Tensile Strength Testing

[rsterne]

I have just noticed that I used the incorrect cross section for some of the vertical coupons below where testing was done with a 0.6mm nozzle.... The corrected layer-to-layer strength (Z) is less than I originally published  ....  I have replaced the last 4 charts and corrected the comments for the following filaments:

3DXTech CF-PETG
Polymaker PA612-CF
Anycubic PLA
3D Printing Canada PLA

The biggest change is that the Z strength does NOT approach the X strength, I should have known there was a mistake when I saw that....

I have mostly worked my way through the materials I have on hand, so I installed a 0.6mm nozzle to do more testing.... I have found that using greater extrusion widths increased the layer-to-layer bonding, as per the CNC Kitchen video testing that Stefan did.... He found the maximum strength occurred at about 150% of nozzle diameter is his tests, and I have confirmed that.... So, that has become a goal for parts that require maximum strength.... I tested quite a few materials, looking at extrusion width in some, layer height in others, and in some, both.... Here are all the charts, with a few comments after each one....



This one shows an increase in strength with a 0.4mm nozzle going from a 0.50mm width to 0.60mm (150%), and the same results with a 0.6mm nozzle peaking at a 150% width (0.90mm) and then starting to decline....



This one shows the peak at 1mm width (166%) in both directions, and a very sharp decline in strength in the horizontal (X) direction above that....



This one, using the same filament and a 1mm width (the peak above) showed a decrease in the vertical (Z) direction as the layer height increased, with a slight increase in horizontal strength from a layer height of 0.25mm to 0.30mm, and then a steeper decline above that....



This one showed an increase in Z with the 0.4mm nozzle at 150%, but a decrease in X.... However with the 0.6mm nozzle, both peaked at 150% (0.90mm)....



With the same material, using a 0.6mm nozzle and two different extrusion widths, 0.9mm and 1.0mm, the horizontal (X) strength peaked with a 0.30mm layer height, and the vertical strength with the 0.90mm width also showed a slight peak at 0.30mm layers, but using a 1mm width the layer-to-layer bonding decreased as the layer height increased.... The next three tests focused on varying the layer height while using 150-166% of nozzle diameter extrusion width (at this point, a "proven" maximum)....



This clearly shows a decrease in strength with increasing layer height.... Note that for all three of these charts, the extrusion width was 0.90mm (150%) for the 0.25 and 0.30mm layer heights, but I used 1mm for the 0.33 height (3:1 bead)....



This shows a fairly dramatic decrease in X with increased layer height, the layer-to-layer strength was relatively constant....



The Nylon acted a bit differently, the strength in both directions increased with greater layer height.... The broken "Z" coupons showed no layer lines with the thicker samples, and a somewhat "crystalline" appearance.... Next I tested two PLAs....



At a 0.20mm layer height, the strength was a lot lower.... Greater strength in Z but less in X with the 0.6mm nozzle....



A different PLA, showing a similar graph....In both PLA tests, however, the horizontal (X) strengths were lower than what I got for the same filament when testing with a 0.4mm nozzle and 0.20mm layer height (the typical printing profile for PLA)....

Now for my conclusions.... First of all, print quality depends mostly on layer height, and is basically unaffected by the extrusion width, up to 150-160%, as Stefan also found.... Requirements on the appearance of layer lines will vary with personal preference and the job at hand, but the "standard quality" using 0.20mm for PLA is good enough for me, although for shallower angles (eg. on the top of a sphere), thinner lines definitely help, so the "adaptive layers" function in Cura is helpful for that.... However, I am convinced that for carbon fibre filled filament, you should not use a layer height less than 0.25mm, which because of the matt finish doesn't show the layer lines too much anyways....

In terms of the extrusion width, I have not seen any significant deterioration in print quality by using widths of 150-166% of the nozzle diameter, and it certainly speeds up the printing process and increases the strength in the vertical (Z) direction, by up to 50% or more.... This seems to be a no-brainer, both stronger and faster.... This means an extrusion width of 0.60mm for a 0.4mm nozzle, and 0.90mm for a 0.6mm nozzle.... With some filaments there is, however, a small loss in the horizontal (X) strength, and it may really drop off above 166%....

