Printing speeds

[edward]

This post about Nicholas Seward's Simpson printer has me wondering as of late: http://forums.reprap.org/read.php?178,2 ... msg-245989

(watch the video for full background on my question)

He states that he is running infills at 180mm/s. I beg to differ, based on what I see in that video. Recently, for larger parts, I've been running 110mm/s perimeters and 150mm/s infills, with 0.2mm layer height and an E3D hotend. I'm often bumping the feedrate slider up a bit due to impatience and a desire to see what she can do. My print results are damn good, aside from a few burps here and there where the machine hiccups and pauses (I should probably stop giving it so much soda). I'm running Johann's Marlin with autolevel and print over serial, so I think that the hiccups might be due to serial buffer issues, but it really isn't the point.

Basically, at these speeds, my machine is running ridiculously faster than that video (I'm working on getting some videos together). Add to it the impressed comments of users in the Google Groups thread (https://groups.google.com/forum/#!topic ... 1WCwsBTye0) and I really begin to wonder what speed others are using.

So, the question: what kind of perimeter and infill speeds are you guys using? After seeing the video, do you feel that the machine is running impressively fast?

If my machine is running ridiculously fast, then I'm bragging Otherwise, I'm genuinely curious.

[edward]

Sooooooooooooo.........nobody here has any thoughts on printing speeds? Really?

[lordbinky]

I haven't seen the video. But the difference between their claim of 180mm/s and your 110, doesn't suprise me if ONLY print speed was the setting turned up on Seward's machine.

Like most things, a single value isn't telling the whole story. A print speed setting of 110 with jerk and accel at 100 is going to be much more impressive than print speed of 500 with jerk and accel at 1.

To realize the limits of my build and learn the practical affects of the settings I turned up each of the movement settings to crazy high values and then left one low. That's also how I learned that turning up my jerk settings removed blobbing from square corners for me.

I think my jerk is at 60 (80 makes my machine start to wobble) and the accel is some absurdly high value right now, so I don't go over 100mm/s. I'll refine it soonish but I got onto a re-calibration kick recently trying to get bed leveling and print_radius just right for really fine layers (the machine danceing will also encourage an overhaul). Hm, odd my first response to increased printing speeds is to drop the layer height down so the prints take the same overall time, instead of using the speed to print more objects.

[edward]

Thanks for the response blinky. Through all of the discussion around here I haven't really seen to many numbers about operating parameters, so I was curious. I think you nailed why they aren't so useful in that form.

You're response was a bit of a "duh" moment for me. I've been playing with the jerk setting, I think after seeing one of your comments about how it removed the corner bulges when increased, and have noticed that it seems to have the largest effect on overall speeds. I don't know why that didn't stand out as so obvious. Since your response I've also studied the jerk implementation and realized that it is not true jerk (that is, the derivative of acceleration, or rate-of-change of acceleration), but rather a speed limit for movement transitions. That is, a setting of 60 essentially means that the motion planner de/accelerates such that the velocity at the transistion between two consecutive movements does not exceed 60 mm/s. This is why higher values cause more - here it is - jerky operation: the machine is making closer to near-instantaneous transitions.

I suppose that you already knew this, but it helped me, and hopefully others, to put it into these words.

I now also get the feeling that more people need to understand how all of the motion parameters affect your *actual* speeds during the tweaking of the first few prints. Maybe I'm wrong and I'm alone in this confusion, but, some Youtube searching reveals a lot of people claiming this speed or that (seriously, look, you'll see people claiming 300+ mm/s movements) when it is obviously no where near. Add to this that these parameters are highly individual-machine specific, and maybe I'm starting to realize that quantifying the print characteristics is less useful than judging the quality of the output.

I suppose that is the efficiency-minded engineer in me, acting with a similar mindset to that which you described.

