Speed question

[Chen Nicholas]

Hello there,

Im a prusai3 user and just updated my ramps to the smoothie board. It seems that I got some luck to hook up everything correctly and made it work.

However, I have two questions regarding the extruder calibration and speed.

For the extruder part, when I calibrated the stepper motor for my extruder, I found that the movement was very slow. I checked the marlin setting for the extruder and found this

#define DEFAULT_AXIS_STEPS_PER_UNIT   {80,80,400,750}  // default steps per unit for Ultimaker
#define DEFAULT_MAX_FEEDRATE          {500, 500, 2, 25}    // (mm/sec) 

In comparison with the default setting in Smoothie board, it shows:

extruder.hotend.steps_per_mm                    140              # Steps per mm for extruder stepper
extruder.hotend.default_feed_rate               600              # Default rate ( mm/minute ) for moves where only the extruder moves 

Marlin shows the feedrate is mm/sec and smoothie board shows mm/minute. I wonder if this is the case caused the slow movement of the motor? Also, the default motor setting in smoothie board for XYZ seems quite similar to marlin, but the extruder one is very differet ( 750 vs 140 ) I wonder if this is also something to do with the feedrate as it calculate mm/ minute?

Second question, I was told that Smoothie board makes the marchine print much faster than the ramps, I wonder in what sense it can be much faster? Is it simply just becasue the processor is better?

I assume that we all know that there is speed setting in any gcode generation program, is it just by tuning the speed with high value, then my machine will move faster? If this is the case, I can do this with my old machine and smoothie board doesnt give me any advantage. If I should change any setting in the config file to make it move faster, what is it?

Im printing the same object in the printer with ramps, it took me 10 mins, but the one with smoothie board took 12 mins.... I wonder how this would happen. Please help me with providing some technical details.


Many thanks.


Best,

Nicholas

[Mark Rehorst]

First, you only use the default settings if they make sense for your printer.  If Marlin was set to one value and smoothiware is set to a different value, the printer is going to behave differently.  

Steps/mm is on the input side of the extruder.  You calibrate it by jogging 100mm or so of filament through the extruder and checking how much actually went in, then adjust the steps per mm value appropriately.  If you didn't change anything but the controller board, an the old board was running at 16 usteps per step, the smoothieboard will behave identically to the old controller (assuming all steps/mm, speed, acceleration, jerk/junction deviation parameters are set the same).  With the same config parameters as the old controller, the smoothieboard will give the same print results.

The default rate of 600 mm/min is 10 mm/sec.  That is a LOT of filament to move through a hot-end unless you're using a very large nozzle diameter.  10mm of 1.75mm filament is about 24 cubic mm.  If you have a 0.4 mm nozzle, that's enough filament to lay down a track that's 191 mm long in that 1 second period - i.e equivalent to printing at 191 mm/sec.  Very few printers can print that fast.

Smoothieboard doesn't make the printer go faster unless you specify faster print speeds.  Smoothieboard can generate up to 100k pps to the motor, Arduino/RAMPS is limited to 40k pps.  The printer speed limit is usually a function of the motor performance and the moving mass and other mechanical factors, not the motor driver or controller board.  You can try to drive it faster but it may not keep up.  What the 100k pps really gets you is the ability to use finer microstepping without losing speed (hence the name "smoothie").  Of course, the on-board drivers are fixed at 16 usteps per step, so you won't see any improvement is smoothness in switching from one 16 ustep driver to another, but you can connect external drivers that will allow finer microstepping.

If you want to print faster you specify it in the slicer.  The smoothieboard config file has settings for maximum speed limits and acceleration and junction deviation.  You can set the speed limit very high but if you set acceleration very low the printer may never achieve the maximum speed you set.  So printing speed is a function of settings in the config file and settings in the slicer.  The slicer settings will not increase speed over the maximum value set in the config file.

SmoothieBoard provides all sorts of advantages, but an automatic increase in print speed is not one.  You have to set the config parameters to match the physical performance limits of your printer.  Then you can tell the slicer how fast you want to print (within the limits of the machines capabilities).

[Triffid Hunter]

very few printers can print at 191mm/s? damn, I've been using 200mm/s infill and 500mm/s travels for years, have printer speeds really not increased in that time?

[Mark Rehorst]

That's a hell of a lot faster than most.  What material do you print at that speed and what does the printed part look like?  Is the printed part mechanically sturdy?

