Cura print time errors

[Simonsrk 0]

Hi all!

 This is my first post on the forums, so be gentle!

 

 I’ve been into 3D printing for several years now and am very lucky to own a resin printer and a filament printer. However it’s the latter i really have a problem with, as Cura just doesn’t show the right print times for anything!

 

 Now it could be Cura, or it could be my Anycubic Vyper, however, if it says something will print in an hour and a half - add another hour to it. Large things can take an additional hour, small things an extra half an hour.

 

 What’s happening here as I’m not exactly sure how to marry everything up with the correct timings.

 

 Thank you 🙂

[Slashee_the_Cow 274]

Cura just gets it wrong. I don't know how exactly, since it's creating gcode which tells it exactly how far to move at what acceleration rate (unless the values you've given it for speed or acceleration are higher than what the printer can do/is set to do), but it's definitely not just you. I've had it be wrong by about three hours on one print that took over a day. It can't know how quickly your extruder/bed heats up so that's one thing except that the only one of those I change for my prints is printing the first layer slightly hotter so I don't think we can blame that.

 

I find it's often off by roughly the same percentage of time on a lot of prints, so I just add that in my head when looking at the time.

 

Not just Cura either - there's an OctoPrint plugin I have installed called "Print Time Genius" which claims to run your code through a simulator to give you an accurate time. That's managed to underestimate by over half an hour on a "90 minute" print 😛, I only have it installed because another plugin I use to tell me the time until the next filament change requires it.

Edited February 13 by Slashee_the_Cow

[Simonsrk 0]

This is so annoying.

I'm currently going through my settings and ran a test print last night of a standard six-sided die. Nearly an hour and a half l later (1hrs 26 mins) it finished!

There really is something amiss with the settings as it was supposed to be a twenty minute print.

Getting fed-up now!

[Slashee_the_Cow 274]

Okay, that's definitely well outside the normal range of wrong. Any chance you could share the Cura project file (.3mf, in Cura get it ready to print then go to File > Save Project) and the gcode file you already printed to see if any of the boffins around here can figure out what's up?

[GregValiant 1,260]

There have been two post-processors that are very similar.  I combined them into one and made some other changes.  One change is to allow a "fudge factor" that will adjust the print time that Cura comes up with (when you use the "Display Progress" option).  The "adjusted time remaining" gets displayed on the LCD.

This script will be included with Cura 5.7.  Unzip it and put it into your Configuration Folder and the "scripts" sub-folder.

DisplayInfoOnLCD.zip

On my own prints, I started out with the "fudge factor" at 95% but with 5.6.0 I had to adjust it to 100% as there is pretty much no difference between the Cura calculated print time and the actual print time on my Ender 3 Pro.

 

As Slashee says, much is dependent on your actual settings in the printer for the "Max Acceleration" (M201) and "Max Speed" (M204).

If those numbers are in the printer definition file then they should be correct in Cura.  If they aren't in the definition file, then everything becomes a guess.  There is no way for Cura to query the printer to find out what the settings actually are.  If you enter an accel of 3000mm/sec² in Cura and the printer has a max of 500mm/sec² in it's M201 register, then the Time Estimate can be off by a lot because Cura is acting on incorrect information.

For small models "Minimum Layer Time" becomes involved and the uncertainty can grow.

[Slashee_the_Cow 274]

Please note that despite how awesome @GregValiant's scripts are from a functionality perspective, they cannot overcome crappy design decisions like how most recent Ender printers don't actually show messages from gcode (or a connected terminal) on their screen. If you print from SD card it'll just show its own progress screen with no place for messages from gcode (well, no customisations at all). If you print over USB (which you shouldn't unless you have a dedicated system running OctoPrint or something because there's a multiplicative relationship between how long you've been printing and how likely your main computer is to hard freeze) it just stays on whatever screen it was on when you started printing (mine spends a lot of time on the main menu).

[Simonsrk 0]

I’m running an Anycubic Vyper for my 3D prints and aside from it running beautifully, this is the only problem I have. 
I always thought Cura to be THE go-to for slicing, lately I’m seeing its flaws!