The effect of layer height on strength is something that Stefan looked at, and his conclusions were that you wanted to stay below 50% of the nozzle diameter, that the strength decreased above that.... About half of my tests with the 0.6mm nozzle confirmed that (ie maximum height of 0.30mm), the rest didn't.... When using CF filaments, you just shouldn't use anything less than 0.25mm, IMO.... In many instances, going to a 0.6mm nozzle lost a bit of horizonal (X) strength compared to a 0.4mm nozzle, but that may be a result of the small cross section of my coupons, which are only 3x5mm.... I have a gut feeling that in the horizontal (X) direction, lots of small lines, rather than a few larger ones, give greater strength, up to a point.... This may be the reason for the decrease in strength using a larger nozzle.... In the "Z" direction, however, where layer-to-layer bonding is critical, the higher pressure caused by a 150% (of nozzle diameter) extrusion width is also increased by a small layer height, but once you go above a certain height, that pressure drops and you lose strength....

So, remarkable increases in layer bonding strength can result from using 150% extrusion width, and if you go to a 0.6mm nozzle of course the printing time drops by about a third.... If you combine the two, you can print in less than half the time than with a 0.4mm nozzle.... Here are some suggestions  you might like to try....

Non CF filled filament
0.4mm nozzle....0.20mm layer height with 0.60mm extrusion width....
0.6mm nozzle.... 0.25-0.30mm layer height with 0.90mm extrusion width....

Carbon fibre filled filament
0.4mm nozzle.... 0.25mm layer height with 0.60mm extrusion width....
0.6mm nozzle.... 0.25-0.30mm layer height with 0.90mm extrusion width.... for faster printing, 0.33mm layer height with 1mm extrusion width....

Bob

[TorqueMaster]

Bob,

I appreciate all the testing you've done -- great data to show general trends, and specific data on so many brands.  Thank you!

I  have been mulling the data, and I've seen a "disconnect" a few times -- always with CF filaments.  If you use say, a 0.6 nozzle, with 0.60 line width and 0.30 layer height, and compare that to the 0.4 nozzle with 0.60 line width and 0.30 layer height, there is a VERY significant increase in Z strength (layer bonding) for the smaller nozzle, even though the "bead size" laid down is identical.  Did speed of print mm/s change?  Did flow rate mm3/s change?  That could be significant.  Speed and flow aside, it may be that forcing the molten CF infused filament through the smaller nozzle causes a more turbulent flow, and the CF fibers are "jumbled up" rather than "laid straight"  -- causing a rougher surface and more vertical fibers to aid in a stronger bond to the next layer?

I looked and did not see a lot of evidence to prove that non-CF filaments have the same remarkable increase in strength when the line width is 150% of nozzle width, versus using a larger nozzle at 100% line width.   Can you test a few regular PETGs and see if they also benefit?  Cura's layer view preview will also show the speed and flowrate -- obviously those numbers would be most useful in the area of the specimen intended to be broken.  Ideally the speed and commanded flow should be the same for both nozzle sizes.

Here's the CF graphs that exhibit what I'm talking about.  On one, I added a data point from one of your other graphs, rather than add it here too.

[rsterne]

When you print with a larger nozzle and lay down the same width line (at the same layer height), the pressure in the larger nozzle is less.... I think basically the higher the pressure in the nozzle, the greater the Z strength (eg. your first chart)....

The PLA I tested with the 0.6mm nozzle, the X strength was less than with the 0.4mm nozzle, but the testing I did with the 0.6mm nozzle had a 0.9mm (150%) width for the 0.20-0.30mm layers, and a 1mm (166%) width for the 0.33mm layers.... The earlier testing I did with PLA was a 0.2mm layer and 0.4mm (100%) width, and the Z strength went from about 50% to 100% (with X reduced), or from about 5K to 8K, a big increase with the wider lines....

Where do you find the flow rate in Cura, I have missed that?....

Bob

[TorqueMaster]

Bob,
I found your two PLA graphs, one 0.4 nozzle, the other 0.6.  As you can see, when comparing the Z strength of a 0.60 line by 0.30 layer height, the 0.6 nozzle is the (slightly) stronger layer bond.  In this case, the higher pressure is not making it significantly stronger.  The CF filaments however, do seem to be benefiting from the smaller nozzle.

Cura Flow is in the Layer Preview screen Color Scheme pulldown.

[rsterne]

Thanks for that, Bob, much appreciated.... Yes, the flow rate in the test section of the vertical (Z) coupons is very slow.... Perhaps that is why on some coupons I was getting failures further out towards the ends instead of in the middle, where they should break.... The variations in strength on those charts are so small, they may well be within statistical (testing) error.... I would not put any faith in comparisons of data within 5% or less (maybe even 10%).... For example, on that 3D Printing Canada PLA, the wavy line for the vertical strength could just as likely be a straight line, slightly higher on the right, instead of up-down-up-down....

Bob

[TorqueMaster]

On the flow, no prob, Bob!

