SOLVED Different speed for different printers in Cura

[buteomont 1]

I have 3 printers; a Hictop, an FLSun, and a Tronxy.  If I load an STL file into Cura and switch between the profiles of these three printers, it reports different print times for each one.  The Tronxy print time is usually about double that of the other two.  This is with all having the same print speed, travel speed, layer thickness, etc. 

There is apparently some other difference between the printer profiles that I can't find.  Where should I look?

[nallath 1,120]

Have you had a look at acceleration / jerk settings?

[buteomont 1]

Good point. I did, and they were a bit different - 100 mm/s2 on the Tronxy vs. 3000 on the FLSun.  I changed the Tronxy values to 3000 and it made a small difference, but it took only 9 minutes off a 5-1/2 hour print.  The  estimated time for the FLSun to print the same object is only 3 hours 5 minutes. 

[nallath 1,120]

Next candidate is the minimum layer time.

[buteomont 1]

The Hictop had 3, the FLSun had 5, and the Tronxy had 10.  I changed the Tronxy to 3 but it didn't shorten the print time at all.  Good thought though.

Isn't there some way to compare everything between these printers?  Some configuration files somewhere that I can load into a text editor?

[buteomont 1]

I've added a screenshot of the time estimations for the FLSun and Tronxy, for comparison.  The difference doesn't seem to be related to any one aspect.  Surely there's a way to do a value-by-value comparison of everything between the two.

[GregValiant 1,251]

Travel time is a lot different.  Is Combing turned on for the Tronxy?

Have you un-hidden all the speed settings in Cura - because the "skin" times are near the same but the inner and outer walls are quite a bit different.

 

 

 

 

[buteomont 1]

Yes, combing mode is set to "All" on both of them.

I entered "speed" in the settings filter for both printers and all of the settings are the same or very close to the same. I've changed some of them to be exactly the same but the print time did not change much.  Here's a comparison screenshot.

 

[GregValiant 1,251]

There are a few Tronxy profiles so I just picked Tronxy_x.json.  In that profile the max acceleration is 120 for xyze.

M204 lines in the Gcode set the acceleration.  When I played with this, the "Travel Acceleration" and "Initial Layer Print Acceleration" didn't update when I changed the overall "Print Acceleration" in Cura to 3000.  They stayed at my default of 500.

 

Within the Gcode file, this resulted in an M204 S500 line right before every travel move and M204 S3000 right after a travel move.  I don't know if you've done this yet, but unhide all the Accel settings and take a look.  I'm guessing that "travel acceleration" is still at 100.

 

[buteomont 1]

I did check the acceleration and jerk times, and they were originally set to 100 like you said.  However, changing them all to 3000 (to match the FLSun) only decreased the print time about 9 minutes over a 5-1/2 hour print, which is nowhere near the FLSun's estimate of 3 hours and 5 minutes.

The specific Tronxy I am using is the model X5SA.

[Torgeir 242]

Hi buteomont,

 

I've see that the print quality is set to "normal" for the left and "draft" for the right, in my experience with Cura draft always print faster than "normal" quality.

Think there is a lot of underlying parameter differ in Cura between these various modes.(?)

 

BTW very interesting topic.

 

Thanks

Torgeir

 

 

 

[buteomont 1]

Yes, I chose the "draft" profile because that is the one with 0.2mm layers, the same as the "normal" profile on the Tronxy X5SA.  The Tronxy doesn't have a profile called "draft".  I wouldn't think that the name of the protocol would have any effect anyway.  I've been through every setting in those protocols that I can think of but still can't account for the huge difference in print estimates.

 

Surely someone out there in Cura-land knows where all of these settings are stored - it would be much easier to just compare two files for differences!

[Torgeir 242]

Aha. I've see what you mean..

Do you share a project file?

 

Well, you don't need any profile to select Draft mode, just select this quality (draft) in Cura then adjust for same speed etc.

 

Thanks

 

Torgeir

Edited October 4, 2020 by Torgeir
Addet text

[Torgeir 242]

Ok., Loaded your Tronxy X5SA into Cura (4.7.1).

 

As "draft" is kind of rough and fast print this will be close to like the low quality for your printer in Cura.  Try to use your speed, layer setting etc. in here and the printing time should differ.

 

Thanks

 

Torgeir

 

 

Edited October 4, 2020 by Torgeir
Corrected text.

[buteomont 1]

But that's the whole question, the print times are different!  I want to make them the same for all three printers.  Why are they different?

[tinkergnome 927]

3 hours ago, buteomont said:

But that's the whole question, the print times are different!  I want to make them the same for all three printers.  Why are they different?