[Slashee_the_Cow 274]

9 hours ago, Simonsrk said:

I always thought Cura to be THE go-to for slicing, lately I’m seeing its flaws!

No slicer is perfect (in any way, let alone time, in my brief dalliances with other slicers the time accuracy is about the same). Please don't leave us, we'd hate to lose you. As I said, if you can post your project file and your gcode we can see if there's anything obvious as to why Cura would get it so wrong. @GregValiant loves diving into gcode to try and find anything wrong 😄 

[jaysenodell 13]

I've run 18 prints today and I tracked the "variance". 3:05 seconds. This was on a "34 minute" estimation with the "exact time" taken from my octoprint log. That 3:05 the exact time it takes for 

G28 XY
M190 S65
G28
G0 Z10.0 F3000
G0 X0 Y0
M109 S210
G1 .... ;; Draws a line down the side then back up. 

 

That said, Octo indicate 4 different completion times that varied from 28min to 90min for these jobs. I suspect there is somehing in firmware limiting the printer compared to the settings you are using in cura. 

[Slashee_the_Cow 274]

1 hour ago, jaysenodell said:

That said, Octo indicate 4 different completion times that varied from 28min to 90min for these jobs. I suspect there is somehing in firmware limiting the printer compared to the settings you are using in cura. 

I deliberately turned down the maximum speed and acceleration because how fast the printer can go is basically a pissing contest between manufacturers right now and the default speed and acceleration settings were actually pulling my model apart.

 

That's why it would be nice if we could see your gcode @Simonsrk, so we can check to make sure Cura is putting the right speeds in.

[Simonsrk 0]

Okay, so, here's a calibration XYZ block.

Not sure about the settings helping, but you never know.

I think my top layer is too thin as I'm getting an incomplete finish on the top.

Any suggestions would be gratefully appreciated.

 

Cheers.

AI3M_xyzCalibration_cube.3mf

[Slashee_the_Cow 274]

1 hour ago, Simonsrk said:

Okay, so, here's a calibration XYZ block.

Not sure about the settings helping, but you never know.

I think my top layer is too thin as I'm getting an incomplete finish on the top.

Any suggestions would be gratefully appreciated.

 

Cheers.

AI3M_xyzCalibration_cube.3mf 108.77 kB · 0 downloads

I'll have a look at the calibration cube profile, but what I really, really want to get is a gcode file that Cura has gotten the time significantly wrong on so I can see if there's anything in the gcode that would cause that.

[Slashee_the_Cow 274]

Okay, so notes on the profile. And I'll warn you now, I'm great at nitpicking, so please don't take it personally:

