UM1 resolution

[illuminarti 18]

Nobody ever prints at 300mm/s on a UM1. And you've never heard anyone say that they do. Well, you might have heard me say that I tried, I guess. But it doesn't work. And that's mostly not a function of the electronics or the firmware, but because of how plastic behaves, and how the mechanical components of the printer behave when the head has that much kinetic energy.

Good quality prints are typically done at more in the range of 30 - 50mm/s - maybe even less. And for smaller prints you have to go even slower than the target speed in some parts, to allow the plastic time to cool.

A fast print destined for some mechanical purpose where the finish quality is less important might be printed at 100 - 150mm/s. But as noted previously, that has implications for layer height, because the extruder system cannot physically extrude more than about 10mm³ of plastic per second.

Again, different parts of the print require different segment speeds, which is why the feedrate parameter is important. And so I was addressing your difficulty understanding 'how Fxxxx feedrate in the gcode would affect anything I'm doing.'

 

[jhinkle 1]

Thank you -- that was a great reply providing exactly the information I was looking for.

Joe

 

[illuminarti 18]

That reply contained exactly the same information that was in my previous reply, and in several that I made earlier in this thread and the other one.

My bigger point is that you seem to be assuming that everything about 3D printing comes down to electronics and firmware. And it doesn't; frankly the electronics and the firmware are already in pretty good shape. The challenges that remain in 3D printing relate mostly to how plastic behaves when it is melted and extruded.

And I think you can probably be a great help in solving those problems, but first you have to understand what the problems are. :-)

 

Thank you -- that was a great reply providing exactly the information I was looking for.

Joe

 

[jhinkle 1]

Just to clarify ...

I must have looked like an idiot in my earlier post stating nozzle velocities in the 200 to 300 range.

The UM1 specification clearly states print speeds 30 - 300mm/sec.

Many of the posts I've read are talking about hi-end print speeds.

The reason I asked the question ... 'how Fxxxx feedrate in the gcode would affect anything I'm doing.' ... was when I looked at Cura feed rates and compared those to what people are talking about --- there turns out to be a magnitude of 10 difference --- hence a sanity check.

Also take in the "Marlin" (correct fishy spelling this time) - default JERK is 20mm/sec (most of the time they divide that by 2 --- so 10 - 20 mm/sec jerk is huge compared to printing at 30mm/sec.

I am now comfortable with the magnitudes of the numbers which reassures my understanding of the system.

Thanks again for your replies.

Joe

 

[illuminarti 18]

Joe -

Just to confirm... 'Jerk' as used in Marlin is the maximum allowed instantaneous velocity change at a corner; not the technical sense of rate of change of acceleration. So if you have two line segments which are co-linear, or almost so - such as tiny segments of a large circle, then the velocity change is much less than the 20mm/s jerk, so the head is allowed to carry on without decelerating and re-accelerating at the corner. If you have a right angle corner, then even though the segment speeds are identical, the velocity change is much greater, since you switch axes, and come to a dead stop in one direction and then accelerate away in the other direction. In that case, the planner will reduce the speed at the corner to keep the total velocity change at or below the jerk setting. This reduces ringing effects caused by the head oscillating as it dissipates the energy on an axis when it stops suddenly (it also helps keep the printer a lot quieter, and reduces wear and tear on the mechanics).

Jerk is a speed limit that is applied to the move planning. When the speeds are low, it doesn't apply. When you're printing at most normal speeds, it has a smoothing effect in the corners.

 

[jhinkle 1]

Thank you.

I had the correct understanding of what jerk is and when its applied.

I was only commenting on the magnitude.

Thank you again.

I should be all set for now. Time to actually build the controller and test its performance compared to Marlin.

Thanks again for everyone's replies.

Joe

 

[illuminarti 18]

Now I'm confused.

Why is a jerk setting of 10 or 20 mm/s surprisingly high compared to a print speed of 30mm/s?

 

[jhinkle 1]

It was just a feeling.

THAT is where my ignorance starts -- that relative magnitude.

Once I can control the process, I will have a better feeling for the magnitudes --- Thanks for you input.

Just your comment inferring that 20mm/sec jerk compared to a nozzle velocity of 30mm/sec is valuable feedback.

Thank you again and I'm sorry if any of my questions were irritating in any way.

Thanks to all for your insight and replies.

Joe

 

[illuminarti 18]

No, Joe... not irritating at all. :-) I'm happy to help, really. I'm just not sure I'd approach the challenge in quite the same way. ;-)

But we're here to help, so don't hesitate to ask if you have more questions.

 

[gr5 2,167]

I understand his point completely. If jerk is 20mm and your print speed is 30mm/sec there is almost no acceleration. It *starts* out at I believe half the jerk speed (not sure why not at the jerk speed) for the very first line segment and accelerates up to 30mm/sec. So a 20mm jerk is pretty high (lenient) compared to a 30mm/sec printing speed.

Joe I think you'd like these photographs (with words) that tell a lot about certain aspects of 3d printing - easy reading since it's mostly all pictures:

http://umforum.ultimaker.com/index.php?/topic/1872-some-calibration-photographs/

There's some printing in the purple cubes at 200mm/sec. But quality isn't going to be quite there if you want something that is a piece of art. But for functional stuff that can be ugly, 200mm/sec is reasonable as long as you keep the temperature high.

 

[jhinkle 1]

gr5:

Thanks -- that helped.

Joe

 

[jhinkle 1]

I made another observation that I would like to confirm.