 

Are the estimations of the CuraEngine different or the actual print times on each printer?

 

Be aware that different printers may be driven by different firmware (versions) with different limits for feed rate and acceleration.

There's not only a default value for those things, but also a maximum. For example: for Marlin firmware take a look at gcodes M201 to M203.

Are all printers in questions using the same motion system? (delta printers - e.g. - are a whole different topic regarding motion and speed calculations...)

There are probably much more firmware settings to consider, every printer has it's own capabilities and limits.

🤷‍♂️

 

If you're interested in how the estimation for Cura is calculated, take a look at the "TimeEstimateCalculator" of CuraEngine:

https://github.com/Ultimaker/CuraEngine/blob/master/src/timeEstimate.cpp#L18

 

 

Edit:

the configured maximum values for Tronxy printers are set very low (in comparison) - i guess Cura will respect these limits in all its calculations:

https://github.com/Ultimaker/Cura/blob/master/resources/definitions/tronxy_x.def.json#L25

 

Edited October 4, 2020 by tinkergnome

[buteomont 1]

Ok, now that makes some sense.  So if I set the acceleration values to, say, 3000, Cura will internally limit them to 120 and not notify me?  That seems kind of an unfriendly thing for it to do.  But that would explain a lot of what I'm seeing.

 

I just tried an experiment that verifies it - I set the printer to the Tronxy and loaded a model that purports to take 22:34 (with the FLSun it says about 10 hours) and changed the acceleration value from the default 100 to 120, and the print time dropped to 21:29, as expected.  Then I changed the values to 3000 to match the FLSun, and the print time did not drop any further at all.  There is no indication on Cura that my changes were being ignored.

 

Unless I can find a way around it, this is a real bummer. ☹️   Do you know of a way to override the acceleration value limits?

[gr5 2,172]

@buteomont - I think there is some confusion.

 

I believe Cura has 2 acceleration features that are mostly unrelated.  One of them is to calculate how long the job will take.  The other is what gcodes to create.

 

And your printer also has a built in max acceleration.  A 3rd feature but not in cura - it's in your printer.

 

So the values in the json file I believe are only used to calculate how long your print will take.

 

The acceleration values in cura should create gcodes that get sent to your printer.

 

You can verify this by setting acceleration to say 1234mm/sec/sec and then look for "1234" in the gcode file that is created.

 

Your printer also has acceleration values for each of the 4 axes (x,y,z,e).  Default values.  I think it might also have max acceleration values.  Your printer I'm guessing has Marlin firmware as 99% of printers under $10,000 have Marlin or a spinoff/fork of Marlin.  Many versions of Marlin let you change the default acceleration (and max acceleration) in the "motion settings" menu.  Many versions of Marlin don't and those versions you have to change it in Configuration.h file and rebuild Marlin.

 

Another test would be to print a hollow cube and put in the gcode that changes acceleration on each layer.  gcode files are easy to read and easy to edit.  Your ear can easily tell if it's printing faster or not.  At some point it will stop speeding up.

[gr5 2,172]

By the way there are is also a max velocity (on your printer) for each axis and there is also a "jerk" value for each axis.  "Jerk", in Marlin, is not the same meaning as jerk that physicists and wikipedia uses.  This is a very bad name because normally jerk means something else.  Jerk is in mm/sec which is a velocity.  Jerk is the maximum magnitude of instantaneous speed change at a vertex.  Each movement gcode is a new x,y,z,e position in 4 dimensional space and the printer makes a straight line to the next point with each gcode.  The idea of jerk is when printing say a circle, each change in direction is small so there is no need to come to a complete stop at each vertex.  Instead it slows down to the speed indicated by jerk.  The sharper the corner, the lower the speed.  If the corner is 180 degrees (reverses direction) the minimum speed (aka junction speed) is half the jerk value (e.g. 10mm/sec if jerk is 20).  If the corner is 90 degrees it will slow to 14mm/sec if jerk is 20.  If the corner is only 1 degree change (like points on a circle) the max junction speed is quite high and won't limit speed at all (something like 200mm/sec maybe).

[buteomont 1]

What I wound up doing is creating a new printer definition from scratch, and setting the values as in the original built-in Tronxy X5SA definition.  Now adjusting the acceleration values makes a huge difference, and I can bring the print estimates down to near the other two printers.  I believe the mystery is solved.

The actual printer must have some limits built-in as well like you said, because the actual print time is not quite as low as the calculated one.  It is, however, much faster than it was before and so far I can see no discernible problems with the print (although I've only printed one thing so far).  If I can figure out the firmware limits I'll include them in the printer definition.