• QUALITY:
  • This isn't necessarily wrong, but your layer height is higher than average, which most people usually use to print quickly, so they use wider lines. Most of the standard profiles I've seen in Cura use a layer height of 0.2mm with a line width of 0.4mm.
  • Setting the initial layer height as higher than the rest of the print is great for bed adhesion, but in this case the difference is so small it's not going to make a salient difference.
  • Using different line widths for infill and support isn't really a good idea without a good reason, especially with differences this insignificant. It means every time Cura changes from support or infill to something else it has to change the flow rate, which is especially bad with a Bowden extruder because it takes a bit of a time for changes to take effect. This can result in overextrusion or underextrusion, and neither of those is a good look (lumps or missing/really thin bits respectively).
• WALLS:
  • I don't know what sort of hammer you plan on hitting your prints with, but 7 walls is way, waaaaaaaaaay too many, especially on a small print like this. It could be part of your problem at the top, because all of those walls get in the way of the actual design.
    • Even on bigger prints, I usually use 2 walls with an alternate extra wall. Having infill (and not too high a density) is actually your friend! PLA is brittle, so if you have a solid lump of it there's no way to distribute any impact force and it's possible the whole thing can shatter. Infill can absorb some of the impact because there's room for things to move.
  • If you're using a 0.4mm nozzle (come as standard on most printers) you can often get away with 0.26mm as a minimum line width. Lower minimum width = more detail.
• TOP/BOTTOM:
  • EIGHT TOP LAYERS?!?!!?!?!!?!?!!???? On a print this size, that makes the top about 10% of the total height. Top/bottom layers are like walls: the solution is rarely "more", unless the problem is "I printed a single wall and it shattered when I hit something".  You want your "Z" to look better? Use fewer top layers. Gives details much more room to breathe.
    • At a 0.2mm layer height I usually use four top/bottom layers. If you're going to use 0.26mm layer height that's equivalent to 3, but 4 would still work. Again, this is for anything, regardless of size, and the main problems I've had with strength have been things being too hard to deliberately destroy (like when I've made moulds for gluing stuff together, the mould will inevitably get stuck, so requires forceful removal).
    • Same thing about brittle as I said about walls is true here: the more plastic it has to go through to get to an area that might absorb some impact, the less likely it is to actually reach it.
  • Five bottom layers: nowhere near as egregious, but still overkill, especially at your layer height.
  • They might look pretty similar but Lines is a better top/bottom pattern than Zig Zag, because it adheres to the walls better. It just means that it has to make small, non-retracting travel moves at the end of each line. That's not a bad thing. It's not moving anywhere near far enough to leave a glob behind or anything.
• INFILL:
  • For best visual quality you should turn off Infill Before Walls and reset Infill Overlap Percentage to its default calculated value (10%). Both of these mean that instead of printing straight walls and filling the area between them with infill, it prints infill (which you're doing wider for whatever reason and the higher overlap percentage will stick out further) which the walls will then form around, meaning the infill pattern will be visible on the outside.
  • 25% infill percentage: higher than I use for most things (20%) but well within an acceptable range so you get a pass on that.
• MATERIAL:
  • You're underextruding literally everything by setting the flow lower than 100%. It's not uncommon to set things like support interface a bit lower to try and make it easier to remove, but if you feel the need to change all of these, you need to calibrate the E steps on your printer, not fudge it in slicing. You'll get much better results, especially on things like top/bottom where correct flow is important for a smooth finish.
• SPEED:
  • There's no good reason to change the outer wall speed to the inner walls (or most of the print, really). It's another case of flow rate changes potentially causing defects.
  • Same applies to Top/Bottom speed, Top Surface speed. Yes, slower is good for fine details, in theory. You're not going for fine details here. The letters on the calibration cube are definitely not "fine".
  • Most filament spools will have recommended print settings on them, make sure you follow them (unless you know exactly why you're doing it differently). I say this mostly because I've never seen PLA recommend anything faster than 65mm/s.
  • Acceleration control: 1500mm/s² isn't necessarily bad, but you need to keep an eye on it. I've had a print head accelerate so fast away from a bit of a model that it actually pulled the not-quite-set plastic with it and warped it at a lower speed than that.
  • Jerk control. 13mm/s is WAY. TOO. FRIGGIN. HIGH. It controls how much the speed can instantly change at a corner, and the higher it is, the less accurate your movement is (including the potential of vibrating the printer enough to move the print head a step or two out of place, resulting in layer shift).
    • I usually keep it at 5mm/s, but I'll admit I'm pretty conservative. Personally I've only ever seen bad things happen at over 8mm/s.
    • Think of it like a car with the steering sensitivity turned up 300%: when you reach a corner at speed, it's going to be nearly impossible to make the turn properly and not hit a wall.
• TRAVEL:
  • If your printer is properly configured, then retracting at a layer change can do more harm than good. You run the risk of underextruding at the end of one layer and overextruding at the start of the next. Or the other way around, depending on how well timed its extrusions are. Again, not something a Bowden extruder is good at. It's also bad for the filament because it's running it back and forth through the gears more and grinding it up.
  • You only need to lower Retraction Prime Speed if you're overextruding after retraction moves.
  • Z hops aren't really needed unless your nozzle is crashing into printed parts when you're moving. And your Z Hop Height is so, so tiny that it wouldn't change that in the slightest, just wear out your Z motor with unnecessary movements of a few steps.
• COOLING:
  • Actually not bad if you're just printing PLA. If you're going to use other materials you might want to look into using the minimum layer time.
• SUPPORT:
  • You do realise that one of the ideas of a calibration cube is to see how well it prints without support, right?
  • Support Overhang Angle: A bit conservative IMO, PLA that's being printed properly should be able to stand at least 55° (some of the more daring types even go up to 45°). So not necessarily a bad thing.
  • Support Density: Unless you're having problems with your support not supporting stuff properly, there's no need to increase this from the default. It just wastes filament and time.
  • Just a heads up that if you have Enable Support Floor turned off, if you have support which starts on your model, it's going to be a lot hard(er) to remove without the interface.
• BUILD PLATE ADHESION:
  • There is absolutely no reason whatsoever Skirt/Brim Minimum Length needs to be that high. Not only is it a waste of filament and time (especially when you're using multiple layers, which I'll get to) but if you place things on the wrong part of the build plate you might lose the slicing lottery and it'll try to print outside the printable area.
    • Like 60mm would be fine, as long as your filament is good.
  • I've never understood why the default height for a skirt is for three layers. About the best thing a skirt does is get any crap in the system out before you start printing, and if you still have stuff that needs getting out on your third layer, you have bigger problems.
• MESH FIXES:
  • You can turn Maximum Resolution down to like 0.1mm or so if you want to try and eke every last detail out of something you can.
• SPECIAL MODES:
  • Nothing to see here. In a good way.
• EXPERIMENTAL:
  • Conical support at that angle isn't going to make much of a difference.
  • Be careful with your Small Feature Max Length: the higher it is, the more thing get marked as small features that don't really need to be and just increases print time for no benefit.