[dsnettleton]

Just to clarify, the jerk settings you're talking about are in the printer firmware, while the speed settings are in the slicer. So using Repetier firmware with EEPROM, I can change the jerk settings on the fly with my LCD screen. Is that right?

[edward]

Yes, you are correct. I have found that it is best to leave all of the slicer acceleration settings at 0 and only adjust those in the firmware (EEPROM or Configuration.h).

IMHO the slicer should stay out of the way when it comes to the details of getting from point A to point B other than maximum speed, but even that is a "suggestion" because your firmware settings might not allow it. The firmware should be doing all the rest.

I don't have an LCD so I can't answer to that point, but yes, you can change the firmware acceleration(s) and jerk settings in the EEPROM.

[lordbinky]

About print speeds:
What I think is funny, 90% of the work has been done to judge the practical speed of the configuration!

All that's needed is to print a heavy piece of paper with a 10mm square grid, slap that on a piece of cardboard, and place it in the background behind the printhead when taking a video. We could then objectively identify and confirm the print speeds in these videos (if you trust the people not to tamper with the video).

Edwards previous post:
I'm with edward on this, I also avoid introducing additional software controls to the print. Nothing inherently wrong with it, it is just makes things less predictable for me.

When edward said earlier the definition of 'jerk' in firmware not the same 'jerk' as in physics, but acceleration is the same for both. That is a good example of how confusion around a settings can happen (You just can't get that level of ambiguous if you couldn't so easily slap conventions in the face. Yay English!) and without checking you don't really know.

There are also usually differences between how firmware and external software handle things (which is often the case when you have different levels of computation power to work with) and that isn't even considering it is different people doing the implimentations... I do like Kisslicer's destring implimentation better than the firmware's. I really wanted to like the firmware's though.

[cope413]

Yes, I've noticed the same thing. I was just on the RepRap forums and saw someone claim to be printing at 120mm/s and 300mm/s infill on his bone stock Solidoodle 3.

The thought of getting anywhere close to those speeds and getting decent print quality on my cartesian machine makes me laugh.

Most have no idea of the firmware settings and assume that their slicer settings are their actual printing speeds.

Binky - you said that at 80 Jerk, your machine starts to wobble.

What is wobbling/vibrating? Do you think the dampers will help minimize that and therefore allow you to increase that?

I'd like to start pressing the limits of the machine now that I'm comfortable with the new hotend and not worried about jams or cranking up the temperature with higher speeds...

[edward]

Yeah those speed claims are becomming pretty hilarious. More than ever I really want to measure actual speeds so that I can realize just how funny they are.

While this thread started a bit slow, I'm glad it has picked up some. The funny thing is that I feel even less qualified to desribe the speed at which my machine operates. Not that it is that important, as I'm thrilled with the quality of the prints I'm getting. I just can't resist the urge to push things for efficiency's sake, and bragging rights

So, here's my current setup: running Johann's Marlin for G29 autolevel with mag-arms and an E3D 0.4mm hotend, usually running 0.2mm layers, sometimes 0.125mm.
- Maximum feedrate: {300,300,300,175}
- Maximum acceleration: {5000,5000,5000,4000}
- Default Acceleration: 4000
- Retract acceleration: 3600
- X/Y jerk: 80
- Z jerk: 80

The accelerations I pulled from somewhere dark, but they seem to be working well. Right now my machine can print so fast that it looks like it's doing the Harlem shake, but surprisingly, the quality is the best I've ever gotten. I mean other than some retraction hairs that I can't seem to eliminate, things are awesome.

I seem to be settling into setting my perimeter speed to 90-110 mm/s and my infills to 140 (in Slic3r). My first layers are 30-45 mm/s, depending on detail level, and my solid fills are something like 70% speed. Even with these settings I usually end up bumping the feedrate slider to the right.

Binky - you said that at 80 Jerk, your machine starts to wobble.

What is wobbling/vibrating? Do you think the dampers will help minimize that and therefore allow you to increase that?

