For the same cost, you could print out several sets of MPCNC parts in PLA, compared to one set in any of CF filaments. I love the CF PC Blend filament but I think it’s overkill for this and doesn’t gain you as much as it costs. I may use it on some specific components, but I haven’t decided for sure. I just happen to have 2 more rolls so we’ll see.
2 Likes
Also, I didn’t optimize and used the same speeds for everything (added that info). The CF layer adhesion could be improved some by slowing down the print and messing with the cooling settings some. It’s a balancing act between bridging well and good adhesion so it takes some testing.
I still love that this post is you saying hello in this forum. Like, hi, I’m Paul, I am posting for the first time, here is a dump of really interesting data that I spent a lot of time and money on.
Legend. Thanks a lot.
5 Likes
LOL. Love that.
Everyone’s got their thing. I love testing stuff and data. I’ve been lurking a while, reading up and learning. Just thought I’d share my work as someone else might benefit. It’s enjoyable to set up an experiment and see where it goes.
6 Likes
Awesome Paul! You really ran through the matrix on this one. That pile of specimens is a testament to that, and you did a good job presenting/explaining your data. For a ‘non-engineer IT guy’, you do science very well.
FWIW, I have switched to carbonX petg for most of the important parts on my primo. I haven’t had much problem with layer adhesion using the settings recommended by 3dxtech. IIRC, 245C/68C, 0.4 nozzle, .45 line width, .25 layers, 40mm/sec. I run a big-ole 3-wire PT100, which may shift the ideal hotend temp from what other machines do. You’re totally right though, that the CF only interferes with adhesion… so may tradeoffs there’s no perfect filament.
I’m also curious if you plan to add carbonx pla-cf into the mix? That stuff is a PITA to print it loves to jam, but I’m guessing it will be an edge above for non-heated parts. I had to swap my all metal e3d for a teflon lined heatbreak to use a spool of that stuff up… which is why I haven’t bought it since (lots of work for me to print it, and the all metal works with all other filaments I use).
1 Like
Thank you! It was a lot of work but not too bad. Once I got the itch to test, it needed to be scratched
One thing these tests don’t do is give a real world comparison of the MPCNC parts, under actual milling loads, and in dynamic situations. I suspect the increased strength of the CarbonX PETG in the one direction is beneficial for most MPCNC parts. There are some pieces which I wonder about like the Core and the Corner leg locks. Just because it seems like the direction of the layers and my one-thumb-in-the-air-left eye-squinting analysis at the loads there would cause the PETG-ness of the CarbonX PETG to be a liability. But it would be stiffer in the other directions…so it’s a guess what the net result would be. I think for a lot of the parts it really is a great (i.e. superior) filament. The corner top/bottom for example and perhaps the trucks look like they would benefit the most…but again just eyeballing. I don’t think many are able to print entire sets of parts and build multiple machines to test defection so we’ll have to stand on what everyone is experiencing.
While layer adhesion does seem lower with the CarbonX PETG, the standing loaded samples exacerbated that since the CF wasn’t really helping at all. It was basically just PETG at that orientation…well, PETG with lower layer adhesion than regular PETG. I don’t think it’s something that would be evident in a full MPCNC part which is dimensionally large enough you can’t just bend and snap it (most aren’t anyway). So it goes back to the question of whether it matters and does the extra stiffness in the directions the CF help enough that it’s just a moot point.
I have almost bought a roll of CF PLA a few times. I’d like to test it out. Maybe after the new year. As you can tell I’m a bit of a filament hoarder.
The CF samples are very interesting. I would expect the fibers would be oriented parallel to the extrusion which is consistent with the results.
The flat vs. edge differences could possibly be attributed to perimeters vs. infill. The edge parts are all perimeters (or anyway a much higher fraction) while the flat parts have infill at a 45 degree angle.
I would guess that a flat part with a very high number of perimeters could have the same stiffness as the edge part. If so it would confirm that the fill orientation is the reason the flat part is less stiff.
Slots in a part can force the orientation of the perimeters, which could make a significant difference for CF parts (besides the ordinary benefit of extra density in selected areas). That could make for interesting designs specific to CF.
2 Likes
100% agree Jamie. If I had oriented the infill for the flat pieces along the length of the sample, the edge and flat would be much more similar.
1 Like