Puzzling extruder speed problem
-
I'm experiencing some extremely repeatable underextrusion that increases with extruder feed rates above 500mm/min with no skipping of the extruder motor and no slipping of the filament. I've gone through a round of support emails with E3d but we didn't get to the bottom of it. The faster I go, the more it underextrudes.
I have an E3D PT100 sensor on a Duet Ethernet with PT100 board as my hotend thermistor, a normal EPCOS100k thermistor for my bed and my chamber heater monitor. I've got an E3D Titan extruder with a common stepper motor, not the smaller pancake motor. I've got a Volcano block with the 0.8mm and 1.2mm nozzles. At room temperature, all thermistors are within 0.3C of each other, and I've confirmed that temp compared to room temperature with a Fluke thermocouple. Then I kapton taped my chamber thermistor to the volcano block, and graphed the difference in temps up to 285c. Turns out the taped on chamber thermistor reads consistently 20C low once it gets above about 100C, but I'm not sure if that's normal for that thermistor, the less contact with the block than the cartridge PT100, or some other factor. Anyway, I did some more testing with extrusion rates and temperatures.
I tested marking out and then extruding 1000mm at various feed rates. I started at 440mm per minute, which was giving excellent print results, but is a pretty slow print on a big layer height. I was able to easily confirm my steps per mm were correct this way as well.
I found that at 245C on the coolest reading thermistor, 440mm per minute got perfect results. No measurable slippage.
Leaving it at 245C, I tried 500mm/min, 600mm/min, 700mm/min, and 800mm/min several times each, cleaning the hobbed bolt with a brush between each test. I used new ABS of a brand I've had excellent success with.- At 500mm/min I was getting 5 to 10mm of underextrusion (10mm before my mark still sticking out of the point I measured from).
- At 600mm/min I was getting 15 to 25mm underextrusion,
- at 700mm/min I was getting 25 to 40mm underextrusion, and
- at 800mm/sec I was getting 100mm underextrusion.
The temperature graph was steady, there was no drop or increase in temp more than about 2C during these tests.
I then returned to 600mm/min (a little underextrusion, same as before) and repeated the testing, increasing the hot end temp 10C per try. I went from 245C up to 285C in 10C increments, and over that 40C range, the underextrusion improved by only about 5mm, still underextrusion by about 20mm.
I dropped the temp back to 265 (it was starting to stink at 285), and then began retesting at 600mm/min, increasing tension by 1 turn on the knob each try. I got to the maximum tension where I was no longer able to release the filament by pushing the lever over, with no more than 5mm underextrusion improvement. (still underextruding by 15 to 20mm).
I tried pushing 1000mm at 800mm/sec on my long-suffering hard-used delta that has a direct drive bowden extruder with a large motor on it, and I was able to extrude at 15mm/sec (900mm/min) on a standard 0.4mm nozzle V6 hotend. On that one I was able to extrude fast enough to see a temperature drop without any grinding but I didn't remember to note the feed rate for comparison after 900mm/min. The 96oz/inch motor will actually stall and start skipping before the filament grinds. I've got a standard 40oz/in stepper on the titan, with no skipping or stalling ever occurring that I can tell. I've even pulled on the filament so that I could experience what stalling or slipping would be like, and it's a very audible clicking noise with pulsations through the machine. None of that occurs during the normal higher speed underextrusion.
I've tried turning the microsteps down by half at a time all the way down to 1/2 (adjusting the steps per mm to be correct), and turning interpolation off and on at each setting, with no change in the results.
Does anyone have any ideas what could be wrong? I know a volcano setup is capable of more than this. I've found people being able to get 40mm/sec (2400mm/min) out of theirs.
Any ideas? I've been working on this for weeks and I'm all out of things to check. There's nothing mechanically or electrically wrong that I can find, I've even rewired the motor twice and used two different (though identical model) motors!
-
Just some rambling thoughts….......
What size nozzle are you using when you push the speed up? It's all about melt rate. Typically a standard E3D V6 will top out at about 8-10 mm^3 sec (that's cubic mm /sec) and a volcano at about 20 to 30 mm^3/sec. So for the same size nozzle, you might expect up to around 3 x the extrusion rate (but it might be less with a small nozzle due to pressure build up). The area of a 0.4mm nozzle is around 0.126mm^2 but the area of a 0.8mm nozzle is around 0.5mm^2. So for a given melt rate, the maximum speed that can can get with the larger nozzle would be around 0.25 of that which you could get with the smaller nozzle. i.e 40mm/sec with a 0.4mm nozzle would equate to 10mm/sec with a 0.8 mm nozzle for the same volume flow rate.
-
Hmm, that's certainly something to consider. I'm going to go off in the corner and play with a calculator for a while.