The OP will definitely never make a Prusa I3 go that fast- the bed moves in the Y axis.

I see a lot of G+ posts by people building machines to run at 200+mm/sec.  They post a lot of video clips of the machines moving the carriage around at high speeds but they never seem to post photos of the printed results.

[Chen Nicholas]

Hello Mark,

Thanks for your instant reply. That was great. Just to make few things you said more clear.

You said  “The default rate of 600 mm/min is 10 mm/sec.  That is a LOT of filament to move through a hot-end unless you're using a very large nozzle diameter.  10mm of 1.75mm filament is about 24 cubic mm.  If you have a 0.4 mm nozzle, that's enough filament to lay down a track that's 191 mm long in that 1 second period - i.e equivalent to printing at 191 mm/sec.  Very few printers can print that fast.”

In Marlin, the federate for extruder is 25 mm/sec, wouldn’t that be giving much more than 10mm/sec? or the way smoothie board defines the feederate is completely different from Marlin, therefore, 10mm /sec makes sense that giving more than 25 mm/ sec?  The filament Im using is 1.75, and nozzle size is 0.4. – Exactly same as your assumption.  I used to use the travel speed 150, so if you said 10 mm/sec is enough for flowing the filament out with the speed 191, then in the case of 25 mm/sec, I assume that more filament can flow out and the speed can go even higher than 191?

If not, is that the case you mentioned the hardware itself limits the speed ?

“Smoothieboard doesn't make the printer go faster unless you specify faster print speeds.  Smoothieboard can generate up to 100k pps to the motor, Arduino/RAMPS is limited to 40k pps.  The printer speed limit is usually a function of the motor performance and the moving mass and other mechanical factors, not the motor driver or controller board.  You can try to drive it faster but it may not keep up.”

So the question here is what is the highest speed and acceleration  the Smoothie board can be applied and Marlin cannot, which has been tested?   ( I wonder how those people who posted videos with their smoothie board testing on youtoube set their printers, those printers look moving very fast ) https://www.youtube.com/watch?v=U_qct8ecnas  is the performance of  a delta printer different from prusa technically?

Also, regarding the stepper motor driver, are you saying that I will need to use a higher resolution stepper motor driver such as DQ860MA  to make the machine runner faster? Or the finer step motor diver just for keeping it stable  when driving faster?

However, last question I still haven’t figured out why it moves so slow when I calibrate my extruder? The speed becomes normal when I print stuffs, but while I calibrate, it took ages to complete a 100mm extrusion? I did have this problem with my ramps before, and normally it happens because of the stepper motor driver, so after changing it, it’s okay…. I wonder if this is the case?

Smoothieboard does look amazing, so I really wish that it can really make a printer performing better since the cost of this board is much more than a ramps.  Please let me clarify the point mentioned above.  Many thanks, Mark.

Best,

Nicholas

[Triffid Hunter]

Mark,

The prints look better than most I've seen.

I print both PLA and ABS at that speed on a MMax which also has a moving Y bed like the prusa i3, and fwiw a somewhat larger one at 230x290mm

http://youtu.be/wksLdtMm6Ko and http://youtu.be/YV8mXkf_uco may pique your interest, and http://i.imgur.com/vbYYkVe.jpg may give you a better idea of the results I get printing at those speeds.

[Mark Rehorst]

The feed rate you're talking about is the maximum limit that is set, not the rate at which the machine is feeding filament when printing (unless you set that limit to a very low value).  The actual feed rate varies during a print depending on how fast the extruder is moving, and it is constantly accelerating and decelerating.  Likewise, the maximum feed rates for the axes are upper limits, not necessarily the speed at which the machine will print.  They are there to keep the controller from trying to exceed the physical performance limits of the machine.

Yes, if you set the max feed rate of the extruder higher, you can theoretically get it to squirt out plastic faster, which can theoretically mean you can print faster, but there are real physical limitations that may prevent ever reaching those higher speeds.  Forgetting about the physical limits in the axes for a moment, and just looking at the extruder, there is a limit to the maximum speed at which the controller can drive the motor.  If, for example, you use a 200 step per rev motor and 64:1 microstepping (requires an external driver), it will take 12800 steps per rev.  The Arduino can generate a maximum of 40 pps which means the motor speed can't get higher than 3.125 rev per sec.  The smoothieboard can generate 100k pps, so in theory, the motor could be driven at 7.8 revs per sec.  If the filament drive gear is 10mm dia, than each rev represents 31.4 mm of filament, so maximum filament feed rates are 98mm/sec and 245 mm/sec.  That of course means nozzle output line length is even higher.  Using the standard 16:1 microstepping would multiply all those rates by 4x.  So you can see that the extruder drive speed is NEVER going to be the limiting factor.  Either board can drive the extruder much faster than it would ever need to go to print.  Reality is that you have a limited power going into the heater and you won't be able to melt plastic as fast as the motor can theoretically push it out.

