I would like to add that since doing more testing and looking at more results that I think the flow rate in this testing is limited by issues with providing consistent extrusion force due to the extruder design rather than the true flow limits of the hotend itself. For best up to date information continue watching other force extrusion testing as we all begin to understand this kind of data much better.
Great work! First steps towards closed-loop extrusion, where we know EXACTLY how much we're actually extruding. BTW, it'd be helpful if you publish the data in terms of extruded mm^3 / s, instead of mm / min of filament.
I design coffee brewing equipment and one of the issues that occurs is temp dip due to the introduction of cold water. You can overcome it by turning on the heater before (best) or right as the water starts to flow. The temp probe is blind to what’s happening but we know cold is coming and can compensate. It might be a good idea to make a PT1000 sensor that checks the temp of the filament as it leaves the hot end to check the variances in temp during extrusion. Someone smarter than me can do the math and determine how much energy is needed to melt filament at any speed and program the heater to respond to speed then use the existing sensor to keep the hot end stable at idle. The main issue here is having the filament at a know temp so the calculations can be calculated. That said the dip might be heater catching up because it’s blind and took time to catch up. Take the heater and turn it on full power and start running filament until it skips, then you will know the actual flow limit.
Would be interesting to see if a command to turn the heater on just before extruding would smooth the peak. A feature like this may help keep a consistent temp during printing where the filament starts and stops.
I've been thinking of ways to heat without using PID tuning so it can be removed from the equation. It'll need a well tuned machine, but should be possible.
@@Vector3DP Who is the Gcode wizard? Monitor the extruder motor, do the math, maintain filament temp, and it should be possible. Optionally you could count forward pulses of the stepper to do the energy math and leave the PID for ide stability.
On duet boards you can use M309 to configure extra heating for every amount of filament extruded, and they recommend calibrating it by extruding at maximum volumetric flow and watching temperature, it there is a drop at the beginning you should increase this parameter a bit.
Yes, definitely wanna see more, especially tests within groups: different extruders same hotend, different hotends same extruder, and even different filaments of the same type with the same setup. Are all PLAs the same? All ABSs? You have opened Pandora's box my friend, and we thank you for it! 😁
About the strange and erratic dips in the curves at high temperature: they could be vapour bubbles, which can form when three parameters are right: Pressure, temperature, traces of moisture in the filament.
wonder how changing the extruder would change it love to see how the lgx works with the same hot end. do the larger gears give a better flow rate due to a more consistent force?
THIS! This is what I had in mind for months. Not just finding the hardcore limits of hotends but "how quickly can I print without compromising quality?". And you nailed it. Congratulations. I'm really looking forward for other materials and hotend tests. Also, please note all these procedures can be automated so that you can test and get results within a few minutes and minimum effort :brain exploding:. Are you considering sharing your design. I'd really appreciate it. And for those aspects that you pointed out in your video: - About the first peak might be due to slightly lower temperatures in the tip of the nozzle, as it's in contact with cold air and not protected by the silicon socket. Once the tips clears and warm filament reaches, the pressure gets lower. It's somtheing I usually notice when extruding manually when I change between filaments or during cold pulls. - About the drops seen during extrusion, they seem to be at low flowrates and/or high temps, so I think they might be produced by humidity of the filament that evaporates in the tip of the nozzle. That creates a low demand of force until more filament reaches the tip of the nozzle and builds up pressure again. Sorry in case my english is a bit poor, as it isn't my native language.
Awesome testing methodology! This is what I expect to really help with tuning max speeds in the near future, specifically with the Prusa XL implementing a load cell returning analogue data to the mainboard. With how proactive the community is with their firmware development, I wouldn’t be surprised if they develop an “auto-calibration” feature for finding maximum extruder feed rates for filaments and making it easier for people without lots of background knowledge in 3D printing to have their machine tuned to the ideal settings.
The constant ripple in the blue and yellow line, is I think the teeth of the extruder gears, making the filament change speed. You can see the same pattern in both, but the blue line(rotating faster) is closer.
I think its showing the concentricity error between axle and outer surface of the gear wheel. Where the radius is larger, it wants to push filament faster, therefore creating a force spike. With lower feedrate the hotend is able to melt it better, so that spike disappears there /the wave is longer, we dont see it as well...
