Incredible Flow Testing Resolution Reveals SO MUCH

[@Vector3DP]

Please note that there is a mistake on the graph shown from around the 5:40 mark. The X axis has been labelled as "Flow rate (mm³/s)" but it should be "Time (seconds)"

[@RobGadeke]

Potentially the 0.8 outperforms the 1.0 because of cold core issues? Could be interesting to see if CHT changes that relationship

[@smittery]

This was my first thought as well

[@TTPF_]

I think the same, because I could get linear flow increase when I upped my Volcano nozzle from 0,6 mm CHT to 1,0 mm CHT.

[@Vez3D]

Good analysis man! Very good video.. I think, however flow test should also be tested while printing and seeing quality of plastic layed down.

[@Vector3DP]

I don't disagree that there is value in testing real printing but you need an objective test to determine performance and i've not found a good objective measurement for that. Measuring strength, brittleness etc could be an option, but might not be very accurate or reliable. Also, considering we're had the option of looking at prints and printing faster since day one, I can't help but feel that it's the subjectivity of that method that has landed us in the bigger number better mentality we have at the moment.

[@Vez3D]

@@Vector3DP agreed. But I think Miragec has found a good way to do that. Im sure you saw his video too ?

[@Vector3DP]

Yep, watched every second of that. I think I'll be hated for being this critical, but unfortunately, I think it was quite misleading. The extrusion force data actually had no effect on the flow results (I know this because I asked Mirage C) so every result would have been identical if there were no force measurements at all. The video gave the impression that force was being used to determine something, but it wasn't, it was just plotted to make a pretty graph (which was very pretty, hats off for that).

[@user-yk1cw8im4h]

@@Vector3DP Mirage C is way better than you.

[@piloterror1644]

Measuring mass is a method of initial and periodic calibrations on polyjet printers I have worked with. Maybe you could print a large material volume (to minimize start and stop deviations) vase mode part to weight at different flow rates, and you would then have a print to also dissect a bit. Every large industrial filament printer I have worked with has had an encoded geared dc extruder drive, plenty of motor data, and filament encoders.

[@klschofield71]

Really enjoying all these in-depth analyses of extruders/hot-ends. Seems like with all these data -sets you could put together a quick reference chart for optimum flow rates (great datum by the way) for various hot-end/extruder combinations; similar to your proposed models for identifying ideal design parameters (got a good chuckle, cause my desk is exactly the same). There's already tons of examples out there for various CNC applications comparing speeds, feeds, cutter material, and stock composition. I'ld buy that poster.

[@ThantiK]

I wonder if an extremely steep throat angle would be of any additional benefit. Whosawhatsis suggested to Sanjay a long time ago to move from the 120 degree drill tip, to a 60 degree drill tip for the E3D ends, and it resulted in quite a performance increase. I'd love to see a nozzle that used an even smaller throat angle - and I wonder if that might be able to help solve the cold core issues as the cross-section gets smaller over a much longer distance; both increasing the distance needed to fully plug the nozzle, and also reducing the diameter of the molten area and helping push that heat into the center. Unfortunately, not something that I have the tooling for. :(

[@leesmithsworkshop]

I really appreciate all the time and work that went into this video.

[@mattisonsiri228]

That was a fantastic video! Thanks for adding some rigor to the field! I would be interested to see the average heating power vs flow rate. The 100 W spec for the heater is the maximum power when on 100% of the time (btw I've found that the resistance of heating cartridges can vary quite a bit, so 100 W may not even be the maximum power). Did you notice any temperature sag at high flow rates? If so, then a higher power heater will help as you've likely hit the maximum power for that resistance and output voltage of your power supply. If there is no temperature sag at high flow rates with a 50 W heater, then switching to a 100 W heater will likely not have much of an effect as the 50 W heater was not being used at 50 W to begin with. This can be used to determine if the bottleneck is the thermal conductivity between the hotend-nozzle system and filament, the power of the heater, or the extrusion force of the extruder-filament combination. Keep up the great work!

[@pascal789456123]

Did you PID tune for the 100w heater ? Could explain the wacky spikes in extrusion force if the temperature varies a lot

[@Vector3DP]

Yep, pid was tuned for both heater setups.

[@radioactiveolix540]

I have a Mosquito Magnum hotend, best hotend I've ever seen!

[@tvathome562]

Wow this has got my head in a spin... I'm not sure extruder force equates to a 'good' print. (the force of the extruder could be applied slowly but up to higher pressure than a quick burst at a lower pressure (non-Newtonian fluids vs Newtonian). Actually flow vs expected makes sense. As does 'print artifacts' caused by insufficient melting of the filament cross-section when printing fast but I have no idea how to measure that. I'm not sure how common it is but I've seen speed benchys printed with Form Futura Volcano high-temp Pla so that definitely fits with your findings.. Making me question everything I though I knew....

[@Marcos-tj8nk]

I think that in a standard bed of 20x20 not so much flow is needed. I have a bondtech CHT nozzle and use it on special prints, because, even If I wanted to print at a constant high flow, I have to add some time between layers to avoid melting problems (specially for tech materials printing without fan). So I can only print at a constant high flow on tech materials when the pieze has a very simple geometry without detailled zones. So for me the bondtech CHT nozzle made the volcano and mosquito obsolete at least for small beds

[@DamienRobertsonYYC]

It could be that there is an optimum maximum nozzle size. It might be worth testing with a 3mm filament, an see how that changes everything.