The maximum print speed is limited by a lot of factors including the motor performance (torque goes down as speed goes up), the moving mass- you have to be able to start and stop the motion precisely or you'll have crappy looking prints.  If you set the print speed, acceleration, and/or junction deviation (roughly equivalent to jerk) too high, the motors won't be able to keep up and the print layers will shift.

No, a higher microstepping driver will not speed up the machine, it will reduce the maximum possible feed rate.  What it does do is smooth out the motion of the motor, reducing noise and vibration.  Most printers aren't operating anywhere near the maximum possible feed rate of the controller, so with the smoothieboard you can use higher microstepping ratios and still maintain your realistic print speeds.  

[Mark Rehorst]

The print looks good and your machine is blazing!

I have a little bigger set up (317x305 bed) with MUCH higher moving mass, so I can't print nearly that fast.  The screw drive I put into the Y axis isn't helping me speed-wise.

I'm going to have to look into making a smaller, faster machine.  I like deltas, but stereo lithography machines are interesting, too.

[Ryan Carlyle]

Chen, the huge benefit to Smoothie is when you're 1) using a Delta where lots of extra math is required or 2) printing extremely high-resolution models. Smoothie can churn through gcode segments very, very quickly without any motion planner stuttering. That lets you print curvy perimeters at much higher speeds without decimating the model mesh down first. People using Marlin/RAMPS tend to have to run complex perimeters at very slow speeds to avoid pause-zits and motion planner jerks. Smoothie lets you run high-res perimeters as fast as your printer hardware can handle. 

For example, slice this insanely high-poly model http://www.thingiverse.com/thing:151251 (warning, artistic nudity) with a slicer that does not automatically merge/simplify paths (like S3D) and try to print perimeters at >100mm/s with various printer platforms. Marlin will choke. Sailfish and Repetier will do ok at moderate speeds (say <60mm/s). Smoothie and Machinekit will do great.

Triffid, how much time are you actually spending at 200mm/s during printing moves? The second video you post uses very reasonable perimeter feedrates and a "low-speed" infill pattern. (Lots of turns means acceleration drops the average speed a ton.) But 3k acceleration IS quite fast for that style of bot.

Do you recall the layer height in that video? Most of us with nicer printers these days are limited entirely by PLA melting/cooling rate, not motion hardware. My CoreXY can move at 250mm/s without any issues whatsoever, and I'm sure my Delta could print at way over 500mm/s because of the "fast" arm geometry, but I don't ever WANT to do that. At 0.2mm layers with a 0.4mm nozzle, extrusion paths that run over 90mm/s causes visible PLA temp sags (lack of gloss) and poor layer adhesion due to insufficient residence time in the hot end. (PTFE-lined nozzles with reasonable hot zone length are even worse.) Plus all the usual PLA cooling challenges, of course.

On Saturday, April 4, 2015 at 10:08:07 AM UTC-5, Triffid Hunter wrote:

[Ryan Carlyle]

Just for fun, I'm running a 200mm/s perimeter+infill print on my Smoothie Delta right now. (The lady bust linked above.) The curve fidelity is very good, but with a bowden drive and no pressure advance function, I'm getting an unacceptable amount of blobbing at corners and infill direction reversals.

Even more fun... just dialed it up to 200%. Which I suppose means it's doing 400mm/s? Having some extrusion issues there. Motion fidelity is fine though. https://www.youtube.com/watch?v=y6hRDkPsKVM

[Ryan Carlyle]

Raises a question... when you crank up the speed dial mid-print, does acceleration / junction deviation scale up as well? Or do you keep the same accel settings and just raise the top mid-segment speed?

[Arthur Wolf]

On Sat, Apr 4, 2015 at 10:39 PM, Ryan Carlyle <temp...@gmail.com> wrote:

Raises a question... when you crank up the speed dial mid-print, does acceleration / junction deviation scale up as well?

