why acceleration?

Hi all. I am having a bit of trouble with my Mendel reprap. I have been running at print speeds around 75mm/s. I notice that when I print some parts that have rapid short zig-zag fill patterns, I seem to loose steps on my Y axis because the part becomes shifted in the Y after the current layer. My belts are tight and my motors have enough current without overheating. I have to back off down to about 50mm/s for this problem to go away. I notice that there are videos of machines on youtube which can achieve speeds in excess of 150mm/s and they seem to use acceleration. Would enabling acceleration increase my maximum build speed? What are the disadvantages of acceleration?

Thanks

Acceleration control means that desired speed is achieved through a progressive increase/decrease in the steps pace. Conversely, not having acceleration means that stepper is driven at a constant pace which is the desired speed. As there is some mass involved here, you need a lot of torque on your motor to be able to "instantaneously" accelerate to target speed your axis in the first step. That is why a high speed is not working well without acceleration.

However, if you activate acceleration on your firmware the acceleration required to reach the desired speed is spread across multiple steps, which does not require that much torque from the motor, giving you a smoother motion and less chances for your motors to skip steps.

misan

The problems are mass, inertia and friction.

The elements of the Cartesian bot have all 3;

1. When starting/speeding up, the mass of the system need to be accelerated to the desired speed and at the same time, the friction overcome, to do this in one step requires enough force, or in the case of the stepper motor, torque, plus the interface between the stepper motor (gearing, belt connections etc) also needs to be able to handle this force. Acceleration allows you to achieve higher operating speeds without having to increase the force or reduce mass and/or friction.

2. When stopping/slowing down, it's the reverse the inertia of the system needs to be reduced to that of the desired speed and the braking force is also equivalent to the force capability of the system.

Think of it like a car where you need to accelerate and brake, a more powerful car (same mass etc) will accelerate faster (brakes obviously separate), but these things never happen instantaneously.

Your motor skips steps as it's trying to step the motor again before it's completed the previous step, because the real world acceleration is lower than predicted acceleration in the firmware.

Thank you. I'll have a go at using acceleration.

I just have a feeling that it may be something else other than this thats causing the issue.

Please report back how this goes I am keen to see how acceleration affects the top speed of a machine.

I have always wondered how necessary acceleration really is, particularly with very short moves on stepper motors with relatively light loading, a belt drive which removes some of the shock and positive gear ratios.

With default settings in the firmware say for a 10mm move how long is the acceleration and deceleration phase?

I know this depends on speed and acceleration settings, but just a rough guide would be usefull. I was wondering if there was a way to approximate acceleration, maybe just a couple of speed steps to full speed rather than calculating it all out.

I am not sure I should mention the microstepping debate here!

Do they cross over so that maximum speed is never reached and is this why the machine runs better?

Is the smoothness of the machine noticable and what speeds do you all run at.

Acceleration required will vary from machine to machine.

Variations in design, motors, machine setup, parts used, driver type, with current setting and operating voltage etc all impact the performance of the system.

Regardless of the machine, you can always go faster with acceleration than you can starting instantly.

---

[www.hydraraptor.blogspot.com]

By "always go faster" do you mean top speed or time it takes to complete a line segment? I'm pretty disappointed in actual build time reductions for large increases in potential top speed.

There is a maximum step rate that you can start at without stalling the motor. That is the pull in rate and it is determined by the pull in torque of the motor and the mass and friction of the system.

Without acceleration the whole line is at the speed. Once the motor is moving though you can increase the step frequency and go faster until you exceed the pull out torque of the motor, where again it will stall. You can give any step sequence that doesn't exceed the pull in rate and the motor will follow (ignoring resonance effects). When you operate above the pull in rate the motor is said to be in slew mode, which means it can't stop instantly without losing steps, it has to decelerate. It no longer follows arbitrary step sequences.

With acceleration you should complete line segments faster. However with constant acceleration you do have to limit the acceleration when accelerating to higher speeds, because torque drops with speed, so it is not optimal for very high speeds. S shaped acceleration is better in the respect. At lower speeds you have more torque so can accelerate rapidly but as you go faster torque drops so you need to accelerate slower to avoid stalling.

I think most of the speeds quoted nowadays are bogus because they are never reached due to the acceleration and the maximum move distance.

To make a system move from A to B twice as fast with constant acceleration needs 16 times more power!

---

[www.hydraraptor.blogspot.com]

To answer hazuki's original question

Quote
Would enabling acceleration increase my maximum build speed?

And put some numbers to nophead statement

Quote
Regardless of the machine, you can always go faster with acceleration than you can starting instantly.

The blue Mendel in the video is running without acceleration at 72mm/s infill as it won't run faster without beginning to miss steps. It's counterpart in the top right of the video is running under acceleration at 100mm/s infill as the extruder won't melt the filament consistently any faster. On paper the Mendel running acceleration looks to be ~39% faster. However, comparing actual print times on a certain tray of parts, the blue Mendel completed the print in 80 minutes and it's counterpart completed in 61, so the gain using these particular acceleration settings is more like 24%. So speed ratings for acceleration are never actually as good as they sound. Given long acceleration curves the 100mm/s machine could very well print slower than the 72mm/s machine, however when configured within optimal range as nophead said, the machine will always run faster and smoother with acceleration than without.



The white and green machines in the video both run at 120mm/s infill under acceleration, which seems to be the upper limit for consistent prints under our current acceleration configuration.

To respond to hasuki's second question

Quote
What are the disadvantages of acceleration?

Poor configuration will result in slower printing......... That's all I've got. Our machines run smoother and our belts last longer. Our prints complete faster and turnout better.

---

[www.eckertech.com]
[blog.eckertech.com]
[forum.eckertech.com]

If you are limited by the volume of plastic you can extrude, at least you can decrease your layer height, increase you speed, and print better quality parts in the same time.

Thanks again for the replies. So I installed new firmware (Sprinter) and had a go at printing. I notice acceleration/deceleration when I move the axes manually or when the machine is moving during a print but not extruding. I don't think I've noticed acceleration when the machine is actually printing. I'm not sure if this is because the acceleration and deceleration phases of each trajectory are just so fast that I can't notice them. To restate, If I tell X to move 100mm, I can not only see but also hear the steppers wind up and then wind down at the beginning and the end of the trajectory. But when I have a part that say, has a 100mm perimeter, I don't hear or see any windup and it looks like a simple constant velocity trajectory.

long max_acceleration_units_per_sq_second[] = {1000,1000,50,10000}; // X, Y, Z and E max acceleration in mm/s^2 for printing moves or retracts
long max_travel_acceleration_units_per_sq_second[] = {500,500,50,500}; // X, Y, Z max acceleration in mm/s^2 for travel moves

I'm guessing that the firmware uses a much larger acceleration for when the machine is actually putting out plastic. I just probably don't notice it.

Anyway, after messing around a bit, I do notice that everything runs smoother now, and I'm quite happy with the performance of my machine!

BTW I do love that video, eckertech.