I think your waves in the force diagram at the top two feed rates are due to a concentricity/roundness error of your driver wheel in the extruder... where the radius is a bit larger, there will be more filament going through the nozzle, taking out more heat, therefore restricting flow and rising force. With lower speed the problem disappears because the nozzles pid tuning can work against it... and also because longer waves are harder to see... Does, that sounds resonable? Do you have a way to measure the runout on the wheel when turning the motor?
This is a very interesting experiment. As for the small variations at low flow rates, it is rhythmic and thus may be tied to your stepper motor steps. I'm guessing you're running a 1.8 degree stepper, I would be very interested in seeing the data from a 0.9 degree stepper. Wonderful job, I really enjoyed it.
One reason your first peak from slipping might be higher is that is when the driving gear is clean.bas soon as you start slipping the gears will fill up with material and make slipping worse. I've had similar issues trying with consistency while calibrating flow rate, if i didnt clean the extruder gear after a run where i new it had slipped.
I don't think I've ever seen you this excited Adam 😂 To answer your question.. YES.. "Please sir, may i have some more?". I can't believe we've gone this long without actual quantifiable data and repeatable testing for measuring hotend performance. It's about god dang time! Thank you! Now you just need to procure ALL THE HOTENDS :D I'm guessing there's a bunch of interesting observations to be made about different filament brands and types as well :)
I've been sitting on this data for 4 months because i didn't know how to make this video too. I was so excited that i couldn't control my own thought process 😅
Testing filament itself is harder. I need to the rig to be ultra reliable and consistent before that's possible I think. But should be doable. There is literally SO MUCH STUFF one could test with this. It blows my mind every time i start thinking about it.
Very clever and incredibly informative. If you don't test you don't know. BTW, there is a term for the friction which prevents stationary surfaces from being set in motion. It's called 'sticktion' (I kid you not - well know to suspension tuners) and it may explain your initial pressure spike.
The peak in force seen at the beginning of extrusion is, in my opinion, due to the breaking of the more or less soft, swollen and sticky plastic plug formed just after the heatbreak, when there is no movement. When this plug has been forced, we enter a dynamic regime where the filament is lubricated by a layer of molten plastic, with much less resistance. The size of the plug, its temperature profile along the filament axis, as well as the coefficient of friction with the extrusion channel walls, all depend on the hot-end and its technology. Your test device will be able to show the real or supposed qualities of the different offers on the market in this respect. BRAVO!!!
You will also be able to determine how long it takes for the plug to reform, which also depends on the hot-end. This information is valuable, especially when using several different hot ends, or a tool changer. It would also be useful to know the "negative" forces exerted during retraction, or during extraction of the filament, depending on the temperature...
I wonder if the initial peak is similar to "water hammer". You need to get the filament to move, both in the hot end and also pull the filament from the spool. I wonder if a "soft start" of the motor could make a difference here. Though I have no idea if it will be of use in real printing conditions.
I LOVED THIS!! I'd love to see a contrast with other heat blocks, especially a round one like the REVO from E3D, or different materials of the blocks, like copper vs an aluminium vs brass. Like you said, there's so much you can test with your new machine 😱😃
surely a Volcano would be a great example to clearly show how the block material can change performance, as it has a decent amount of mass and a is easy to source on a wide variety of materials with the same dimensions.
Great work! My guess for the peak at the beginning is slightly cooler filament at the nozzle tip (which is definitely cooler than the point the thermistor is measuring) that needs to be pushed out before the hot/soft flow kicks in... Could test that with a poor heat conduction nozzle like a stainless... Should worse / the peak higher I assume.
Nice work, super interesting. Now to find a way to measure while printing without the results being influenced by the lumps and bumps under the nozzle. :-)
Wow! Incredible job! It is very interesting to compare hotheads and extruders in the case of regular peaks, retract testing and pressure advance algorithm. Thank you!
The waves in force vs speed graphs up and down are most probably caused by the hotend heating/cooling cycle as the filament goes through and cools it. It will take some time until the thermistor detects lower temperature than required and the heating block receives power and manages to reach the temperature again. When the hotend PID is tuned properly and has enough power to cope with the filament volume, the wave is smoother and the extruder pushes more constantly the filament, as the temperature in the hotend is kept almost uniform.
