CoreXY performance (max print speed) comparison
-
General question,
Has anyone done a comparison with say a Voron 2 CoreXY design between the various different Duet board generations and a typical klipper setup to show the benefits or lack there of for print speed, print quality, input shaping, pressure advance, etc.?
I know it would be a ton of work to swap out boards, rewire and reprogram everything, but I think it would be a really interesting to see what the max performance would be with just swapping out the boards. Same steppers, hotend, motion system, just the board changing out. I would think it would make an excellent youtube video for any of the duet content creators out there!
For instance, I wonder if 300Mhz would make a difference at really high speeds with input shaping turned on or if we would find that 120mhz is fine to handle whatever a typical coreXY could throw at it .
Another one that would be cool to compare would be closed loop steppers vs. stock setup vs like an AWD setup for all the same metrics. I know personally on my rig that I'm losing steps printing at the speeds my rapido UHF hotend can handle with a 0.4mm nozzle. Closed loop would fix that problem and maybe allow me to push the system closer to it's max throughput. From a quick cost analysis, there's a huge premium to upgrade everything to handle a closed loop setup but maybe a comparison video would show that it's totally worth it for the extra performance and reliability that you can get from the system.
I'm curious what the community thinks of this idea or if it's already been done.
-
After thinking about it a bit more, upgrading to AWD is really about the same cost as adding 2x 1HCL's and 2x steppers with encoders when comparing to 4x LDO nema 14's or nema 17's, the plates, and an 3HC.
This is probably the better comparison to make. Any thoughts from those running closed loop voron's or have seen both motion systems that can speak to the pro's and con's? I don't think I can return the new motors I just got in, but it would be interesting to keep a closed loop upgrade in mind for the future.
-
@michaelr123 If you are missing steps on the axis motors at print speeds (which is something I'm surprised to hear) then changing to closed loop control will not in itself allow you to print faster. The reason for missing steps at high speeds is that the motor is operating at a point where the torque drops off too much to accelerate the mass at the desired rate. The most common cause is back EMF due to rotation. This is a function of stepper motors in general so fitting a shaft encoder will not change the behaviour of the motor. It will detect the missed steps but it won't prevent them if you drive the motor too hard. What will help is either using motors with inherently more torque, or choosing ones with lower inductance or driving them with a high voltage (or any combination thereof) but the same spec motors fitted with a shaft encoder won't help to deliver higher speeds.
Server motors do not lose torque in the same way that stepper motors do so that might be another option.
The maximum step pulse frequency is largely irrelevant unless you run at (say) 256x micro-stepping. As an example take an axis speed of (say) 500mm/sec with 80 steps per mm at 16x micro stepping then you have a pulse frequency of only 40kHz.
Regarding CoreXY vs other kinematics, there is an inherent advantage in that for anything other than 45 degree moves, both motors contribute to motion. There is also a mechanical advantage in that with 45 degree moves, the belt arrangement is such that it gives a leverage effect increasing the motor torque by sqrt2 (roughly 1.4) although this means the motor runs at higher speed so it might reach the point on the torque curve where it start to drop.
-
Right now, I'm running a duet 2 with 2 extra external drivers patched in and I was trying to increase the current to my XY motors via software... but I just realized that I can't use the UART function to these steppers... because they're external
I'm pretty sure that explains why I was still losing steps as I wasn't adding any additional current. Swapping in a 6HC+3HC should fix that. It also seemed like updating to 3.5 RC2 made some changes to how the system performed, but I should be able to overcome that with more current.
As for the closed loop steppers, I figured you would be able to push higher peak powers while also staying cooler on average. Is that not how closed loop steppers work? This would in theory allow for higher power limits, and therefore better accelerations.
-
@michaelr123 this is me running a VzBoT 235 on RRF. 2100mm/s at 100k acceleration
https://photos.app.goo.gl/K93Cm8AZEX9y5uWN6Owns various duet boards and is the main wiki maintainer for the Teamgloomy LPC/STM32 port of RRF. Assume I'm running whatever the latest beta/stable build is
-
Now that's fast!! lol.
I don't know as much about VZbots, but looks like fairly standard nema 17's and metal plates? is this closed loop?
I'm working on a archetype - breakneck tool head with a rapido UHF. I'm hoping to be printing around 50mm^3/s with a 0.4mm nozzle! So far I'm all setup to go plaid with nema 14's and upgrade to aluminum plates on my 300mm voron 2.4
-
@michaelr123 it's a 4WD setup, using 2 X and 2 Y NEMA 17s. I was over juicing the motors somewhat at 4A and they're ran on 48v.
It's not closed loop. I would have to go for external closed loop drivers and the duet 1HCL caps out at about 800mm/s in closed loop mode.