Also be sure to look at the same Color Scheme pulldown for Line Width.  If you are using Cura 5.x, it may use varying linewidths around your base chosen linewidth, sometimes up to almost 2x.  If you are setting your base line width to 1.5 nozzle width, Cura may be increasing it to 3x nozzle width!

On the PLA curves, I think you are missing my point.  Yes, the PLA 0.6 nozzle using 0.60 line width and with 0.30 layer height has almost identical Z strength to the 0.4 nozzle using 0.60 line width and 0.30 layer height.  There is no Z-strength advantage to using the smaller nozzle.

On the PETG-CF, on the other hand, there is a sharp increase in Z-strength in the same nozzle/line width /layer height test comparison.  The smaller nozzle is showing a huge increase in Z-strength compared to the larger nozzle.

The question is -- is that because it's CF, or because its PETG?  So that was why I asked if you could run the same tests with some regular, non-CF, PETG.   Hope this makes sense. 



 

[rsterne]

Got it!.... I will try some PETG with different Extrusion Widths, etc. when I get some time.... Right now I am tired of testing and have a project I'm working on!....

Bob

[TorqueMaster]

Sounds good Bob, thanks.

[rsterne]

I had a look at the other Dropdowns in Cura, it is very interesting to see what actually happens!.... I found no change in layer height (unless using Adaptive), nor anything really in line width, except the center "dot" in the test coupons.... However, the test sections of the coupons are printed MUCH SLOWER and with MUCH LESS FLOW because they are so small in area!.... Of course I can't print them large enough to avoid that, or they would be wayyyyyyyyyy too strong for my machine to break them!.... What difference that makes in the actual strength on a part printed "normally" (ie as intended) I have no idea, and it may be a bit of a stretch to think that all filaments would give proportional results at my test conditions as when printed at intended speeds and flow rates.... 

Bob

[TorqueMaster]

Hi Bob,
I had been wondering if you had a chance to look at all the useful info in the pulldowns. 

For linewidth, explore a hollow model with a gradually increasing thickness outer shell for instance -- even a solid cone may work with 100% fill and concentric fill pattern on top/bottom  --  the linewidth will probably vary through most of the range the default parameters are set for.  For a 0.50 default linewidth, most of it is printed at 0.50, but part will have lines as narrow as 0.425 (sometimes even lower) and as high as 0.85 - 0.89 -- to fill the gaps better.  The print speed typically varies inversely with the linewidth such that the flow remains constant.

I just checked something on solid prints -- the variable linewidth mostly affects concentric fills.  If you use lines or zig-zag fills, it may not show up at all.  I find that concentric often leaves fewer unfilled gaps than the linear fill patterns.

I believe what you are seeing with the less speed and less flow in your vertical specimen necks -- is you running up against the "minimum layer time" and/or the "minimum print speed" in the Cooling section.  Are you looking with just one on the plate, or several?  If you printed multiple specimens "all at once" I believe the necks should print the same speed and flow as the rest of the specimen.   If not, you may need to add more specimens (or timewaster columns) or set min layer time a bit lower. 

There are also special settings for printing "small features" --  which I have not messed with and I believe are disabled by default  -- could also affect your print speed and flow in the necks.

That's the $1000 question -- how the slower printing (if it really was slower) affected neck strength -- and does it affect all filaments the same?  I would venture to say the existing test results are still quite close, but the error bars around them may expand a bit.   Maybe pick ONE of each type of filament (PLA, PETG-CF, Nylon, etc), tweak the setup so a specimen prints at the intended speed (with ample time between layers), and compare the new results to the originals.  If the results are still about the same, great, note which direction strength changed and call it done.  If they change dramatically... well ... just hope that they don't. 

[rsterne]

I was only printing one specimen at a time (well 1 X and 1 Z, but the Z is only 3mm tall, so if doesn't affect the test section of the Z coupon), as sets are done at different temperatures (for the most part).... My testing method seems VERY consistent, and even with only 1 sample per variable, the curves are pretty fair.... Yes, the minimum layer time is the likely culprit, I think.... The "official" tensile test specimens (can't remember the number) doesn't have a huge test section.... It isn't an issue printing in X, only in Z....

Bob

[rsterne]

TorqueMaster.... I found an error in the last 4 charts I published, I used the wrong cross section for the Z coupon.... The correct strengths are significantly lower than what I previously published.... In particular, the PLA layer-to-layer (Z) strength was nowhere near the X strength, sorry for that mistake.... 

I did the tests you requested on plain Eryone PETG, here are the results with both nozzle diameters.... I used extrusion widths from 100-150%....