I've been experimenting with weight for damping. I tossed a big chunk - I'd say ~3lbs - of aluminum on the top of my machine and it seemed to help. I couldn't quantify anything, but it certainly made things quieter. I also have a 5/16" aluminum build surface which I'm sure helps, too.

We're starting to get to the original point of me starting this thread. I too want to push the machine but didn't know where to begin talking about it.

[cope413]

You can approximate your actual average printing speed fairly easily by measuring filament used and using total print time. I've done it on some calibration cubes.
I used the remnants of a spool, measured a length of 2000mm, printed with it, then remeasured the strand and divided by the print time.

It's not super accurate but its a fairly decent estimate.

It's actually fairly interesting. At 20 Jerk, When all speeds set at 40mm/s, my calculation worked out to 38.5mm/s. at 80mm/s, I got 62mm/s. at 100mm/s I got 68mm/s.

I bumped my Jerk up to 60 and my actual speeds got much closer to the desired settings.

Haven't touched it since I got my new hotend, and I have the dampers that will be arriving soon. I'll be playing with it in the coming weeks to see what kind of speeds I can increase to.

[gestalt73]

This whole time since I first built the printer, I've been tinkering to maximize my print speeds.

I'm doing mostly internal/structural and rapid prototyping prints, so I'm currently running a 0.70mm nozzle with 0.333mm layers.

I currently top out at 70mm/s perimeters, 70mm/s infill. If I go faster, it's filament starved for the first cm or two. and the infill gets really spotty. appears filament starved.

In the firmware, I have x,y,z max speed at 300mm/s, x,y,z jerk at 60, x,y,z acceleration at 6000, extr 1 start, max 80mm/s, extr 1 accel 6500 (I think I cribbed those from another of lordbinky's posts)

For slicer settings, I have x,y,z speed set to 300mm/s, and print and infill start at 40, 20 increase per layer to max of 70 for large prints, 40 for small prints.

With these settings, I'm flinging an average of 42 cm^3 of plastic per hour.

I'm always looking for tips to increase speed.

I'm not sure if I'm hot end limited, power limited (hot end power fluctuates between 80%-100%), or extruder limited. I suspect it's a combination of at least two.

What happens if I go faster, is that the infills get messy, and perimeters are filament starved for up to 2 cm (pits and thin)

[Flateric]

I have spent many countless hours on this very topic. Here is the fastest I have ever managed, this is before my mag arm and dampers and with a e3d hotend. I could not get the stock hotend to maintain quality anywhere even close the same performance levels.

http://youtu.be/4QBdDmWD5kc

[edward]

You can approximate your actual average printing speed fairly easily by measuring filament used and using total print time. I've done it on some calibration cubes.
I used the remnants of a spool, measured a length of 2000mm, printed with it, then remeasured the strand and divided by the print time.

It's not super accurate but its a fairly decent estimate.

It's actually fairly interesting. At 20 Jerk, When all speeds set at 40mm/s, my calculation worked out to 38.5mm/s. at 80mm/s, I got 62mm/s. at 100mm/s I got 68mm/s.

I bumped my Jerk up to 60 and my actual speeds got much closer to the desired settings.

Haven't touched it since I got my new hotend, and I have the dampers that will be arriving soon. I'll be playing with it in the coming weeks to see what kind of speeds I can increase to.

I'll have to try this. I like this idea, and it sure does make it easy to compare printers with a completely objective metric.

What happens if I go faster...and perimeters are filament starved for up to 2 cm (pits and thin)

I've had this, too. It was really bad with the stock hotend. I have to use the "Extra length on restart" setting in Slic3r to combat this (I'm at 0.2mm at the moment), but I don't like having to. Something doesn't seem right about it.