2 года назад
Now i have seen the whole video ! This is Breakthrough in 3d printing!!!! This should be possible for a 3d printer to detect problem and correct it before it get out of hand. Really love to see more testing!! Great jobb!
I hope you can set up a good array of tests/benchmarks. And then perform them on different setups v6/dragon sf/dragon hf/rapido/revo... bmg/lgx/lgx lite/Galileo/sherpa... Cht nozzles
My hypothesis for the initial 4kg of force. the initial slip happened at 4 kg because your extruder gears for still clean, after it slipped the first time it clogged the teeth with material thus reducing their gripping power by approx. 40%. I am new to 3d printing so I could be out to lunch but it would be easy to verify with that tester you built.
Great work! I had planned to compare various filament feeders with the traditional methods .... now I'd like to build a test rig like yours. Will you be sharing a BOM/buildlog?
And then i thought, why hasn't someone else come up with this idea yet? This is awesome, keep on doing your thing!, you are on the right way! I am very certain this will improve the whole field of 3D-printing a lot over time...
Very interesting - still lots of unknowns to unpack. When I saw the one-time pressure dips at the beginning of some of the low pressure curves, it made me wonder if the software had paused/delayed moving the stepper for a moment. You might want to consider probing one or more of the stepper motor wires with a scope so you can confirm that you’re getting the continuous stepper motor movement you’re currently assuming you’re getting. Even further along those lines, probing the current in one or more windings could be informative, as well as being an easy way to check for motor slipping without having to be there for the whole test. Would be interesting to see how closely motor current corresponds to pressure in general. Lots to think about!
That's pretty sweet. Only think I think that could make it batter is an encoder wheel before the extruder to measure flow and pressure. Definitely should test different hot ends and nozzle combos. Id like to see the Slice Engineering Mosquito Magnum+, The dragon with high flow, with and without Bondtech CHT High Flow Nozzle.
My first thought is that the wobble in middling flow rates can be attributed to the heater PWM. But at 12:45, the periods look to match at all temperatures, but with a random starting offset. We'd expect different periods and a pattern in onset of the pattern at different temperatures. This could be saying that it's due to eccentricity in the feed gear. If the period matches up nicely with the period of filament slipping, that's a stronger argument for the drive gear. If there's eccentricity, there will be a spot on the gear that's weaker than the rest, which is where slippage will happen. Having the actual heater drive signal would be killer to distinguish between the two. A neat test rig, with lots of potential!
Well this is awesome! Please do more tests and with other popular combinations. Dragon hf, rapido, mosquito and orbiter or other extruders! Just awesome!! As for the initial spike it must be the axial compression of the filament as it gets pushed until it equalizes.
Great video! Some ideas on the data: 1. The 4 kg peak load at 6:17 is highest because the filament has not yet been damaged. Once material is sheared away, the drive gears cannot develop as much grip / extrusion force and later peaks are lower. Not sure why this peak is different from those later in the video, as there are many potential related factors (gear tooth incident angle at the moment of slip, velocity dependence, upstream filament tension from the spool, etc) 2. The initial peak-and-valley pattern in all tests is likely due to the viscoelasticity of the material, and the initial transition to plastic flow. When extrusion begins the material is uniformly stationary, thus a shear layer (boundary layer) must be created around the plastic slug of material before it can flow. Initially, it's more energetically favorable for the material to simply compress and store energy elastically. However, with increased load, some of that energy begins shearing the boundary layer. Peak load occurs at the moment when the boundary layer fully develops and plastic begins to flow freely (also, the built-up elastic energy is released). Just before the peak load moment, the extruder is fighting against BOTH shearing (flow) and built-up elastic compression of the plastic slug. Once the boundary layer fully develops, most of the compression is released and the extruder ONLY needs to overcome shearing. Also - due to viscoelasticity, it's not surprising that the peak force is rate-dependent. 3. Fluctuation frequency correlates strongly with flow speed, suggesting a periodic source of error in the drivetrain (e.g. imperfect gear mesh) or at the drivetrain interface (gears gripping the filament). Many other potential disturbances - such as structural resonance - can likely be ruled out as their natural frequencies would remain more or less constant regardless of flowrate. 4. The source of dips at 10:00 is unclear, but don't forget about upstream disturbances. One example is fluctuation in tension of the filament entering the extruder, as it is being pulled from the spool. Extruder force will change depending on whether the upstream filament is slack vs. taut, and if the spool must be "yanked" due to being misshapen or rolling intermittently. Even if the extruder is not slipping, upstream tension on the filament can propagate along its length and affect measured force at the hotend. It can also change the "apparent" peak gear force, as gears not only work against extruder back pressure, but also spool feed tension. If I had to nitpick (and it's hard, again, great video!) - I'd just ask you to indicate the data sampling rate on any graphs where time is used as an axis, so we can immediately check the time-resolution of the plots. (Also, make sure the chosen chart style doesn't use any sort of smoothing - these charts might not be, just hard to tell given the nature of this data)
I think the initial peak is higher as the extruder contact with the filament is clean. It then gets filled with plastic so has lower coefficient of friction (static and dynamic).