I can also run the same printer and electronics on klipper -
@jay_s_uk, that is absolutely nuts, LOL! Can you actually print at that speed or is this just about non-printing moves? If printing speed is different, what is your best printing speed?
-
@jens55 i haven't attempted printing at that speed but its on my list.
i'm running a goliath hotend with a 0.8mm CHT nozzle so hopefully flow won't be an issue -
@jay_s_uk, a lot of filaments give a range of speed that the filament is made for. Absent of some high speed filament (never actually seen speed specs on these), most filament recommends speeds below 100 mm/sec. I can see pushing the speed to maybe 300 mm/sec but I can't see how 2100 mm/sec could possibly work.
I am looking forward to hear about your speed tests to see what is actually achievable. -
@deckingman said in CoreXY performance (max print speed) comparison:
Regarding CoreXY vs other kinematics, there is an inherent advantage in that for anything other than 45 degree moves, both motors contribute to motion.
Isn't that similar to standard systems? Slightly reworded: Regarding other kinematics vs core xy, there is an inherent advantage in that for anything other than axis parallel moves, both motors contribute to motion.
That said a big one has to be reducing moving mass - aside from Ultimaker style systems.
Running 3 P3Steel with Duet 2. Duet 3 on the shelf looking for a suitable machine. One first generation Duet in a Logo/Turtle style robot!
-
@DocTrucker said in CoreXY performance (max print speed) comparison:
@deckingman said in CoreXY performance (max print speed) comparison:
Regarding CoreXY vs other kinematics, there is an inherent advantage in that for anything other than 45 degree moves, both motors contribute to motion.
Isn't that similar to standard systems? Slightly reworded: Regarding other kinematics vs core xy, there is an inherent advantage in that for anything other than axis parallel moves, both motors contribute to motion.
That said a big one has to be reducing moving mass - aside from Ultimaker style systems.
Slightly off topic, but there is some irony with the CoreXY motion system, as I’ve seen several recommendations in the Voron community to print squareish parts at 45deg on the build plate to aid in eliminating surface artifacts. Lose out on some mechanical advantage, but reduce motor contributions to surface artifacts on 45deg moves.
-
@DocTrucker said in CoreXY performance (max print speed) comparison:
............. Isn't that similar to standard systems? Slightly reworded: Regarding other kinematics vs core xy, there is an inherent advantage in that for anything other than axis parallel moves, both motors contribute to motion.............
Not exactly. For pure X or pure Y on a CoreXY both motors contribute equally whereas on a say a Cartesian only one motor contributes. As the angle between axes increases, one motor will contribute less on a CoreXY but the first motor will contribute the same (everything else being equal) until we reach 45 degrees where only one motor contributes (but with the mechanical advantage of sqrt2 due to the gearing effect). On a Cartesian as the angle increases, one motor will contribute less as the second motor contributes more and without the mechanical advantage due to gearing.
-
As an update, I checked my driver's and found one of my motors was only running on like .5A...
Cranked them both up to 1.5A and everything got so much better, quality, resonance, max speeds. I'm not sure if I somehow bumped the potentiometer at one point or what happened there.
Unfortunately my 6HC board is v1.01, so no 48v for me. Maybe I'll swap it out it out for a 1.02 some day, but I've got plenty on my hands to figure out Can bus for the first time and swap everything over so I'll save that upgrade for another day. Now that I've got my duet 2 system running so well I'm a little sad to tear it apart, but it'll be fun to see how much more performance I can get with a duet 3 system.
Anyway, I still need to buy some cht volcano nozzles for my rapido, with standard brass volcano nozzles from E3d with abs at 250C I've found my max throughput to be around 45mm3/s for general printing to avoid under extrusion. .8mm nozzles really don't need to move very fast to hit that flow rate. .6mm nozzles move faster at that rate, but nothing too crazy for a stock voron. Getting a .4mm nozzle up to that rate is where I'm guessing AWD comes into play.
I know rapido uhf's claim they can hit 75mm3/s, not sure how possible that really is, but I would imagine with cht nozzles, cranking temps, and switching to high quality extruder like an orbiter will make improvements.
-
@michaelr123 said in CoreXY performance (max print speed) comparison:
............................. I know rapido uhf's claim they can hit 75mm3/s, not sure how possible that really is, ..............................
A lot depends on the filament. Something like PLA with a low viscosity will give a much higher volumetric flow rate than PET-G for example. When I was experimenting with 6 input, single nozzle multi-material hot ends, I tried some high flow rate prints using all 6 extruders in mixing mode, just out of curiosity. With a 1.5mm nozzle, 1.65mm layer width, 1mm layer height, I got and average flow rate for the entire print of 73mm^3/sec https://www.youtube.com/watch?v=gc8AciHjf4I but the best I could get with PET-G was around 44mm^3/sec using the same hot end, nozzle, layer width and layer height combination.
