Unbeatable performance on a budget - can the E3D BigBox pull it off?
Unbeatable performance on a budget - can the E3D BigBox pull it off?
Man, that thing really IS big.
It seems like E3D has kept up their track record of building good kit, too. Now if only I could justify buying a third printer.
Imho,this printer is blame for e3d. They should do it better. But hotend is ok
In the video he says that the 8mm smooth rods are not the stiffest for use in a printer with this build envelope. I’m in the process of designing an ultimaker style printer with a 300mmX300mm build area, should I spec out 10mm rods instead of the 8mm and 6mm rods the ultimaker uses?
There is as always a good reason for not walking that path, and in this case it is the added weight. If you listened carefully to Thomas, he did say the results were very good.
So maybe thicker, heavier rods only seem better.The added advantage of stiffer rods would only come into play at high speeds, but with the added weight high speeds are not so easy. Not with Nema 17 motors anyway. Bigger is not better in this case.
@Bradford_Short i used 12mm on mine at a cubic foot print area
With a 500 gram x-carriage sitting atop two 8mm solid steel rods, you’re looking at about 30 microns of deflection. With 10mm rods, about 15 microns of deflection.
But this deflection is constant, and I doubt your bed is within 30 microns of flat anyway so really, 8mm is fine. Plus there’s the speed penalty as previously mentioned.
@Tim_Elmore you’re looking at static deflection, but the bigger issue is the dynamic part - once you start lugging that extruder around (accelerations are around 1G typically, plus the “jerk” constant) and factor in resonances, you will be left with visible ringing artifacts on corners of the print.
Has anyone pushed the speeds of a pro yet? I have a dual so it wouldn’t be a good test point
+Thomas Sanladerer I was looking only at the Z-component, which is basically static. As far as dynamics, a 1G acceleration would produce an equivalent deflection in the direction of the acceleration. Also: isn’t the default Marlin XY acceleration is 3m/s^2, or about 0.3G?.
30 microns seems pretty small here too - with typical extrusion widths around 500 microns and filament diameter variation at a few percent, the deformation from acceleration is about equal to the extrusion width variation.
Plus if you want to look at just dynamics, making everything thicker carries a weight penalty (rods + motors!).
@Tim_Elmore the jerk is what kills you. It’s an “instant” velocity change, i.e. “infinite” acceleration. In reality, the acceleration is finite, because the steppers apply finite torque. But the true acceleration is on the order of multiple G’s, depending on mass vs motor torques. Then the other factor to consider is that a suddenly-applied force creates an oscillating deflection that initially peaks at 2x the magnitude of the equivalent static or gradually-applied force. Between these two effects, ringing deflections are typically much larger than gravity sag.
Well at least the jerk is adjustable in firmware, but gravity isn’t!
I’m not sure I buy the multiple Gs either. A typical NEMA17 has around 40N-cm of torque. Lets say the pulley radius is 1cm, to make the math simple. That sounds like 40N of force (and 8Gs on a 500g extruder!) but… microstepping. Each microstep contributes 10% or less (depending on rotor position) of the holding torque at 1/16 stepping, so wouldn’t that first step be more like 4N of force (1G) neglecting friction? You’d have to do many microsteps to get to full motor torque, and by then the carriage better be moving.
Is there definitive insight on the contribution from belt rigidity? People say “near zero”, but perhaps 30 microns would also be deemed “near zero”.
I have a ICM20608-G, maybe I should strap it to the BigBox end effector and get some numbers.
As for bigger rods being heavier, it’s true, but weight increases with the square of radius while stiffness increases with the cube of radius. So you do net gains from bigger rods… Just not as much as you would hope. When you do all the math, to achieve constant deflections you need to make rod diameter increase roughly LINEARLY with span length. A 25:1 ratio for L:D is the usual rule of thumb for precision motion.
@Tim_Elmore The key thing to remember is that steppers only generate torque when they have position error (or more specifically, phase error between rotor position and coil energization position). To generate max rated torque, the stepper has to be a full step off position. So in a typical cartesian drivetrain, that’s ~0.2mm of error before you actually get to peak torque. SO, if we imagine the motor is spinning along at 20mm/s carriage speed, and you suddenly command it to dead-stop, the rotor+drivetrain mass will try to keep going and will only start building braking torque after it has overshot the target position. This is analogous to dropping a weight onto a spring. Once the rotor overshoots the energized coils, torque rises rapidly, until the work done “compressing the spring” so to speak equals the change in kinetic energy from 20mm/s to 0mm/s. (Then it rebounds, oscillates and damping absorbs the energy.)
That highest “compression” position error will be less than one full step, or else the stepper will skip steps, but it can be quite a bit of error and quite a bit of peak torque. Your microstepping level doesn’t really matter in this “dead stop” scenario – just the stepper’s torque/error function and the drivetrain inertia.
What gets really complicated is figuring out how the rod flex and stepper overshoot error affect each other. I have yet to see anybody make a really realistic estimation of how much ringing comes from rod flex and how much comes from stepper restoring torque effects. It would depend on some pretty holistic drivetrain design decisions – eg bigger rods mean more stepper error but less rod flex.
Then there’s belts too. Between the extruder carriage mass, rod mass, rod flex, belt stretch, rotor inertia, and stepper restoring torque, you have more or less three elastically-coupled damped oscillator systems. Which is nutty. It’s simply not realistic to simulate without getting into frequency-domain transfer function type stuff.
@Ryan_Carlyle Re: steppers require phase lag to produce torque - Yes, of course. The cross product of two parallel vectors is zero. Not much torque to be had!
But this is different from rod flex, which was the original topic. It just means steppers are an ideal motor with a spring between the motor and the load, and that springiness is a contributing source of the ringing that manifests as ripples in the print.
Actually, you can distinguish the factors to some degree. If you need all available torque to decelerate the load, then that’s 40N of force and 0.2mm of error from the stepper alone. Then you’ve got the deceleration induced deflection, which would be 40N/0.5kg = 8G. That’s 0.24mm of deflection, or roughly equal to the contribution from the stepper motor. This assumes all force goes into acceleration (no friction) and the rods are point loaded and point supported. Real loading and support are more distributed, so the 0.24mm is a bit of a worst case scenario while the 0.2mm from the stepper motor is a bit of a best case scenario since it ignores belt stretch.
If you make the rods larger then the rod deflection contribution gets smaller, but the stepper contribution gets bigger (for a particular jerk setting, and until you jump to a larger motor when then has a weight penalty). As you said previously, you see some benefit but not as much as one might hope.
For my uses, the 8mm rods are fine for the BigBox. 10mm would be better, sure - but not drastically so and hey - the BigBox is open source so one is free to modify as they see fit!
The Ulitmaker machines from v2 on are made from aluminum composite panel, often times referred to by the brand name Dibond or generically as sandwich panel. It’s two thin sheets of AL (sometimes coated with a polymer or film) and a non aluminum core.
So… how long untill someone designs printed parts for 10mm Y rods on the BigBox? Any bets?
I did not realize how big that printer was, wow!
Ultimaker 2 has left and right side panels made from acrylic, top, bottom front and rear are aluminum sandwich composite.