Great job! Have you considered the on and off switching of the heater cartridge as it tries to keep a constant temperature? You would see a force increase as the heater cartridge turns off and then see the force drop when the heater cartridge is turned back on.
Couple thoughts to add to the discussion. ¹ is molten filament a non-Newtonian fluid? Could perhaps explain the initial resistance, then once it starts flowing it flows much more easily? ² do the periodic dips relate to the hotend cycling? Would adjusting pid change how they look? ³ mm/s³ would definitely be a more useful metric as it's easy to use with the graphical displays in slicers and you can set extrusion limits based on mm/s³ (in SS) ⁴ what were the criteria you used to determine your best extrusion rate? It felt kind of arbitrary, maybe I missed it - but having simple prints to go along with the data could help with developing some useful standards. Then a person could test a handful of settings, run it through Excel to build a graph, interpolate, and find some "ideals" for things like draft quality, best quality, speed, strength, etc. This is some really good stuff! Like others, I'd love to see some comparitive data on competitive hardware - extruders, hotends, nozzle materials, heater cartridges, ... The ability to help find an ideal configuration for each of the parts in a printer is extremely powerful and intriguing, and as you said - the amount of data you get from this is phenomenal. Last question, lol. Did you ever combine the load cell data with the extrusion test weights? Was there enough there to characterize how force on the extruder translates to underextrusion?
My first thought on the initial spike in the graph would be the equalization of the compression of the filament against the extruder gear. A lot goes into forming the footprint in the filament that the gear will be pushing through all the way until on the other side of the wheel and compression lets off.
Nice testing method! My guess for the first peak would be heat creep: while pre heating the point where the Filament is soft moves up the heat break and it requires more force to break that loose. After it has loosened the 'meltzone' is lower down, so there is more space with low friction where the Filament is still solid.
Same for the dips at low flow. Heat creeps up and slowly increases friction (and force) until the whole blob of molten Filament rips off and moves down. Because it's already soft, there is a period of low force. The wideness of the dip is how much of molten 'heat creep' Filament there was.
I guess max force is lower after skipping is because at first gears create grooves in filament, but when gears slip, they literally shave those groves and now the bottom part of the filament is not gripped so well, only upper one works correctly. It is basically what happens when the nozzle clogs - you open the extruder and find lots of plastic shaving near gears.
Hey! This is great. Stefan’s measurement protocol to s very useful for a functional, in-service printer. But this…. You came up with a whole purpose-built test rig. Marvelous. But you are missing a few parameters: Get a wheel that is coupled to a rotary encoder to measure actual filament speed and throughput on a continuous basis. Under-the-curve calculation will yield total mass of plastic extruded (if you know density and trust the diameter). Also, get am ammeter on the hotend heater to measure heat/energy going into the hotend to melt the filament. Some math, coupled with the extrusion rate, should indicate how well the hotend is actually melting the filament (is the filament completely melting? Is it merely softening? Is the outer perimeter of the filament melting, but incompletely at the core?) Also, depending on how you’re controlling all this… if you use Trinamic drivers, they are capable of dumping a BUNCH of diagnostic data via UART - so you could track skipped steps, current, etc….
