@@Tarbard When it comes to preventing phishing I can't see any difference either, and the browser always knows what domain it is on before releasing the password to the site.
@@Tarbard These sponsorship segments are based on a contract where the video creator must go by the provided script, and can never say anything bad. That's why when a creator does one of those Raycon segments they have to say the audio quality is good even when it leaves a lot to be desired.
@@soundspark I have no problem with adverts being positive and scripted/approved, I just wondered if I was missing something regarding nordpass' features.
You've created the most delicious looking benchmarking tool with those coextrusion blobs! I love pushing speed when printing models, but I want to print quality models, not just print quickly. Thanks, as always, for such clear information on filament flow rates and your new profiling tool!
Great video! One other thing that I've done is use Prusaslicer's "Max Volumetric Flow Rate" to limit print speed. Then I can just set infill speeds to something like 120mm/s and let the volumetric flow rate automatically keep the speed at the max.
Great video. Thanks for the mention. If you increase the extrusion width on my Extra Fast Profile you’ll also see lower numbers as well. I leave adjusting the temp to the user as each filament is different. I also still include many settings that give small improvements in quality. So there is some compromise. So many people have asked for a Prusa version of my profile. I’ll just send them here. You cover it well.
In case anyone else is using Klipper, I had to edit a couple of things to get it to run. In M109 R## the "R" needs to be replaced with "S", then I needed to disable some machine software limits under [extruder], adding max_extrude_cross_section: 5 and max_extrude_only_distance: 500.0. (For safety I'm going to comment them out when not running the test). Hopefully this helps someone!
Hey it helped me! Figuring out the right value for max_extrude_only_distance is easy enough, but max_extrude_cross_section is not as straightforward, so thanks for providing something to work with.
@@dsp4392 no problem! I hate when people don't post their solutions to issues like this, it just ends up with everyone going on their own wild goose chase 😅
Thank you so much, Couldn't get this working until I saw this post. Surprised it hasn't been added to the web page for the gcode generator so that people see it.
THIS is the content that makes 3D printing so amazing to me, even though I've yet to buy my first printer. Seeing this makes me want to get a cheap Ender just to tune it patiently. Tools like this make it possible.
As a tinkering kind of guy myself. Out of the Pro and V2, the V2 is the better model for tinkering. You'll have to get comfortable with modifying C header files either way, specifically the C pre-processor. A lot of the potential of these printers just doesn't come into play until you start to tweak the configs which are written in C header files almost exclusively using the pre-processor. It's kind of like an ini file on steroids.
Very valuable info! I especially liked how you referred back to the real flow measurement results to determine the actual print temperature. I usually take an educated guess and start praying 🤣
I think the greatest pearl of wisdom here is determine the flow rate of the part, then adjust temp. I generally default temp based on material, but give a lot of thought to adjusting speed based on the needs of the print job. Definitely plan to switch my order of operations here.
hey!! I tried to make a sub-10min benchy on my ender3 a few month ago (settled on 10'33" for the moment) one of the parameter that can be useful is the skin line orientation, making it more aligned with the X axis allows the light print head to do most of the rapid accelerations ;)
stock bed, stock motion system (but with 95% rated current), I modified the print head a bit with a Sherpa extruder and bi-metal heatbreak + I modified the v6 nozzle with 0.4mm brass wire like in your videos (I kept the stock hotend and fans, the quality of the benchy is quite poor to say the least) then I mostly played with all the cura settings to optimize what I could, didn't post anything because there are 6min ender3 benchys out here using klipper (way too complicated for my mecanical engineer mind) but maybe one day.. 😶 I'll follow this comment with my settings when I have time to do so
So glad you are part of this wonderful community everyone knows Stefan is the man for the technical info and I thank you so much for your work good sir !!!
Wow, simply awesome. This is the most impactful 3d printing video I've watched this year. I've got an Ender 5 Plus with a direct drive Bondtech LGX through a mosquito hot end. I had no idea it could push so much volume. I'm still tuning but so far I'm blown away with the rates it's achieving. Thank you!