[@Sovol]

Great video! Keep on the great job man👍

[@nerobro]

I wonder if this relates at all to "time in the melt zone". As you're moving.. LOTS of plastic, you might not be melting a lot of the plastic that's extruding. Also, would "printing against something" make a difference? would it self limit the unmelted, or less melted plug down the middle?

[@Error90]

Great VIdeo! First time viewer. Keep up the great work!

[@Dave-wv9vc]

Great stuff, I'd love to see you try different large diameter nozzles to see if the 0.8 > 1.0 diameter holds for nozzle types other than Bridge-Master. I suspect the internal geometry of the Bridge-Master is responsible for the confusing result.

[@shaunmorrissey7313]

But what's the print QUALITY like?

[@jakeengland1430]

Literally the same as the cheapest working hotend a VZ bot with a magnum + WILL print the same as and ender 3 hotend if both are tuned correctly (assuming the same extruder) the difference is in feature set and theoretical/practical max speed.

[@brandoneich2412]

Can this utilize CHT nozzles? Wonder what kind of a change that would offer.

[@Vector3DP]

I believe so yes, it should offer a substantial improvement as it overcomes the normal cold-core problems. Not tested them yet though, so I can't say for sure.

[@JonS]

How are you getting the standard deviation? Maybe I missed something, but it appears you are calculating the sigma from just three samples. If that's the case, then you will have an incredibly-low statistical significance, meaning those values will be mostly noise. Yes, you can fit a Gaussian distribution to just 3 points, but it's not meaningful. Again, apologies if I'm misunderstand your method.

[@Vector3DP]

It's calculated from a whole run, so about 40 seconds worth of samples at 20hz, so about 800 values.

[@Nitram_3d]

are you running into limitations due to lack of tourq from the stepper motor

[@Vector3DP]

Nope, not at all. The motor I have is huge, maybe like 56mm body or something.

[@Nitram_3d]

@@Vector3DP the tourq drops at a specific rpm... Have a look at it so that you are certain..

[@JernD]

Very interesting video! I have subbed for more videos/tests like this! In the last 2 days I have built a heat transfer model for 3D printer hotends. One key parameter that has become increasingly apparent from my model is that SMALLER internal diameters lead to increased heat transfer rates to the filament (this is because of the increased surface-area to volume ratio). This may partly explain your odd results with increased force for the largest nozzles you tested. The temperature gradient must flow across a larger radius of plastic with a poor thermal conductivity at the end of the nozzle. My model does not examine extrusion force, which is obviously a key factor in real world performance of hotends. Let me know if you are interested in seeing my heat transfer model. Also, as previously noted, I believe there is a mistake on the graph at 6:20 in that the x-axis should be labeled Time/Seconds instead of flowrate.

[@Vector3DP]

Hi John. I would be interested in seeing that model. To my understanding there has not been enough experimental data yet to determine accurate performance from a 3D model but if you're in the right place with that information it would definitely be interesting. I 'm not sure the nozzle size would affect heat transfer to the material too much because it's only this dimension right at the tip. Yes, the 6:20 error has been shared. That should indeed be Time (seconds) on the x axis.

[@3D_Printing]

Learned I have too much money?

[@rarefind3d]

That first plot is very confusing with the x axis labeled as flowrate. You call it time in the video.

[@Vector3DP]

timestamp?

[@JernD]

@@Vector3DP at 6:20 you have the individual lines labeled by mm^3/s but the x-axis is also labeled as mm^3/s. This is incorrect as pointed out by RareFind, as the x-axis should have been Time / Seconds.

[@Vector3DP]

I'm afraid it's not incorrect. Perhaps you have misunderstood what is being shown here. That's my fault if i've not explained it well enough. Perhaps try watching it again and it might click.

[@JernD]

@@Vector3DP GAH sorry I made a mistake, the timestamp was supposed to read 6:20 -- am I still misunderstanding or is there a mistake? I edited my comment above so others aren't confused.

[@Vector3DP]

Oh balls. Yes you're right. That is a mistake 🙁 The X axis there should infact read "Time (seconds)".

[@jleadbetter29]

4 to 5 seconds to normalize...not 45 seconds as I initially heard it.

[@Vector3DP]

Nope, you heard correct. results would change unless the test was around 45 seconds long.

[@bagibadoo439]

I like your content but i believe you would benefit from shorting down your clips. Going from 30 min to say 10 min would probably be better.

[@ThantiK]

This type of content isn't really distilled down for "the masses". This is for people who are deeeeeeep into this hobby, engineers, etc. 10 minute segments isn't enough to get all of this information across.

[@bagibadoo439]

@@ThantiK yeah that's me but 30 min for what can be destillerat to essentially one flow rate plot is too much. Cnckitchen does it pretty well, he puts detailed info about test setup and raw data in his Patreon and focus on the results in the video.

[@sacredserpent9667]

The nerve...