I've used .8mm nozzles for all of this. heck, the reliable speed is slower with this volcano/duet/titan-direct setup with a .8 nozzle than with my v6/smoothieware combo when I had a .8 mm nozzle. I saw clear symptoms of melt-rate problems with v6/smoothieboard/direct drive bowden with a .8mm nozzle there, the motor would audibly skip if I went too fast. Nothing like that here, no symptoms other than measurable underextrusion. wouldn't I see some evidence of slippage, grinding, or a more normal indication of loss of steps if that were the case? I can get those symptoms to show if I go reaaally fast, like 2000mm/min, I start getting extruder motor clicking and skipping.
e3d's support contact doesn't seem to think I'm hitting anywhere near the melt rate limits.
Maybe it's finally time for me to convert to volumetric extrusion
-
Well, you can calculate the melt rate easily enough. I guess one way if you are just extruding filament in air would be to take the area of the filament multiply it by the length to get the volume and divide that by the time. If you are printing, use the area of the nozzle x the layer height x the print speed. Multiply the results by any extrusion factor if you are using anything other than 1.0.
However, it does seem that there must be something else amiss if the volcano is no better than a standard V6 with the same nozzle size. Have you measured the nozzle to make sure it is actually the size that it says? Any sign of a partial blockage anywhere?
Have you tried running M122 and looking at the diagnostics? That might tell you if you are getting missed steps.
How about motor itself? What is it's rated current and what are you driving it at. For info, I have three Titans feeding a Diamond hot end and use these motors with no problems. http://uk.omc-stepperonline.com/nema-17-bipolar-step-motor-29v-07a-18ncm255ozin-17hs100704s-p-260.html. They are rated at 0.7 Amps so I run them at 600mA.
Just some other random thoughts
-
ok, so working from the speed that I just start seeing an issue at,
915mm of 1.75mm filament (that's how long my stupid measuring stick is, double checked with another tape measure) extruded over 91.5 seconds is 17.5mm/sec cubed but 15mm/sec cubed is fine. Interesting.
yeah I've checked for obstructions and even have gone as far as completely replacing my entire feed path including the nozzle with genuine e3d stuff. No change.
I keep forgetting about M122, I'm going to go look at that.
I've got a 1.3A 40 oz-in motor running at 1.3A but I question the quality. Came in one of those cheap printer kits. I'll replace it with something well known before I start chasing too many more things.
-
I get a different answer (but I'm a 63 year old who has forgotten much of the maths I learned). Area = Pi x Radius ^2. Radius of 1.75mm filament is 0.875mm That squared is 0.765625 x 3.142 (Pi) =1.20559375 mm^2. That x length of 915mm is 2,201.11828125 mm^3. That divided by 91.5 seconds is 24.0559375 mm^3/sec. From what I have read, a volcano can do 20-30 mm^3 /sec so it looks about right to me.
One other thought. I've hardly ever used ABS (the fumes trigger my cluster headaches) but is it harder to extrude than PLA? I'm assuming it is, as the melting point is higher. The point being that you might get a higher melt rate with PLA.
-
Have you tried different brands of ABS, some flow much better than others. 40mm^3/sec requires some flowy polymers.
-
What heater cartridge are you using? Maybe it's not 40w and can't keep up with the flow rate? I have a cloned volcano and with 0.8mm nozzles I can't go faster than 35mm/s, it just stops melting if I go any faster. As such for a large object it's not really faster as a 0.4mm nozzle running at 80-100mm/s would be as fast. However the large object with coarse layers is very strong, one is a lookout from paw patrol that my son has dropped an amount of times that would have destroyed an object printed with a 0.4mm nozzle. So I don't see volcano as a way to go faster, rather a way to print big things with big nozzles which are stronger.
-
I get a different answer (but I'm a 63 year old who has forgotten much of the maths I learned). Area = Pi x Radius ^2. Radius of 1.75mm filament is 0.875mm That squared is 0.765625 x 3.142 (Pi) =1.20559375 mm^2. That x length of 915mm is 2,201.11828125 mm^3. That divided by 91.5 seconds is 24.0559375 mm^3/sec. From what I have read, a volcano can do 20-30 mm^3 /sec so it looks about right to me.
One other thought. I've hardly ever used ABS (the fumes trigger my cluster headaches) but is it harder to extrude than PLA? I'm assuming it is, as the melting point is higher. The point being that you might get a higher melt rate with PLA.
yup, looks good to me. I tried to take a shortcut and got a wrong number. I underestimated just how much plastic that was flowing. I guess what needs adjusting is my perspective! I can't find anywhere if Reprapfirmware supports volumetric extrusion, do you know? I might be able to get this whole thing automated for me. It would be nice to be able to define a maximum volumetric rate, and no matter which nozzle size I put in there, it'll handle the speed for me.
I've been printing almost exclusively ABS for 2 years, I find PLA to be less tolerant on everything except shrinkage (abs is far more tolerant for retraction length settings, multiple retractions over the same length of filament, and a relatively wide glass transition temp, doesn't require as much if any forced cooling for small slow details, etc). I've discovered that some brands are super stinky and make lots of fumes, and some brands make pretty much none. I'll try a roll of PLA I've got kicking around, that might prove this out.