I just started with 3D printing, getting a Prusa i3 MK3S+ about two months ago. I think my journey to great prints is immensely sped up just by watching and learning on one channel, this one. So much information and still digestible for a newbie. Been watching this channel for about a year to learn stuff before deciding which printer to get. Stefan never disappoints, every video is crammed with great info and tests. The presentation makes it easy to understand and enjoy 👍. Keep it up, I impress my friends with your knowledge 😁
I'm glad you made this; it validates a comment I left on a video last year about how we've been tuning our printers wrong for so long. Previously the common wisdom was to just change print speed, but this ignored that the printer has to first accelerate to that speed and often never gets there. My ide (based on a teaching tech video) was that we should instead be focussed on finding out our printer's max extrusion rate for a given filament and temp, work out a given nozzle's maximum print speed based on that and then instead adjust acceleration and jerk to affect print speed.
Would love a video from you reviewing best low-budget, mid-budget and high-budget printers. I'm thinking of getting one and a lot of the videos on this topic are either outdated or are from sources I don't trust. I feel like you're a well-respected expert in this area, so it would be great to see a 2022 review video from you.
Another thing where you can save a reasonable ammount of time are travel moves. They don't need to move smoothly as there's no filament laid down that can show wringing/ghosting and underextrusion is alao not a problem for obvious reasons. I push my Ender 5 Pro to 500mm/s at 2000mm/s² for most prints with larger elements and/or fewer travel moves per layer. If the print doesn't take longer than an hour I can push the acceleration up to 3500mm/s² without getting layer shifts, but I only did this to see the limits. Most of the time I use 2000, and on parts taking longer than 12h I put a fan at the Y motor (that's the weak point for this printer) or go down to 1500. A nice side effect I observed was reduced and finer stringing with PETG (my prefered filament type) since there's less time it can dropp during travel.
This is golden! I used the previous method you developed to identify the capabilities of my heavily modified Ender 3 with very good results. The Aquiles heel of my set-up is still the extruder, but based on what I found out with your experiment, that soon will be also replaced. Awesome work!
It would be fantastic if Super Slicer could integrate some more of these calibration tests into the program, like they already do with other tests. Then it can automatically use the printer and filament configurations we already have.
There is so much info in this video. New to 3D printing (but have watched some videos here and there) and it gives me a lot of ideas of how to test. Thanks dude
I was just about to use your "How fast can your hotend print?" video since I finally have the scale and a Revo Six in my Prusa MK3S+ with MMU2S. Thank you for having this automated process as I too can be rather lazy
Stefan, thanks yet again for your amazing work and video! Keep it up! While I haven't download your .gcode and tool yet, I did download your profile and compared it to my default profile for a 0.6mm nozzle @ 0.32 mm layer height on my Ender 3 Pro. After copying over pretty much all you setting changes I printed the cylinder that you had on the build plate twice with the same 0.32 mm layer height. Including heat up time for both the nozzle and bed from room temperature, my default profile took 28:49 minutes. Modified with you tweaks, it finished in 17:46 minutes (again, including heat up time for both nozzle and bed). That saved a whopping 11:03 minutes (38.35%)!!! The increase in temps (215 vs 230) sufficiently allowed for increase in speeds (both max and average) as well as volumetric flow rates (again, both max and average). I will definitely need to download your tool and do of my own experiments to gather actual data, but for now I am super happy! Thank you so much!
Great video Stephan, I'm usually deciding on extrusion parameters and setting my feed rate to achieve a target volumetric flow rate. So glad you pointed out matte vs glossy surfaces as well as an indicator of how well material was melted.
I always love seeing new benchmarking/calibration methods, tuning my 3D printer has become a bit of an obsession for me over the last two years xD I tend to be pretty conservative with my printer settings, but just knowing how much of a margin of error I've got is a very reassuring feeling. I'll test this out on my cheap all-metal heatbreak I just got; I've checked the temperatures with an external probe just to make sure heat creep isn't an issue (which thankfully it isn't, so that's a good sign) but a qualitative comparison with my stock hotend will be interesting to see. Also, on the matter of creating a fast profile: I highly recommend getting an ADXL345 accelerometer if you have a raspberry pi or arduino to read the values from it; using the klipper guide on resonance compensation you can figure out how much you can tune your acceleration and jerk settings, and particularly on my Ender3 the stock settings are... conservative, to put it nicely. I can run those values higher with a DD extruder than stock does with bowden, just as a forinstance.
