Comments sought on new build: A CNC Router for RC Gliders and Planes

[Jonathan]

I'll have to dig out the drawing of the gantry (estimated) to work out the mass, ballscrews are 1610's length is 1300mm, feedrate I've not calculated (or did but can't find the calculations on the spread sheet) but it will be cuttng balsa wood, plastic and aluminium.

Just to be clear that's two ballscrews at 1300mm long, not one? By feedrate I meant what feedrate do you want, not expect, so for cutting those materials I'd say around 8-10m/min would be suitable. The mass doesn't have to be that accurate - within 10% is plenty good enough.

[njhussey]

Just to be clear that's two ballscrews at 1300mm long, not one? By feedrate I meant what feedrate do you want, not expect, so for cutting those materials I'd say around 8-10m/min would be suitable. The mass doesn't have to be that accurate - within 10% is plenty good enough.

Yes twin ballscrews 1300mm long! Going to drive it with one Nema 34 8.7Nm stepper (or 4.5 if I can get away with it) and a 2M2280N High voltage digital microstepping driver and a long belt 55kg weight springs to mind but I'm at work and the details are on my laptop at home.

[Jonathan]

Based on:

• 2 of RM1610-1300mm
• 55kg actuated mass

Assume:

• 10m/min feedrate
• 2m/s^2 acceleration
• 20T pulleys for 1:1 ratio, 30:N pulleys for 30:N ratio.
• Cutting force of 25N
• Not pre-loading ballscrews

Variables to select:

• Motor size
• Drive ratio

For the common 3Nm Nema 24 motors the corner speed is about 1000rpm, so lets start with 1:1 ratio and see if the torque calculated is low enough to use two of those motors. Using the script (posted here, and attached), the required torque from each motor is 0.47Nm. Up to their corner speed stepper motors output about 2/3rds of their holding torque, so 2Nm up to 1000rpm for the 3Nm motor, so since you only need 0.47Nm this system will definitely work up to 10m/min (1000rpm*10mm) plus a bit more since the torque will only drop below 0.47Nm at a fair bit more than 1000rpm, so you have a good factor of safety. You could even use 30:20T ratio and get 15m/min feedrate since the torque for these parameters is 0.78Nm, though you would need to used fixed-fixed bearings on the ballscrews to get the required critical speed.

One interesting thing to note is that if you use 30T pulleys instead of 20T, the torque required increases from 0.47Nm to 0.70Nm, so don't use bigger pulleys than necessary.

Lets try one motor driving both ballscrews:

Assume:

• 10m/min feedrate
• 2m/s^2 acceleration
• 20T pulleys for 1:1 ratio, 30:N pulleys for 30:N ratio.
  
  • One 20T pulley on each ballscrew
  • Two belts (length roughly 1/2 distance between ballscrews) with two 20T pulleys attached to motor.
  
  
• Cutting force of 25N
• Not pre-loading ballscrews

Since the system is symmetrical, we can model it as a single ballscrew with the length doubled and twice the pulley inertia. The calculated torque is 0.75Nm (or 1.09Nm with 30T pulleys). That's still within the ratings of the 3Nm motor... so no need for a Nema 34 motor (though it would clearly work, just be more expensive). If you want higher resolution, you could use 16:20 ratio (16T on motor) and the torque required is 0.55Nm.

My script also now calculates the ballscrew (nut, shaft & bearing) stiffness and the torsional stiffness of the belts. For example if we assume 100mm center distance, then for two motors the linear error due to belt compliance is 5.6um, compared to the error due to ballscrew stiffness of 0.7um. That's assuming worst case scenario - so machine operating with maximum acceleration and cutting force. If you use one motor, say the belts have about 700mm center distance then the belt stiffness goes down by a factor of 7, so it's now about the same as the torsional stiffness of the ballscrew. That sounds like a big difference, but you have to consider the stiffness of the whole system as clearly if the gantry is not terribly strong that will make a bigger difference than the belt stiffness (or lack of).

I've attached the scripts with your parameters entered for X - change the file extension from .txt to .m. The script runs in MATLAB, or GNU-octave which is free/open source and can be downloaded here. I encourage you to experiment with the numbers to get a feel for what parameters affect the torque requirements the most. You could put your numbers in for X and Z...

servo_motor_sizing_hussey_1.txtservo_motor_sizing_hussey_single_motor.txt

[JAZZCNC]

Since the system is symmetrical, we can model it as a single ballscrew with the length doubled and twice the pulley inertia. The calculated torque is 0.75Nm (or 1.09Nm with 30T pulleys). That's still within the ratings of the 3Nm motor... so no need for a Nema 34 motor (though it would clearly work, just be more expensive). If you want higher resolution, you could use 16:20 ratio (16T on motor) and the torque required is 0.55Nm.