From the config file:

 

#define DEFAULT_AXIS_STEPS_PER_UNIT {78.7402,78.7402,200.0*8/3,760*1.1} // default steps per unit for Ultimaker

This states that the step/mm associated with the extruder is 760 * 1.1 = 836 steps/mm.

Flip that for .0011961 mm/step.

When I look at a gcode segments for the "robot" that comes with the UM1 -- I see that the number of E steps is about 20% of those steps associated with the X or Y axis (whichever one has the most steps for that segment that has the most steps).

Here are a couple of examples:

 

line 149 ----ESteps 11.971534 Xsteps 60 E/X = 0.1995

 

line 608 --- ESteps 2.349161 Xsteps 11 E/X = 0.2136

That provides what I consider coarse control over the extruder

This is a strong reason to move from 2.85mm filament to 1.75mm filament.

Gcode line 149 above - the length of filament associated with that segment was .01432 mm

The controller is expected to deposit that length of filament over the 60 steps of the X stepper.

The volume of a cylinder = Pie*R*R*height.

That means the volume of plastic in .01432 of 2.85mm filament is .0913532 cubic mm.

That same volume of plastic using a 1.75mm filament gives a length of .03798mm.

Given today's controller, the Extruder stepper would make 32 steps --- that is almost 3 times the number of stepper steps to control the flow at the print head's nozzle.

Any comments of this observation are welcome.

Joe

 

[owen 19]

X is only moving less than 1mm though Joe

 

[illuminarti 18]

Yes. One of the disadvantages of thicker filament is that you get coarser-grained control over the advancing of the filament compared to the thinner filament. On the other hand, there are other tradeoffs that work in the opposite direction, such as the higher pressure that can be exerted from the larger cross section of the 'plunger' that the filament forms in entering the melt chamber.

Having more steps would give finer control - especially to keep the filament flow in step with the head during acceleration and deceleration phases.

However, the line segment you are quoting is about 2mm long, it looks like, based on the amount of filament used (or 1mm if it's sliced at 0.2mm layers) so having ten discrete extrusion events within that short of a gap is still pretty fine control in absolute terms. In practice the extrusion is proportional to nozzle pressure, so that tends to smooth the flow rate quite nicely, so that it's not too much of a problem in practice.

 

[illuminarti 18]

Ah yes, I see you mentioned the number of x steps - I was working backwards from the volume and the 0.4mm extrusion width. Maybe it is a 0.2mm layer height... or 0.15? Whatever, as Owen says, it's a pretty short line.

Once again, the mechanics of head positioning and the mechnical properties of melted plastic dwarf any effects caused by the limitations of the electronics.

 

[jhinkle 1]

X is only moving less than 1mm though Joe

 

The actual length is not important -- I was a showing real-life example from the robot gcode file.

Longer length beads still show the number of extruder steps to be about 20% of the steps in the longest axis.

I understand that the flow of plastic will tend to smooth things out --- but if finer control was available --- it could only improve the result.

Joe

 

[illuminarti 18]

If you really want to worry, bear in mind that the steps-per-mm for e on the UM2 is about one-third of that for the UM1. That actually does concern me a little bit, for the reasons you mention - but I'd challenge you to find any other FDM printer that gives better quality prints, so I don't let it bother me too much.

 

[jhinkle 1]

I'm just trying to get an understanding of the UM1.

Knowing how it works - where the strengths and weaknesses are will enable me to properly tune the machine and quickly start producing quality prints.

It will also point out areas where I can play in the future --- in an attempt to better my machine --- the whole reason I purchased the UM1 over another machine.

Joe

 

[gr5 2,167]

JHinkle - I think your reasoning is correct. 1.75mm filament is better for very fine control and I disagree with Illuminarti's assertion that you can get higher pressure with the 3mm filament. The stepper can only put out so much force and with the larger surface area as the 3mm filament goes into the head means it has to push harder to achieve the same pressure. Pressure, typically pounds per square inch is heavily dependent on the "square inch" part.

But 1.75mm filament isn't going to be as strong as 3mm along it's length and so you can't push it as hard as you can push on 3mm filament so mabye it works out about the same in the end.

JHinkle -keep in mind that the UM1 feeder can apply about 22 pounds of push/pull on the filament before it starts to grind it to powder and this force is commonly achieved for most people as many people like to push their printer to it's limits (to save time).

 

[jhinkle 1]

One of my upgrades I am looking at, once I replace my controller, is to put some filament positioning feedback sensor in place.

Feedback will allow me to see if the filament is moving or just grinding - but the feedback requires more controller resources --- hence my new controller.

I already have an idea for the sensor -- just need the additional controller capability to manage it.

Joe

 

[gr5 2,167]

This is smart and I think someone said UM may have tested something similar maybe.

 

[albert 2]

Very interesting thread...just reading it gave me a much better understanding of the meaning of some of the parameters.

Feedback: Would it be possible to implement feedback on all axes? this way the controller knows the real position of the printer nozzle and not just its assumed position. Would this help to get rid of certain problems and deformations? On the better laser cutters there is some kind of optical encoder riding on the x and y axes to provide direct feedback.

Probably too expensive and too cumbersome....

[jhinkle 1]

My statement was flawed so I removed it ...

Joe

 

[IRobertI 520]

Where did you find information that says UM is going to use 1/256?

 

[jhinkle 1]

I posted what I was doing in the "Alternatives" forum.

Here's my post --- see the end.

http://umforum.ultimaker.com/index.php?/topic/4048-is-there-a-need-for-a-100mhz-32bit-arm-controller/

Did NOT say they were Going --- I take their efforts as just good advanced R&D.

Joe