Nope. You can adjust those separately though.
 

Or do you keep the same accel settings and just raise the top mid-segment speed?

--
You received this message because you are subscribed to the Google Groups "Smoothieware Support" group.
To unsubscribe from this group and stop receiving emails from it, send an email to smoothieware-sup...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

--

Courage et bonne humeur.

[Chen Nicholas]

@ Mark and Ryan,

Thanks for all your detail explanation. That was brilliant.

However, the biggest problem (slow extruder movement)  I encountered which hasn’t been resolved yet. I wonder if you guys can also help me investigate what is the problem that causes this slow movement. That will be very appreciated.

 

These are the things that I’ve already done so far to investigate the problem :

1.       1. Swapping motors and see if it’s the problem of the motor for extruder. The motor is fine.

2.       2. Double check the extension wire I welded. This is also fine. No loose connection or touching to another wire.

3. Changing the number for my extruder.hotend.steps_per_mm ,   689 is the number I used on another printer. I tried both 689 and default value, which is 140. Somehow when I tried to extruded 100mm, it always moved 20mm only. I also tried some random number, the result is same. I assume this is where the problem is.

4.  I have two smoothie boards. I moved everything from one board to another and encountered the same problem. I tried this to identify is it the problem of the stepper motor driver. It seems that both board occurs the same situation. I assume unless Im very unlucky to have both board with bad  stepper motor drivers… so this problem happens, but I guess the chance should be low.

5.. I tried to use different laptops to test if it’s the issue of prunterface since some people mentioned about it, but  no, I got the same result.

 

I got this problem before with my old ramps board, however, I found that it’s the issue of the voltage on my A4988, so it was fixed by either tuning the voltage back to what it should be or changing a new driver. However, Im not sure if this is the same case here as I mentioned in point 4. This also happens to my second smoothie board.

 

Just to make it clear, I mentioned that the printer was moving fine and I tested the printing time. That was only based on the figures shown on prunter face while I loaded the gcode. I did try to print some stuffs with the extruder hooked up, the motor movement seems much normal when printing with a gcode object, but I would like to make sure that my extruder is calibrated properly not just using a random number as more problems will come for sure later. At the moment, I cannot even calibrate the extruder since it only moves 20mm no matter what.

 

I took a video to show what I mean the slow movement of my extruder in the link below, please help me  have a look.

https://www.youtube.com/watch?v=Yc7XKZraDVg&list=FLsKo-DOMj2RONGABZ0iXFmw&index=1 

Thank you very much for all you guys' help.

Nicholas

[Mark Rehorst]

Are saying it prints OK with the steps per mm set to 689 but it doesn't behave right when you send it the extrude command from pronterface?

The laptop and smoothieboard can't share the connection to the uSD card.  Make sure that after you change the config file via the laptop/USB connection that you unmount the card then reset the smoothieboard to ensure that it reads the changes you made. Also make sure that the card doesn't automatically mount on the laptop OS.  That may prevent the smoothieboard from reading the changes you made on the uSD card.

When really weird stuff with no sensible explanation happens, it is sometimes the uSD card.  Try a different uSD card on the smoothieboard.  Just copy the firmware and config files to it, put it into the board and reset the board.  Then run the pronterface test again.

[Triffid Hunter]

@Ryan,

I'm not sure of the average speed after acceleration, I should make a new video with rectilinear infill instead of honeycomb.

The layer height was probably 0.2 or 0.25mm with a width of 0.5-0.6mm, those are my usual settings.

I'm using an arcol v4, and even during long runs of rectilinear infill at 200mm/s I can't see any evidence that it's struggling to keep up, despite its quite short melt zone.

@Mark,

You don't actually need the firmware on the card at all unless you intend to update smoothie to a later or modified version.

only thing that needs to be there is config :)

[Ryan Carlyle]

Triffid, is that ABS or PLA? I can print that fast in ABS with a 0.5mm nozzle, but PLA doesn't come close to that. (Thermal conductivity and heat capacity differences.)

[Ryan Carlyle]

Incidentally, I did some speed tests today, posts here: https://groups.google.com/forum/#!topic/deltabot/UF_Y5IJwNU0

[Triffid Hunter]

@Ryan

I print that fast with both PLA and ABS, no problems with either.

My nozzle is 0.35mm and I use 3mm filament

[Ryan Carlyle]

@Triffid, sorry, but I have to call bullshit. I'm pretty sure you're not printing 20-30 mm^3/sec of PLA through a 0.35mm nozzle with a short melt zone. Something in that doesn't add up.