Somewhat related question, have you guys calibrated your extruder STEPS_PER_MM or are you using the extrusion factor method for calibration as described by Polygonhell? (http://forum.seemecnc.com/viewtopic.php?f=54&t=1163)

I have spent many countless hours on this very topic. Here is the fastest I have ever managed, this is before my mag arm and dampers and with a e3d hotend. I could not get the stock hotend to maintain quality anywhere even close the same performance levels.

http://youtu.be/4QBdDmWD5kc

Finally! Now this looks similar to how I'm operating. So are you running like this now with the mag-arms?

[gestalt73]

Oh yeah, followed polygonhell's extruder calibration tutorial the moment I saw it.

I've had this, too. It was really bad with the stock hotend. I have to use the "Extra length on restart" setting in Slic3r to combat this (I'm at 0.2mm at the moment), but I don't like having to. Something doesn't seem right about it.

So, maybe a hot end upgrade would help? I've been eyeballing that E3D one everyone's raving about.

Yeah, in KISSlicer, I'm running with 9.6mm retracts, and 10mm prime, 120mm/s retract speed. it's a technicque that definitely helps on parts with larger features, but if I'm not careful it makes smaller more detailed features "juicy" with too much excess filament. Now that I'm thinking out loud, I wonder if I could tweak that with min/jump and trigger settings.

The other setting i've messed with, with limited success, is to change the extr1 start speed. I've gone as high as something silly like 200mm/s, which sometimes helps a bit, depending on the part.

[edward]

Yeah the E3D rocks.

I've never used KISS. Is the difference between retract and prime essentially "Extra length on restart?" I'm assuming you're talking about using ABS? If so, are you seeing any little retraction hairs with those settings? What's your extrusion temp?

The other setting i've messed with, with limited success, is to change the extr1 start speed. I've gone as high as something silly like 200mm/s, which sometimes helps a bit, depending on the part.

That's fast. I'll have to try something like this and update with results...

[lordbinky]

Right now my machine can print so fast that it looks like it's doing the Harlem shake, but surprisingly, the quality is the best I've ever gotten. I mean other than some retraction hairs that I can't seem to eliminate, things are awesome.

I seem to be settling into setting my perimeter speed to 90-110 mm/s and my infills to 140 (in Slic3r). My first layers are 30-45 mm/s, depending on detail level, and my solid fills are something like 70% speed. Even with these settings I usually end up bumping the feedrate slider to the right.

Binky - you said that at 80 Jerk, your machine starts to wobble.

What is wobbling/vibrating? Do you think the dampers will help minimize that and therefore allow you to increase that?

I've been experimenting with weight for damping. I tossed a big chunk - I'd say ~3lbs - of aluminum on the top of my machine and it seemed to help. I couldn't quantify anything, but it certainly made things quieter. I also have a 5/16" aluminum build surface which I'm sure helps, too.

When my jerk setting is at 80 the machine itself looks like it's warming up before it tries to dance like the toaster from ghostbusters 2. At that time I think the speed was 120 mm/s and the accel is set to something silly like 4000 as well. The rostock is on feet right now, if I remove those I think it will sit still better. Adding weight to the top sounds like a great idea as well. I have dampers on and it doesn't SOUND bad, the dampers are working great. My problem was that it was just visually ominous, and seemed like a good way to check how well you applied your locktite to screws.

I believe I could increase the jerk if I lower the acceleration. To get a feel for how fast I can get away with that my present idea is to set my speed to something silly like 500mm/s and my jerk to something similar, and set acceleration very low. Then I will increase the acceleration to a point I find just beyond acceptable movement (the dancing rostock speed). Llikely doing dry runs at this point just to see how the printer reacts.

[cope413]

When my jerk setting is at 80 the machine itself looks like it's warming up before it tries to dance like the toaster from ghostbusters 2. At that time I think the speed was 120 mm/s and the accel is set to something silly like 4000 as well. The rostock is on feet right now, if I remove those I think it will sit still better. Adding weight to the top sounds like a great idea as well. I have dampers on and it doesn't SOUND bad, the dampers are working great. My problem was that it was just visually ominous, and seemed like a good way to check how well you applied your locktite to screws.

