What speeds are you printing at?
Roaming the internet I found some infos about printing speed, so i.e. a JHead with 0.4mm nozzle will do about 50mm/s with ABS, PLA will be faster, E3D Lite will do 50mm/s, E3D V6 200mm/s …
30-300mm/s depending on the object and what part of the object.
if you print faster than 120mm/s, could you give some details about the used printer/extruder/hotend? My china E3D clone seems to max out at about 120mm/s, usable at about 90mm/s.
Ultimaker II extended
Nice numbers 
Unfort. I have to admitt that 95% prints that i see on internet are not accetable by my standards
@Kristijan_kolak So maybe FDM is just not for you? Many people seeing my printed things don’t like the rough finish of 3d printed objects.
No,i just like making things right
@Kristijan_kolak depends on the definition of ‘right’. For me that varies, depending on how and where the printed object is used.
volcano nozzle currently fitted, super slow but super strong
Feedrate in terms of speed should only be used when discussing drivetrain design. When talking about extruder / hot end limits, you need to think in terms of FLOW RATE – filament volume per second.
What limits print speeds is the mm^3/sec of plastic being pushed through the nozzle, not just the linear speed the nozzle is moving.
To a good approximation (plus or minus some slicer quirks and calibration) you can calculate volumetric flow rate by simply multiplying [extrusion_width * layer_height * feedrate]. So printing 0.3mm layer heights will hit the extruder flow limit at a MUCH lower feedrate than 0.1mm layer height. Likewise, different extrusion widths will lead to different top feedrates. This is the main reason why people report such a wide range of print speeds. Speed simply isn’t the right metric to compare performance!
Most all-metal hot ends and regular extruders can do ~8 mm^3/sec with PLA or ~15 mm^3/sec with ABS through a 0.4mm nozzle. That’s based on how fast the filament can be heated to a low enough viscosity to be squeezed through the tiny nozzle orifice. Bigger nozzles have higher flow rate limits because viscosity is less critical with a bigger orifice. A 4x2mm PTFE liner cuts those limits roughly in half or less because it’s an insulator. A longer melt zone also increases speed, which is why a Volcano can do around 30 mm^3/sec with PLA.
Thanks Ryan, great explanation!
Yes, and we should talk about ACTUAL material flow rate per second. Not configured. We’re having a ton of acceleration stuff happening in the firmware, meaning that your actual feed/flow rates are different then what you are configuring in software.
So far, I haven’t found the time to put an actual flow rate display in the UM2 firmware. But I hope to find some time to do this soon.
But as for configured flow rate (which is line_widthlayer_heightspeed), the Ultimaker Original seems to max out between 10-12mm^3/second and the UM2 between 7-10mm^3/second. (According to our data at Ultimaker, your mileage may vary, and depends on set temperature as well)
@Daid_Braam Great point.
The best/easiest way to test your extruder’s flow capacity in my opinion is to print a big box with LINEAR INFILL that places the same strand orientation(s) on EVERY LAYER. For example, “grid” fill in Simplify3D. That way, you know:
- The printer has enough gcode segment length to actually accelerate up to the target feedrate for a while on each infill pass.
- The nozzle back-pressure and effective extrusion width due to layer height is constant.
If this test print infill has broken strands in the middle, you’re getting filament starvation, and thus have exceeded the flow rate limit of your extruder.
The other thing I want to emphasize is that different filament materials can have VERY different max flow rates using the same hardware.
- The drive gear/hob/knurl gets a different quality of grip against different filament hardnesses. For example, PLA is hard and brittle so you get a shallower bite than you do with ABS. Some nylons and a lot of TPEs are so soft they’re difficult to grip, which means less force can be applied.
- Different filaments have different heat capacities (which roughly determines how much heat has to enter the filament to melt it) as well as different thermal conductivities (which determines how fast heat flows into the filament). For example, PLA has low thermal conductivity and a high heat capacity so it needs more time in the hot end to fully melt (thus lowering top print speed).
- Different filaments have different viscosity-vs-temp and viscosity-vs-strain curves, which controls nozzle back-pressure and thus how much filament pushing force is required. For example, nylon and PLA both tend to strain-crystallize as they neck down from the melt pool to the nozzle, so the effective throat viscosity is higher than you see with a more amorphous filament like ABS.
It’s all quite complicated when you really dig into it.
@Ryan_Carlyle Just did some quick math on some numbers. With an acceleration of 1500, a distance of 50mm, and a set movement speed of 150mm/s. You’re off about 25% on the actual vs configured speed. Doubling the acceleration lowers the error to 13%. And going to 100mm/s instead of the 150mm/s lowers the error to 11%. So at lower speeds, the estimates are quite ok. But going over 100mm/s, all math is off right now.
At Ultimaker we’ve been using testing cylinders to test the actual flow. While you do have some acceleration effects, you are just going in circles, thus the acceleration and pressure are constant. I think this will give less acceleration related effects then cubes with infill. And are quicker to print.
It’s not a perfect test, but we’ve been using it a lot to compare different machines and setups.
It’s also good to remember that each material has a sort of “processing window”. Depending on the amount of flow you want, the machine you have, and the material you have, you need to set your temperature at a certain amount. Using a single temperature for a material, as we are doing now most often, is usually sub-optimal.
@Daid_Braam All valid points. The cylinder test is good, although how the firmware reacts to faceted arc paths can vary. Does Marlin maintains speed through the full arc? I know Sailfish always slows down at the four “axis direction reversal” points in circles (0,90,180,270 for cartesian or 45,135,225,315 for CoreXY) because GRBL-based planners have so much trouble identifying faceted arcs. So that messes up the cylinder-test flow rates. The longest continuous line of extrusion you can do without any acceleration slowdowns is a diagonal across the build plate.
The other thing to ponder (which might not matter at all, I’m not sure) is that polymer flow is rate-history-dependent, so alternating fast and slow periods might be a more realistic test for real printing than a constant-rate cylinder. Rate changes and retractions cause much more complex polymer shear history through the transition zone.
I’m at 120mm/s on my RigidBot.
As explained above full speed is rarely really achieved. Based on print size and current wait time I’d estimate 400mm/s would remove any care I had but my setup simply won’t cope.
I will be going BondTech soon which should mean a ton of tuning but epic consistency so can start the speed march. 400 mm/s is way out of reach xD
Wow, thats so much useful information here! I asked for print speed because I thought its much more imaginable, and its usually the slicer setting you set or change. But your info is very welcome! I usually use slic3r and since a few versions there’s a max speed setting below the normal settings, so maybe its a good idea to find out the max throughput volume of my hotend/nozzle/extruder combo and calculate a max speed from that?
we produce kits at 70mm/s print and 150 for moves. quality /speed is ok. We just made speed tests with aluminuim smartrapcore at 200 and 300… but the result is really not good at all , the flow seem hard to follow corners and moves.there is certainly a good study to make there but it’s highly technical i guess ? plastic flow…etc…
Thanks for your responses! It seems most people are printing at max 60mm/s, so that was what I expected. Assuming different printers are used, I would guess building your own printer which will print quite ok at that speed could considered a success