[Chen Nicholas]

Hello Mark,

Thanks for pointing to the SD card direction. Yes, you totally got it right. The SD card just somehow doesnt update at all. Well... it's a bit of annoying a problem like this happens.

I tried to use another SD card, however, it doesnt work... I assume that this is kind of hardware problem as smoothie said that updating can be as simple as just saving the text file.... it happens to my both smoothie boards.

The way I made it work is to type directly the gcode command M92 to update the E step and overwrite it from Prunterface.. then problem solved.

However, another weird problem happens, just wonder if you have a similar experience... somehow the x axis is shifting while printing, the x axis motor works fine and the belf is tight.Y and Z axis are both fine.  I just double checked these few details.

Please let me know if you also know how this would happen. Thanks a lot.

Nicholas 

[Mark Rehorst]

I struggled with the problem of a shifting x axis for a couple weeks when I first installed my SmoothieBoard.  It turned out to be a problem with the SD card.  I replaced the card with one I picked up at Walmart and suddenly the problem was gone.

--
You received this message because you are subscribed to a topic in the Google Groups "Smoothieware Support" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/smoothieware-support/wKN1uPgV7S4/unsubscribe.
To unsubscribe from this group and all its topics, send an email to smoothieware-sup...@googlegroups.com.

[Chen Nicholas]

I see... I will get a new one and give it a try. Did you pick up the SD card with higher class? The SD card comes with the board is class 4.

Thanks again for your great help.

[Mark Rehorst]

A new class 4 card worked fine.

On Sun, Apr 5, 2015 at 4:34 PM, Chen Nicholas <nickc...@gmail.com> wrote:

I see... I will get a new one and give it a try. Did you pick up the SD card with higher class? The SD card comes with the board is class 4.

Thanks again for your great help.

--
You received this message because you are subscribed to a topic in the Google Groups "Smoothieware Support" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/smoothieware-support/wKN1uPgV7S4/unsubscribe.
To unsubscribe from this group and all its topics, send an email to smoothieware-sup...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

--

Mark Rehorst, DMD

262-321-8264

http://mark.rehorst.com

[Triffid Hunter]

@ryan

Recorded this just for you - http://youtu.be/xQNJgXV6KAQ :)

[Ryan Carlyle]

ABS or PLA? Judging by what looks like a large brim, I'm guessing ABS? I would be impressed but not surprised if you were hitting 30mm^3/s with ABS. I would be surprised with PLA.

Just sanity-checking your figures... from 1:02 to 1:32 the hob rotates 5 times during linear infill. If your hob is 6 mm effective diameter after including bite depth (a guess) then that means you theoretically extruded 94.2mm of filament in 30 seconds (Probably a bit less due to non-linear extrusion rate at high hob shear force, but let's not worry about that.) For 2.85mm filament, cross-section is ~6.38mm^2, for a total volume extruded of ~601mm^3. That's an average rate of 20mm^3/sec. Naturally, the peak rate would be somewhat higher, so it seems reasonable to say 30.

20-30mm^3/sec average extrusion rate through a 0.35mm nozzle with PLA would be very impressive. From looking at the Arcol v4, I would say "quite short melt zone" is inaccurate though. Looks like around 20mm of length from nozzle throat to middle of heat break. The total melt zone length is comparable to the E3D Volcano. This length combined with the 3mm filament means the filament residence time is over four seconds at 30mm^3/sec. That's a lot of time for heat conduction to melt the filament. Which explains why you're able to print at high rates. Makes sense.

[Triffid Hunter]

Next time I'm printing PLA I'll grab a vid of it running with exactly the same settings ;)

The arcol v4 is an interesting design, there's 8-10mm of stainless inside the top of the barrel before the filament touches the aluminium, not entirely sure how much it helps with the melting since its thermal conductivity is relatively low.

The heaterblock actually wraps around the outside of the nozzle block, and the nozzle has a very shallow angle to aid heat transfer to the tip, so it doesn't have a hot zone in the barrel while the nozzle runs slightly cooler like many other hotend designs

[Ryan Carlyle]

Yeah, it's an interesting hot end design, thanks for point my attention that-a-way. Everyone in my circles is all about E3D, Jheads, and Rep2 style hot ends. Good to see people using something else with very competitive results.