Great, thank you! I've been struggling with the manual method for too long. I have gcode macros in Octoprint to do something similar but never thought of just dispensing a blob. The takes a lot (but not all) of the tedium out of the process. I've also wondered about what to use for the 100% reference. You use the smallest flow rate and I've tried that. Lately though I've been cutting a length of filament and using it's weight for reference. That makes my curve fits look better at least.
To help with stutter without lowering the resolution, enable Meatpack in Marlin and install the Octoprint plugin. Meatpack is a compression algorithm to push gcode faster over the slow serial connection.
This is a timely video because I just finished a hypercube and was wondering how to begin making it print faster using PrusaSlicer. Hope we get to see your Voron doing speed printing soon!
This seems like a compensation you could easily add to Marlin. 2D table of temp vs flow. Little bit of logic to anticipate and account for heating and cooling delays.
Love it! being able to change the nozzle diameter and it change the height quickly would be a great addition, i am doing some testing with a 1mm nozzle and am having to do a lot of manual manipulation to get the blobs to come out correctly.
Wonderful video! I resonant with this video a lot because I am also tuning my voron 0 towards faster printing. This gave me a lot of idea to optimize my profile even further. One thing to add though, is that Prusaslicer will limit print time of each layer under Filament - cooling, and that might slow down the print unexpectedly.
One trick i found when printing with larger line widths is to have a nozle thats not too ponty and a bit of flat. I also piled the edges of the point of to he nozle to have a bit of roundness so that it glides instead of dragging when squishing that filament. And you will get a smoother surface... And also ironing.
You are a legend... I am fairly new but I've just finished upgrading my Ender 5 Plus with Creality's silent board (TMCs 2208) and a Direct Hemera + Volcano as, exactly as you explain, I intend to "move a lot of filament" to print fast despiste not necessarily moving the tool that fast... Until now I had performed only one flow test at 1 temp and 4 speeds with no repetitions... Volcano did deliver 30 mm^3/s I'll make sure to check out your tool. Dank sehr!!!
could you make a toung and groove to improve layer adhession? assuming 100% infill print .4mm lines at .2 layer height and .4 spacing between the lines move up 1 layer and fill in the gaps by using enough filament to print .4 with and .4 hight that would force mover surface area between layers
That tool will be really useful. I tuned printers to print really fast before and agree with all of your observations. Using the tool will give much more confidence on what is possible. Side note: it works great to just set all printing speeds in prusa slicer to the travel speed and put the maximum volumetric flow into the material configuration. I'd like to see an investigation on the reason why printing hotter works so well. Does the temperature gradient between thermistor and inside of the nozzle increase with higher melt rate? Similar to the thermocouple inside the nozzle on your recent video.
Besides moving to Javascript, which you already mentioned was in the plans, I would suggest the option to specify different nozzle sizes (edit, as you’ve pointed out, it doesn’t matter for this test) since I'm pretty much exclusively using 0.6mm nozzles. In any case, great work!
If you want to print faster on Ender 3 and with good quality you should use klipper instead of marlin... You can print PLA 100mm/s at 215°C. Outer walls 75. An you will also not have problems with models with lot of segments and it will be smooth.