A rotary encoder would take this to the next level. Between the pressure and volume data, you could have a hotend tune itself. I wonder how the pressure relates to the internal stresses that Stefan was looking for in his CHT testing. I feel like a lot of these forces relate to linear advance as well. If it would allow you to automatically set those parameters that would be amazing.
Initial force spike - I would hypothesize that the initial spike in force is due to the angular momentum of the system (gears, motor, spool, etc.). That is until everything to the spool is rotating at target speed you have an additional acceleration force to bring them to speed and then everything would settle on the friction and pressure forces of the system.
Some of those spikes are gonna be the pid settings. Because its not measuring at the filament, and its not heating at the filament, the lag involved is gonna create spikes. Because its based on pids or bangbang control it would be rhythmic as the extruder cools from releasing heat into the filament, that heat is lost from the extruder to the filament, it cools, the extruder sees it change, turns on the heater, repeat. What we REALLY need is a PID autotune that extruder about 200mm of filament to decide how much it needs to heat for how much flow, and autocorrects when its running the extruder.
Awesome sauce! Please consider testing different filaments like PA6/12, PC, and other engineering grade stuff. In particular CF filled would be cool to see how the pressure is effected by the additives.
10:19 Perhaps the dips in force are due to steam bubbles releasing from the hot end. Compressed steam bubbles support pressure until suddenly bursting at nozzle exit. It would be interesting to see if deliberately water-logged filament exhibits the same results.
Awesome job. I feel the one thing 3d printing is missing is definitive information. Granted there are tons of combos but it seems each person keeps their perfect settings to themselves. It's such a waste of time having each hobbyist have to go through all the steps themselves. Keep up the good work.
Stepper motors can easily "get confused" and stall when run at low speeds. I assume the dip at low flow rates could be related to that. It happens at some sort of "frequency"/turning speed that the stepper can stall... you could try with some combinations of gears and gear ratio that would give you the same filament speed but at higher RPM for the steppers to eliminate that stalling. Just a thought though. Hope this helps.
Thought on the initial peak. That could be the initial melt time. If you work under the assumption that most of the heat is transferred via conduction, that initial peak the would the "cold" filament essentially crashing in to the nozzle, then at it heats it begins melting and lowers until you reach the point it is melting at your feed rate. The variations in force could be multiple things. Change in velocity from the hobbed gear. Could be slightly non concentric leading to slightly different velocities, or even potentially the same issue caused by the difference in diameter between the peaks and valleys. The stepper motors, they don't apply a constant torque as they step, so you will get changes in force depending on where it is in the step.
You should look into using an arduino to run the stepper and also read the force at the same time. The advantage of this method is you can program a series of moves for the stepper and record the results very consistently. I used this method with a stepper and load cell to cycle load parts I was testing with lots of success
Love this setup! I would be curious to know if your PID settings on your heater cartridge and thermistor explain most of the variation you're seeing before you start to truly skip, especially in the higher flow rates. Especially considering that we tune with no flow... The delay in measurement/variation between inner core of the nozzle and the thermistor might be the explanation here.
My guess as to the source of the force spike at the beginning of every test is the initial force required to overcome the friction between the plastic and the PTFE tube (if present) and the brass of the nozzle.
I'm betting the massive peak is the actual pressure the gearing can produce with clean teeth. After the teeth slip the first time and fill with abraded filament, it drops off. Could also be initial plug being blown out.
Have a look at the Renkforce RF2000v2 printer from Conrad. The printer has a fixed bed and measures the distance between the nozzle and the bed with such a measuring device. There also exists a custom firmware, which changes print speed according to the filament presure in the nozzle, measured during the print
I would guess the pressure troughs are possibly caused by the delay in temperature correction. Basically the hotend starts to heat up and overshoots slightly causing a pressure drop. It could happen pretty quick and depending on how the board is measuring temp ntc100k are pretty darn inaccurate sometimes.
Fantastic work! One thing you're missing is a Gage Repeatability and Reproducibility (GR&R) study. This will tell you what the measurement error of this set up is. Some of the wobbles you are observing (not the big ones from teeth slipping) might be measurement errors.