Umm... told you I was good at nitpicking.

[Simonsrk 0]

Okay, 

Based on your recommendations, I've reset my Vyper to the settings on this 3mf file.

Could you please let me know if there's anything else I can do to improve the quality of my print all the while keeping the timings down to something reasonable? It says it should take 28 minutes with these settings, so I'm going to run it once my current print has finished - currently it's at 67% and has taken 2H 59 minutes for some keys using grey Anycubic PLA.

 

Thank you in advance.

AI3M_xyzCalibration_cube.3mf

[Slashee_the_Cow 274]

Please post the gcode file you're printing too, so we can compare it with the settings from the Cura project.

[Simonsrk 0]

47 minutes ago, Slashee_the_Cow said:

Please post the gcode file you're printing too, so we can compare it with the settings from the Cura project.

- here you go!

I ran the print and it printed in 33min, rather than the 28, so an additional 5 mins isn't the end of the world.

Also, a lovely finish on it too. Now to try the settings on something else!

 

🙂

AI3M_xyzCalibration_cube.gcode

[Slashee_the_Cow 274]

The gcode shows it's trying to print at the right speed - and even on a 28 minute print, 5 minutes is still within a reasonable margin of error.

 

Try to keep the 3mf files and gcode for all your projects and if there's anything that takes a ridiculous amount longer than it says it should post them and we can have a look.

[Simonsrk 0]

Thank you 

[jaysenodell 13]

Just to add fun to this… I’m starting to see jobs finish 60min UNDER cura time. I’m not complaining but I thought I add that to the fun of this conversation. 

[DivingDuck 105]

Then you should check your printer settings in firmware with the settings in your Cura Printer Settings. This is an issue that all slicers have when max speed and -acceleration settings are not matching. You see this more often with Cura because there are some default values that are not from this universe 😉 like default max speed values in the fdmprinter.def.json definition: 299792458000 mm/s. This isn't a problem if you have a good printer profile, but in case it is not complete and do not have  the correct values for your printer then you will face the behavior you see as those default values will then take place in the calculation.

 

Add the plugin Printer Settings from Marketplace and you can check the values and adjust those that are not matching with your printers max values in speed and acceleration.

 

And remember, it is always a good idea to make a backup of your configuration folder first or at least a copy of your printer profile for experimenting with a profile where you don't know how this will end because of missing experience. We all started with low experience once.

 

 

Edited February 20 by DivingDuck