No sorry lovely calculations and all that but I can tell you from experience that 1 x 3Nm Nema 23 motor connected to 2 x screws with timing belts won't handle what will become 60Kg+ gantry and be reliable.!
Yes it will move it no problem and work but when the cutting gets hard or the speed and direction changes are fast it will struggle and the chances of lost steps makes it unreliable IME.
Either twin 23's or Single 34 is the way to go for this size/length and weight IMO.

[njhussey]

I calculated that I only needed one Nema 24 3Nm stepper for the twin X axis ballscrews but on thinking about it and listening to Dean I'll be going with a single Nema 34 on the X axis.....just to err on the safe side...after all it's not my money now so I'll be getting a 240V driver to go with the Nema 34 as well...

[JAZZCNC]

I calculated that I only needed one Nema 24 3Nm stepper for the twin X axis ballscrews but on thinking about it and listening to Dean I'll be going with a single Nema 34 on the X axis.....just to err on the safe side...after all it's not my money now so I'll be getting a 240V driver to go with the Nema 34 as well...

Ok well feel I've got to comment here about theses calculations or calculators. While I'm not saying the calculations are wrong what I am saying is that they are miss leading and not giving the full picture so sending people wrong. I've helped several people who have gone off them and been dissapointed with performance.!!

They don't and can't take into account all the factors that make up a working machine. They don't account for miss alignment or poor materials and how machine is designed and built or the affect this as on friction etc.
They don't account for the affects of resonance. They don't account for Cheap drives running on PC with poor pulses from a ropey parallel port.
They don't account for the DIY affect and the numerous things that can affect how machine performs.

All these things come into play and have a big affect on the overall performance and the calculators can never know these variables so IMO they are only to be used as a very rough guide.

This particular case is a prime example because while the Calculators says 1 x 3Nm Nema23 turning 2 x 1300mm screws connected with belts should all be hunkydory I can tell you with confidence you won't get anything like the performance that as been suggested. Yes it will move them but at much slower feed rate and acceleration than the calculator suggests to give a stable working machine.
So Neil believe me your Not Erring on the side of caution your doing what's is needed to get the level of performance your looking for.!!

Neale:

The single or Twin motor argumant falls down to ONE real differance.! . . How stable and accurate do you want your machine.?
If you wan't very stable with virtually no chance of screws losing sync even with cheap electronics then you need single motor/belts.

If you want to use twin screws then while they do work well they only do so provided certain things are correct.?
They are much more sensitive to motor tuning and need careful attention not to push too hard.
IME They work best and get best performance from them with Modern digital drives and External motion control cards which can provide nice clean pulses compared to ropey parallel port.
IMO it's crucial they have some form of stall detect or Error signal on the drives so if one motor stalls or drive fails the system shuts down. I won't build a twin screw machine without this option and Those that have built using twins motors and had this happen at speed will tell you how scary the big potential for damage.

I build more machines using Twin motors than single motors because it's easier but will ONLY do so provided I can build like the above. I Only build using Motion control cards and Digital drives and still leave a comfortable margin on tuning for best reliabilty and safety.
I never have any issues with twin screws building this way but it comes with a cost. Single motor and belts while more messy and involved is cheaper and more forgiving with lesser electronics.
I never truely trust twin screws not to stall or twist up the machine in some way if pushing hard for exteneded periods and I'm always mindfull while using and setting up. Where has I have complete trust in single motor setup and it never enters my head in use no matter how hard I push, no matter how long I push it for.!!

Both work.!!

[EddyCurrent]

M = Misalignment
Q = Materials quality
D = Design and build
F = Friction etc.
R = Effects of resonance.
C = Cheap drives running on PC
P = Ropey parallel port.
Y = DIY effect
O = other miscelaneous effects

For each of the above assign a value between 0 and 1; 1 = 'perfect' or negligible, 0 = 'total shite'

JazzFactor = M*Q*D*F*R*C*P*Y*O

(Note: if any value=0 then JazzFactor=0 hence idea=total shite so don't even bother.


Actual calculation = theoretical calculation * JazzFactor

Example:

theoretical torque = 0.75Nm

JazzFactor = (0.9 * 0.95 * 0.9 * 0.99 * 0.9 * 0.95 * 0.9 * 0.85 * 0.85)

actual torque = 0.75 * 0.42

actual torque = 0.32 Nm