@@urgamecshk ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-MaUU8stsZPo.html ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-yAfalR7-Tvw.html It is just better and faster. It runs on raspberry pi, not on printer mainboard. Printer mainboard MCU have just very small firmware which allow communication with Raspberry. All calculations and kinematics etc. is done by raspberry with much faster processor, so timing can be much more precise and with much more advanced kinematics and calculations. Also it is easy to configure everything, you dont need to recompile and flash firmware after each change, just save cfg file and restart printer MCU and you can edit it from web interface, because everything is running on Rpi. Linear advance is not supported by marlin on 4.2.7 creality board. But klipper pressure advance works + input shaping. And you can also use klipper with octoprint (+klipper plugin), but octoprint is slow, so you need Rpi 3 at least for octoprint + plugins to have responsive web interface. I prefer fluidd. Again it is faster. Runs perfectly smooth on Rpi2. Disadvantage is only, that you cant use printer without Rpi. You cant print from SD card like with marlin. But who is doing that ? Everyone is already using Rpi with octoprint...
There's a couple of things not mentioned here that I have found to make a difference in the quality at speed on bed slingers like the Ender 3 V2. I'm able to cut about 10 min off of Chep's profile when applied to a benchy with these differences. I also try to keep my flow at or below 8.5mm^3/s. -When over extruding on a nozzle, you can go down to one wall with little strength impact (one 0.6mm wall with a 0.4mm nozzle = two .45mm walls) -Lower your bottom layers to 2, you dont need that many on the bottom usually -0 "Extra Skin Wall", so the top and bottom lines will connect with the wall and not its own wall. I find this gives much cleaner prints and make adjustments more direct. -"Lines" for top and bottom instead of Zig-Zig for the same reason -10% Lightning Infill. Its called lightning for a reason. If you must do Cubic, at the very least do "Cubic Subdivision" and mess with the "Gradual Infill Steps" -3 "Skin Edge Support Layers". This makes up for the lower infill density without adding much more time -50mm/s for Infill, Walls, and Top and Bottom. This is where you are honing in the flow rate. -150mm/s travel. My ender 3 V2 can go faster but I fell it shakes it too much any faster -Standard Acceleration and Jerk (but you can go up to 800mm/s^2 and 10mm/s with little impact. Chep's uses 20mm/s Jerk which I think can be seen in the quality) -Combing off and Z-hop on instead. This cuts down travel time way down, but make sure your retractions and wipe distances are dialed in. -Skirt for build plate adhesion with distance set to 2mm and min length to 50. Why do you need so much purge on the skirt? Hope these tips help someone.
13:40 i use Arc Welder plugin (free), its transforms lots of line segments to arcs, im not exagerating, some NC Files got 90% size reduction while INCREASING quality, i dont know why this is not included on the software from the box.
For many years arc feeds were not supported...so slicers didn't include it. Now that it's pretty much universally supported there is no reason not to use arc gcode. Older 8bit processors sometimes struggle with arc interpolation at high speed...but not as badly as 'the serial port literally can't keep up' with 1mm segments.
Big thanks for this tool. Constantly keep underextruding on my Voron when trying to go fast because I have been too lazy to do extrusion tests. Time to get this voron breaking speeds.
Very useful tool - many thanks. I have found one oddity; which is that on my Ender-3 V2 (with Jyers firmware) the hotend is set to 0C while the last blob is still printing (as if it's reading ahead in the gcode).
So I was printing a benchi at 700 mm/s and the silicone sock on the hot-end melted to it and made a huge mess, but still had a really great quality benchi!
When the first layer is uneven, there will be blockage at the beginning of extrusion, which will affect the accuracy of the flow. It is recommended to quickly raise the Z-axis to 2mm at the beginning of the discharge, and then slowly raise the Z-axis。
Duet has a new way to tune your hotend while extruding filament. (Heater FeedForward M309) I have not used yet but could have a big impact for fast printing. Previous Hotend tuning just gets the block heated, with no cooling effect from the filament. With using M309 while tuning hotend would enable the hot end to Keep its desired temp no matter the flow.