I think you got to measure the filament travel too. As the filament is pushed out at a stronger force. The plastic on the surface where the teeth of the gear is digging into the filament starts to deform. I'm sure you will uncover more interesting data from knowing how much actual material was pushed in.
The Prusa XL uses the same principles as your testing rig to determine if the filament is no longer flowing properly. I would expect to see some changes in 3D printer design and firmware in the future.
Adam, since you asked, it's very interesting. For the first time we measure the dynamic performance of the hotend. BTW, what is your loadcell sampling rate? If you use HX711, you can increase up to ~140sps. Next step is to measure with something that simulate the back force from the printed part. E.g. have some rotating drum below the nozzle.
Yes, i am using Hx711, I am using quite a high rate, not sure its as high as 140hz though. Nice idea on the rotating drum, I do think this will affect pressure in some way.
Super interesting stuff! It would be immensly cool if you could plot the current to the heater cartridge and the actual readings from the termistor in the hotend together with the force. That way you should be able to find out if it is the PID tuning that starts going bananas or maybe the heater just can't keep up
woa! amazing, tecnically you could add a multi fillament loader to the setup and just one click collect data to test/compare different fillaments automatically which is mindblowing to me
Great work! You could raise your filament slip force by increasing your drive gear count to 4 or 6. You'd have to add a small gear between each pair to reverse the direction. Ultimately you will get to a point where you have to improve the temperature control of the hot end. You might be able to tweek the PID values. But at some point you'll have to consider the joules required, the response time, and the future steps in the G-code. I think you are onto something extremely interesting here. Don't stop. There's so much more to learn and room to innovate new advancements in the industry. Document the eff out of it so we can prevent some goon from parenting it. ☮️❤️🌈🧘🏽
Fantastic! I've always thought that flow rate and temperature need to be correlated to create a printing profile that works for both fast and slow printing. This will give a good indication as to what the correlation should be.
Superb !!!! Please do more !!! You are RU-vid at its best. Can’t decide if the name of the test rig or the concept / results and execution are best. The name is sooooooo good :). Ty for making this !!!
Great exploration! I wonder if you can add this load cell to the working printer and test the difference (mounting the extruder to load cell, not the hotend?). The "unobstructed" extrusion can behave differently compared to real-world use where you push filament in a tight space.
Very interesting, you cannot do this for one horend obviously and now everyone is waiting for an gradually growing list of combo’s. Also measuring sale horend and cheap vs expensive or brass vs steel nozzle of cht vs regular and so forth. It also confirms my subjective feel of having to need to dial down print speed from the traditional measurement of extrusion rate testing. That was an eye opener as I was hoping to push my ender6 drastically above 100 mm/s with 0.4 nozzles. Currently figuring out how to mount the dragon mosquito high flow knock off paired with cht nozzle. Curious if that would make it into a test.. or just the mosquito with cht as reference point?
This is the first real data I have seen on the inner workings of a hotend. I can see that load cells will likely become standard sensors. I am curious if this configuration would allow the sensor to be used for bed leveling as well as monitoring pot end pressure.
@@Vector3DP They have mentioned that it is used for detecting both the nozzle to printbed and alignment pin interactions, and it can also detect filament jams, so it sounds like it might be able to get similar data to what you have done here. Nice job on this work, by the way!
This kind of testing and analysis would be useful for the future of 3d printing. Id like to see the matrix hotend tested to see how it compares to the hemera, but also id like to see the results with ABS rather than PLA since that's my preferred material
Maybe take a look at your stepper driver and settings, especially TMC drivers do a lot of funky stuff to reduce noise/power usage which could result in some of the strange effects you are seeing. Also going to a SERVO42C Stepper could improve this rig, as it offers closed loop stepper control and through some custom firmware could potentially give you the actual stepper movement, current etc. to correlate with your data.
perhaps one reason for the 4kg peak was that it could initially do that when clean, but after the first slip there was some residual filament ground off in the drive gear teeth reducing the gripping force?