I hope you can make some sense of these results.... I think the reduction in horizontal strength with the 0.6mm nozzle is due to the much smaller number of strands in the test section.... We seem to be getting than in a lot of samples....  .... In some cases, the X test section printed with the 0.6mm nozzle is delaminating before it fails....

Bob

[TorqueMaster]

Bob,
Thanks for the corrections.  I'm still trying to re-wrap my head around all this. I think I see it like this:

The PLA graphs now show the significant increase in Z strength when a 0.4 nozzle is used at 0.60 linewidth vs a 0.6 nozzle used at 0.60 linewidth.  (both at 0.20 layer) This agrees with the PETG-CF results.

Thanks for doing the PETG (non-CF) tests!  They also agree -- the 0.4 nozzle doing the 0.60 line is Z stronger than the 0.6 nozzle doing a 0.60 line (both at 0.20 layer)  About 5250 vs 4200.   

It does appear it's "always" stronger in the Z when the nozzle is printing lines wider than the nozzle size, and best in that 125-150% zone, and shooting for 3:1 linewidth:layer height looks smart, too.  The caveat would be the graphs where you found layers under 0.20 or 0.25 had lost some Z strength -- a minimum layer thickness is generally required as well.

These are some *REALLY* useful fundamental guidelines for strong parts.  If one prints around 3:1 as above at a thickness of 0.2 of 0.25, one may need to slow down print speed to not exceed max flow of the printhead.  Creality has probably improved this from the Ender 3, but I have discovered my printer's limits -- when printing wide and "thick" for best Z (and overall) strength, I am printing at a relatively slow printhead speed now.  Layer height ignored...print quality improves (less artifacting), and it's still creating prints as fast as it is capable -- win win.

I'm attaching a graph of what happens when you ask a hotend for too much flow.  Ender 3's are supposedly capable of ~10mm3/s.  What they don't often mention is that is for PLA.   When printing in PETG, the max flow is reduced -- best I can tell, it's about the higher temperature needed to melt it (and more heat lost to the room) as well as higher viscosity.  My printer has a lot of hours on it, I do not know if this is typical or not.   The blue line was some typical test prints, the green some test cubes, all weighed vs what their expected weight should be.  They overlap pretty well.

So a possibility on the de-lamination of your X specimens, could be underextrusion if the print speed and resulting flow on them is getting off the ideal line. 

Fun stuff, these printers!

[rsterne]

I haven't explored the maximum flow on my Ender 5-S1, that is something I need to do.... When printing with CF filled material, virtually all the manufacturers say to use a slower speed.... I am now using a default of 80mm/s, which of course only does that on the infill.... The walls, top & bottom are printed at 40mm/s, so it's fortunate that using a 0.6mm nozzle with a 0.3mm layer height and a 0.9mm extrusion width gives me the best strength, along with a quality I can live with.... The CF texture hides the layer lines "good enough" for airgun parts, IMO....

Would I be correct in assuming that to calculate the flow you take a rectangle of (height x width) time the linear speed?.... For example, for my 0.3mm layers with a 0.9mm width at 40mm/s that gives (0.3 x 0.9 x 40) = 10.8 mm^3/sec.... I wonder if that is within the capabilities of the Spider hot end on the 5-S1 ?.... I know my print times are slightly less than that predicted by Cura....

Bob

[TorqueMaster]

....
Would I be correct in assuming that to calculate the flow you take a rectangle of (height x width) time the linear speed?.... For example, for my 0.3mm layers with a 0.9mm width at 40mm/s that gives (0.3 x 0.9 x 40) = 10.8 mm^3/sec.... I wonder if that is within the capabilities of the Spider hot end on the 5-S1 ?.... I know my print times are slightly less than that predicted by Cura....

Bob

I would use the preview pulldown to see the flow or flow range in a given print.  But yes, the method you state should be very close.  I believe Cura models its volume estimates on a slightly bulged/rounded rectangle cross section. 

Sometimes my Cura print time estimates are right on, other times they are a little bit high, as much as ~10%.  It seems like all the info they need should be there, not sure why it isn't spot on.  All my best guesses -- hardware limits, ignoring acceleration times -- would have the print taking LONGER than estimated, not shorter.

[Baco]

https://www.cnckitchen.com/blog/extrusion-system-benchmark-tool-for-fast-prints

i have used this to test the stock hotend on the sovol

i can attest that when using 0.2 LH / 0.8 LW / 0.6 nozzle the hot end clogs sometimes from too much back pressure

but you can always increase nozzle temp and throw another variable in the mix xD