I've recently been wrangling with the maximum volumetric flow capability of my hotend using the spreadsheet produced by Stefan. Firstly, as I've previously commented, Stefan's current spreadsheet doesn't work with delta printers (or any other type that has X=0, Y=0 in the centre of the build plate). Secondly, having written my own spreadsheet based on Stefan's, my next problem has been that all the printed "nuggets" were the same size and almost half the weight that they should be with a range of volumetric flows from 6mm³/s to 20mm³/s. Visually, the extruder was spinning at the same rate for each setting. Looking further, I discovered that, for RepRapFirmware RRF2.03 and later (not for any other printer firmware!), M203 used to set maximum feed rates, has a default *minimum* feed rate of 30mm/min. This can be changed with the I parameter (that's capital i) once one knows of its existence (it's clearly shown in the Duet3D GCode dictionary!) At the default feed rate, with Stefan's method, all volumetric flows are constrained to a minimum of about 24mm³/s, which is optimistic for a V5/V6 hotend (hence my 50% extrusions)! I've now set M203 I6 (with my usual X, Y, Z & E maximum speeds) and the gcode runs my extruder at much more sensible speeds and produces printed nuggets with weights that are where I would expect them to be, over a range of volumetric flow rate from 2mm³/s to 18mm³/s. I've produced some lovely charts from the data and I now know how useless my hotend is! Thank you, Stefan, your method works well. Hopefully, this comment will help others using RRF.
By the way, I see in your start gcode that you first set bed and nozzle temp, then home and lift, and then wait for bed temp. Many people would prefer to use polymer thay most commonly use (not PLA), and many have glass over heated bed that needs to be heated a little bit longer. I would suggest other order: Set and wait bed temp, then set nozzle, home and lift, and wait nozzle.
Thanks, I am about to print mine on a CR6 SE! Recently noticed that the color of filament requires some changes. I want to fine tune my Prusa Slice profiles for my filament colors.
I am using your tool. it's great! choosing an individual start height would be great for individual blobs. Mine has trouble sticking to the print bed and i don't want to work around with changing my Z-Offset
I love all these videos that use Stephon's engineering and testing to speed up my knowledge of printing. But why has there never been a video release for the testing of the Prusament PC Blend carbon fiber filament? I am very interested in this filament after seeing the PC blend test. The carbon fiver version was shown in the regular PC blend video, so I am curious why the CF version video never comes out.
One thing to remember: it's not wise to go beyon certain temperature (especially with PLA) on all metal hotends. While my Dragon HF can go up to 40 mm^3/s without any problems, it's valid only for PETG and ABS, not at all for PLA. I print at 205 °C as I have found that going at 215 °C allows me to go beyond 20 mm^3/s with PLA, I will get a clog due to heat creep (and the fan is Sunon 4020 capable of 7 cbf/s). Even for other filaments I rather stick to low temps of material as I even got clog for High temp CF Nylon which wasn't funny at all to get rid of it, it was stuck in a melt zone which on HF Dragon is quite long. So yeah, I can print at 36 mm^3/s with PETG without any problems at all, but I stick to 20 with PLA. That means, 100 mm/s with 0.6 mm nozzle and 0.3 mm layer height. Maybe copper nickel plated nozzle would help with flow, but I'm using tungsten carbide nozzle all the time because I'm too lazy to change it every time I print CF Nylon. So it depends. Stock Ender hotend is quite limiting if you want to print fast. It's not able to push enough filament (and extruder too) so the best and primary upgrade for any Ender should be all metal hotend (I strongly suggest Phaetus Dragonfly or at least SliceEng bi-metal heatbreak + copper nozzle) and BMG extruder (clones from reputable makers like Mellow, TriangleLabs or Fysetc works great).
Great work stephan! The tool looks easy enough to use as a spreadsheet. Will update my comment once tested on my newly upgraded Marlin TronXY X5SA with cheap all metal hotend (0.8mm) and Marlin FW, (having inset issues with ABL).
Deine Videos sind immer großartig! Im not sure how to get attention for the online hotend flow page...but please put an option for deltas in the gcode. I have a lot of (large) printers and for the majority of my printing, I still prefer the delta!
Are you planning to do a video on Klipper and input shaper? Can't help but notice the ringing on the ender 3 part. Set it up a while ago for my bed slinger and was absolutely amazed at the results.
