external motor: speed be reduced more by adding more motors?
-
Good evening
Im testing a Clearpath Nema23 on a Ballscrew with 10mm lead.
Initially I set the Input resolution in the software to 6400/revolution.
And 640 steps/mm in the Duet Config file.Its running very slow.
To get it up at full speed (36000mm/min) I need to reduce the resolution to 2000/rev and X200steps/mm in duet.Thus the question: If I add 2 more of the same motors, and program them at 2000/rev & 200steps/mm. Will the duet 2 wifi have trouble handling it or how will that be?
Best regards
John Henrik Bergene -
Where did the 36000mm/min figure come from?
Why are the 2000/rev and 200steps/mm settings not acceptable?
Why would you add more motors?
Thanks.
Frederick
-
@jbergene said in external motor: speed be reduced more by adding more motors?:
Good evening
Im testing a Clearpath Nema23 on a Ballscrew with 10mm lead.
Initially I set the Input resolution in the software to 6400/revolution.
And 640 steps/mm in the Duet Config file.Its running very slow.
To get it up at full speed (36000mm/min) I need to reduce the resolution to 2000/rev and X200steps/mm in duet.Thus the question: If I add 2 more of the same motors, and program them at 2000/rev & 200steps/mm. Will the duet 2 wifi have trouble handling it or how will that be?
Best regards
John Henrik BergeneI assume you are are asking if you are hitting the CPU limit?
Probably.The fact that dropping the resolution enabled you to hit your target speed certainly seems to indicate that.
I also assume that you are asking if adding additional motors will increase the load on the CPU, meaning that a 3 axis move can't be stepped as fast as a single axis?
I don't know enough about the control loop to know, hopefully @dc42 will stop by and be able to answer that question. -
Using multiple X, Y or Z motors does not increase the CPU load in RRF.
-
@fcwilt 36000 =600mm/s movespeed.
I need to add more motors to add the Y axis is. 200step/mm is acceptable, but I'm just wondering where my limit potentially is.Edit: just adding: 1000steps/rev (100steps/mm) on a 10mm pitched screw, gives the same accuracy as a 1.8degree motor with 16 tooth 2GT belt.
10 microns.
So what I'm asking for here is obviously overkill (since a 10 microns 3d printer can produce perfect prints) but I'm just curious on the limits.@dc42 since this is only the X axis I'm currently testing, adding the Y axis will presumably limit the movespeed?
Does the Duet 3 have more room on the CPU?
And, I assume microstepping doesn't work as that is handled by the TMC2660 controller itself if I'm correct?Best regards
-
@jbergene moving X and Y together does have a lower maximum step rate than moving X only.
The limit is dependent on how wide the step pulses you set and what else the CPU is doing. so delta kinematics are slower (more CPU intensive) than linear kinematics.
In linear kinematics the Duet 2 can handle 170+Khz per axis on two axis for long moves so at 200 steps/mm thats 850mm/s with the default "fast" step timing
If you need to use longer step pulse widths than the TMC drivers need(which most external drivers do) then this will drop. e,g, at 2.5us pulse width i measured 85Khz for 3 axis moving at once.
You are right "micorstepping" is handled by the driver so if you set the internal or external driver to 16 microsteps per step then do your calculations based on a steps/mm of the microsteps (e.g 160 steps/mm step rate on belt driven 3d printers with 0.9 degree motors driven at 16 microsteps/step with a 20 tooth pulley and 2mm toothed belts).
-
-
@jbergene said in external motor: speed be reduced more by adding more motors?:
Does the Duet 3 have more room on the CPU?
Yes, a lot more. See the step rates at https://docs.google.com/spreadsheets/d/1AWA1wLbOaYzxzdQa5LRZvn9rgEk2BuluHy6-_OnD6FY.
-
@dc42 Is it possible to connect a SBC like RPI and use the RPI onboard computer instead of the 120mhz Arm CPU on the duet board?
to overcome CPU the CPU bottleneck.
Br
JHB