I believe I could increase the jerk if I lower the acceleration. To get a feel for how fast I can get away with that my present idea is to set my speed to something silly like 500mm/s and my jerk to something similar, and set acceleration very low. Then I will increase the acceleration to a point I find just beyond acceptable movement (the dancing rostock speed). Llikely doing dry runs at this point just to see how the printer reacts.

What about securing the base to your tabletop - I'm envisioning something like my drill press table that has adjustable locking jigs - that would lock your base in place?

I haven't pushed my machine to Ghostbuster Toaster levels yet, but I gotta believe that would help a ton

[edward]

I think securing the machine is next for me, too. The weight I added was intended to have a similar effect. I guess when I said that it sounded better I meant that it was the noises due to the machine shaking, not the motors. The weight dampened the shaking enough that I didn't notice as much sound coming from the machine vibrating.

Ghostbuster Toaster levels...it makes me want to say "Toastbuster levels."

[626Pilot]

I can run it as fast as I want. The problem is with extruder heat, blobbing, and stringing. To extrude faster I have to print hotter, or it will jam. BUT, that increases drool. Drool results in blobbing. Unfortunately, every time the printer finishes a line, it stupidly sits in place while it waits for the retract to finish. At the beginning of retraction it will still ooze because it takes time for the pulled-back filament to create vacuum and actually pull the filament backwards. Hotter filament + higher feedrate = bigger blobs.

The other thing is the jerk setting. The higher you have it, the more side loading your printer will be exposed to. That will increase wear on the belts and if you do enough of it I think there's a risk that it could throw your towers out of alignment just a little bit. There is also the question of wear and tear on the extruder system, particularly the Bowden fittings, if you are printing PLA. The stuff is more viscous, which will lead to more jams and dust building up in the EZstruder.

I also have to laugh at the idea of a Cartesian printer that moves the entire build platform back and forth printing anything at 200mm/sec. That is so much inertia. If it was actually doing that it would probably walk itself right off the edge of the table.

[Flateric]

I can run it as fast as I want. The problem is with extruder heat, blobbing, and stringing. To extrude faster I have to print hotter, or it will jam. BUT, that increases drool. Drool results in blobbing. Unfortunately, every time the printer finishes a line, it stupidly sits in place while it waits for the retract to finish. At the beginning of retraction it will still ooze because it takes time for the pulled-back filament to create vacuum and actually pull the filament backwards. Hotter filament + higher feedrate = bigger blobs.

The other thing is the jerk setting. The higher you have it, the more side loading your printer will be exposed to. That will increase wear on the belts and if you do enough of it I think there's a risk that it could throw your towers out of alignment just a little bit. There is also the question of wear and tear on the extruder system, particularly the Bowden fittings, if you are printing PLA. The stuff is more viscous, which will lead to more jams and dust building up in the EZstruder.

I also have to laugh at the idea of a Cartesian printer that moves the entire build platform back and forth printing anything at 200mm/sec. That is so much inertia. If it was actually doing that it would probably walk itself right off the edge of the table.

Nevermind the stress this would place on the freshly created part, especially on taller prints.

[CJGerard]

I did a search and just found this thread. There is some really good info here, but its kinda dated now. So now im wandering how everybody has done since they last posted. Have speeds and feeds changed? Are big blocks of aluminum still the best way to keep the Max from dancing of the table. What type of filament was used to get the best speed/quality?

I've been printing all day slowly working my speed up and ironing out quality issues. I thought I was doing doing well, but im no where close to where you guys are at!

[edward]

The search works! Awesome!

For me, no significant changes. I moved my machine to a different table and used some clamps to hold it down. My speeds are still "high," but more refined. With experience I've been able to recognize the detail-level of a part and choose an appropriate slic3r profile that pushes the speed, but still provides crisp features.