Hi Stefan! As allways very good educative and engeneering video! great work. It's almost all factors well researched, studied and put it on reality, the conclutions matches expected results... but it's allways a factor to deal with... consistencies on filament, because tolerances +- 0.05 mm it's good for normal printing, but with that extreme profiles I think tolerances more than +-0.02mm will be a terrible problem. Cheers!
Good work, but you can still make it faster. For example, prusa slicer has variable layer height. This is built into the current prusa slicer. Now use the concept for all the other parameters: variable infill, variable perimeters, variable speed depending on distance traveled, variable hot end temp. The bigger the part the greater the reduction in time for printing the part. Also, when it comes to top layers change the settings to give the finish quality that will be seen by the eye, otherwise who cares if the infill looks bad, right?
Excellent idea, Stefan. Thank you for doing this. I've been wanting to measure the volumetric flow capability of my printer following some modifications, including a CHT nozzle that I bought after watching your review. However, I'm having difficulty implementing your spreadsheet for my delta printer, which has X=0, Y=0 at the centre of the circular build plate and not at one corner as seems to be assumed by your macro. The gcode output from your spreadsheet places many of the print positions outside the build area. I'm going to manually edit the gcode output to place the stripes and blobs on the build plate. I don't know how difficult it would be for you to edit your macro to allow for different zero positions and build plate geometry, but it would be useful (please!).
If you want go fast use Klipper firmware. But the big problem is the faster you print the more time you need to tune it in. Cooling/Print temp and min layer time for mechanical stability.
Your test is very good and precise. Thank you. However this is for a constant flow what doesn't happen 100% of the time during prints. My reults show 10mm3/s, but, while printing vertt fast it has a pick of 20mm3/s without under extrusion for 1 or 2 seconds - with my Titan stock! You would better test it again in real prints! I'm using Klipper at Tronxy X5SA Pro , PETG 230C.
So In a bid to see how one would do on my Ender 3 Pro I spent the whopping £3 and had one delivered. It looked just like the one in the video, only it came assembled. I installed it into the Biqu H2 and was quite surprised to find it did not fare as well as the original nozzle. My scales only do 0.1g resolution, but 8mm3/s to 14mm3/s all came in at 0.7g on the original, though the last couple were visually smaller the scale couldn't show me the difference. On the high flow nozzle it had dropped to 0.6g at just 10mm3/s and only dropped off more at 14 and above. My conclusion is that the nozzle is just far too loose in the hot end (there's a LOT of wobble until it hits the heat brake) and so it's suffering from a lack of heat transfer. I only have CPU paste to hand which can't handle the temperatures required and I have no desire right now to buy some Boron Nitride paste to improve the contact between the nozzle and the hot end, for now it shall live in the drawer.
Even with small changes to the flowrate the weight of the extruded material changes and you are seeking the first significant drop. But there are still smaller drops below that value. Wouldn't it be nice to adjust extrusion according to exrusion rate? Eg. when it's printing a straight infill line with high speed, the extrusion would get a 30% bump to compensate. Even with your prints I see irregular outer perimeters and even though the same wall-line should have the same speed and extrusion, this might still be interesting.
Awesome evolution of process. One moment of curiosity is less the process and more you jumping between programming languages. Given what you've been doing, and your willingness to publish them open source, you could probably consolidate some of what you do into a library suite. You're one of the few RU-vidrs/streamers who actually does any coding. Many programs didn't just materialize, they start off as a collection of components that are assembled into programs. Between reverse bed leveling, gcode bending, extrusion benchmarks, gradient infill, and the general trend of your tests, you could possibly start to make test suite systems "specify some printer params and base filament specs and profiles will be generated" and perhaps this eventually evolves to test slicers and CAD tools. It may be a bit extreme and way far thinking, but it may be a thing... I strayed too far from the point: you jump between multiple programming languages. There is much that can be done and sticking to one for an amount of time may make the work faster/less frustrating.
Flow rate has been my biggest issue. I'm using a volcano with a 0.8mm nozzle on my Railcore printer. I usually print at 100mm/s but the printer can go much faster if i crank the temperatures way up, but quality suffers quite a bit on some prints.