You could be right with the up peak when the motor driver is turning on. Maybe the driver is forcing the stepper to the next full step. It should be easy to see, if you plott the graph with retractions or drivers turned on before. However, from cooling fans we know that they need a bit more power to start rotating, could be similar with steppers. You’re probably right with issue 6 and the ripples.. measure them in relation to the extruder steps per gear rotation, I bet it’s the teeth on the gears we see here. Keep on investigating, it will at least help the nerds to build better extrusion systems:)
I would go a (micro) step further and even test with different extruder gears (single vs double, for example) to see if this provides a quantifiable improvement and how much.
What if the flat part at the beginning was actually the heatblock cooling down? that would explain why the force after a while is higher, being the nozzle itsefl actually colder than the target temperature.
I wonder if it is possible to get the same information from a stepper driver. The stepper driver could know when it takes more power to complete each step. If that were possible it might become possible to diagnose issues with movement on any of the stepper motors in a printer.
Watching you get excited is awesome I found myself getting excited as well. You make top notch videos keep it up! Also I would love to see more vids on this subject.
Interesting stuff. I will throw my two cents worth in on that initial spike. Although there is molten filament in the hot end, could the filament in the exposed tip of the nozzle be cooler forming a less viscos plug of filament that has to be cleared. A possible test would be to start the system and stop it before it clears that initial preheated filament, then restart it again. The filament in the tip would not have time to cool and there should be no spike. If it is stiction somewhere in the system the spike should still be there.
That is indeed the idea, but the most significant plug is formed around the heat break, when the filament is stationary. There is much less resistance when the filament is moving forward, and the static plug has been replaced by a dynamic seal of molten plastic, between the filament and the wall.
Great content Adam, very interesting! Would love to see more with Brute Forsyth lol. Really is so much that could be done, must have been hard to stop and actually finish a video… You could also use this format to show the affect of things like hot end cooling, as heat creep could have interesting affects on the results. If you’re ever after other materials to put through the rig, give me a shout and can send you some 3DTomorrow filament. As a side note, since you did mention health, probably best to avoid printing PLA above 240C.
Can you log exactly when the heater cartridge was on and off? Also are you measuring the hotend temperature or relying on the settings you configured? Perhaps a second logged thermistor output would provide some interesting insights?
Could the dips and the peaks on the graph be from heating requirements? graph the change in temperature next to these values and see! I know the faster you go, the more heat gets sucked out of the surrounding material to melt the plastic. So you're looking at the metal get too cold and then heat up again. At low temperatures, it would be that the pid settings can maintain temp longer before needing to ramp up heating to keep the hotend stable temp.
I'd be interested to see if the feed rate of the filliment was fully proportional to the flow rate at the nozzle, if it was you could measure the filament feed speed and use that as a target for the extruder to follow with a speedy PID loop i wonder if you could push higher forces whilst keeping the flow rate stable
Interesting analysis, thanks for making the video. The internet would tell you a V6 hotend would flow 10+ m^3/s before running out of steam - converting your summary slide at the end yields much lower numbers, like you mention. Pretty interesting. What would be more interesting is determining how far you can push those values before you can detect degradation in a meaningful way - either visibly, or in print strength, or accuracy. I have a feeling your method is very conservative when it comes to print quality. Thoughts?
This was SOOOO cool. Wow such a simple piece of kit can yield so much. I might have to make one for myseld. I would definately like to see more. Could you possibly test the Mosquito? And maybe see how much better the high flow variant is then the standard flow?
Heat creep up the filament into the heat break can do weird things. Might be the cause of the dips. Longer tests would give time for the hot end to get back to it's set temperature at high flow rates. It'd be interesting to see the frequency response function of the hot end. Look up FRF testing. In this case I'd give it a pseudo-random flow rate then use an FFT to compare the motor speed to the extruder force. This system is highly nonlinear so the coherence of the FRF will be poor and the results harder to interpret.
Could those peeks be because of the heating element system not being quick enough? I assume that those work in sort of pulses in terms of on/off or that the amount of heat energy introduced is like "boosted" into the heating element system and then turned off or at least significantly reduced and not being increased again quick enough. The feedback loop between the thermistor and the heating element may be to slow, given that the thermal mass of such systems that could smoothe this out does not look very large. Measuring the current that goes trough the heating element may give an answer to that and lead to more sophisticated extrusion systems.