I'm still puzzled as to why I dont see any clues to this anywhere other than in the measuring or underperforming infill. On all of my other machines, when underextrusion has occurred for a chemical, thermal, or mechanical reason, there's evidence. Filament grinding, motor stalling, tubes popping out of PTC connectors, etc…
-
What heater cartridge are you using? Maybe it's not 40w and can't keep up with the flow rate? I have a cloned volcano and with 0.8mm nozzles I can't go faster than 35mm/s, it just stops melting if I go any faster. As such for a large object it's not really faster as a 0.4mm nozzle running at 80-100mm/s would be as fast. However the large object with coarse layers is very strong, one is a lookout from paw patrol that my son has dropped an amount of times that would have destroyed an object printed with a 0.4mm nozzle. So I don't see volcano as a way to go faster, rather a way to print big things with big nozzles which are stronger.
I don't see more than 1C variation out of the temp graph during the minute and a half test, I don't think it's a temp drop. I've experienced a temp drop trying to use a .8mm nozzle on a standard v6 block and saw that in the graphs.
strength is certainly extreme. I may just have to adjust my expectations to be in line with your view.
-
Have you tried different brands of ABS, some flow much better than others. 40mm^3/sec requires some flowy polymers.
yup. almost identical. small variation though. I was quoting 40mm/sec linearly, not volumetrically (which, now that I do the volumetric math, is worse). Sill, you'd think E3d would have told me about melt rate limits of that combo when I emailed in. They appeared just as puzzled.
-
Probably one of their first tier techs that was unfamiliar.
Note that you're doing 1000mm of sustained extrusion, not everyond does that as it isn't a fair test - rarely does a print not pause the extruder for corners/travel in a 1000mm extrusion.
Best case scenario, Volcano can do about 40mm^3/sec. That's super flowy polymer, extra high temperature, 1.2mm nozzle. More typical is around 10-20mm^3/sec. At 600mm/sec you're doing ~13 mm^3/sec and only seeing ~2% slip. I would be surprised if that 2400mm/min wasn't slipping. Where'd you see that quoted?
50-100mm (5-10%) of underextrusion is not shocking at high flowrates, E3D has documented this and presented it previously (a couple years back, can't find it now). Just asked Sanjay for the data, will post here once I get it.
-
…........................................
I'm still puzzled as to why I dont see any clues to this anywhere other than in the measuring or underperforming infill. On all of my other machines, when underextrusion has occurred for a chemical, thermal, or mechanical reason, there's evidence. Filament grinding, motor stalling, tubes popping out of PTC connectors, etc...Do all of your other machines use Titan extruders?
I'm currently doing some research into maximum melt rates using a Titan extruders and a Diamond hot end under printing conditions, and finding it very difficult to determine exactly when I've hit the limit. I was expecting to suddenly find clear evidence of drastic under extrusion, similar to the signs you mention, but in reality all I'm getting is a rough finish and ripples on the surface at a certain speed, which get worse as I increase the speed. It seems the Titan is doing too good a job of forcing the filament through "under duress".
-
…........................................
I'm still puzzled as to why I dont see any clues to this anywhere other than in the measuring or underperforming infill. On all of my other machines, when underextrusion has occurred for a chemical, thermal, or mechanical reason, there's evidence. Filament grinding, motor stalling, tubes popping out of PTC connectors, etc...Do all of your other machines use Titan extruders?
I'm currently doing some research into maximum melt rates using a Titan extruders and a Diamond hot end under printing conditions, and finding it very difficult to determine exactly when I've hit the limit. I was expecting to suddenly find clear evidence of drastic under extrusion, similar to the signs you mention, but in reality all I'm getting is a rough finish and ripples on the surface at a certain speed, which get worse as I increase the speed. It seems the Titan is doing too good a job of forcing the filament through "under duress".
nope, my other machines use a 3dator style extruder (fixed tension rather than spring loaded). My first indication that I hit any kind of limit was infill not able to run a complete line across the interior of the part (rectilinear infill).
-
nope, my other machines use a 3dator style extruder (fixed tension rather than spring loaded). My first indication that I hit any kind of limit was infill not able to run a complete line across the interior of the part (rectilinear infill).
Then I would suggest that the Titan extruder is the reason why you are not seeing under extrusion manifesting itself in the same way as your other printers. I'm having exactly the same issues with the tests I am doing on my Diamond hot end with Titan extruders. I'm trying to ascertain the maximum melt rate by pushing the print speed higher and higher. I'll be doing a write up on my blog with accompanying videos when I've finished so I don't want to say too much until I've finished all my testing. However, I can say that I am finding that there is no clear black and white distinction as to when I've reached the limit of how fast the filament can be melted. Rather it seems to be a gradual "thinning" of the filament bead as I continue to push the speed up. Eventually, there will be distinct ripples in the surface but I don't get much in the way of "clicking" and no sign at all of any grinding of the filament. These damned Titans just push and push. It's actually quite annoying for what I'm trying to do.
