SoliForum - 3D Printing Community — Cura estimated print time inaccuracy

Re: Cura estimated print time inaccuracy

2015-04-02T02:29:44Z

That makes me wonder if RH does its own computation instead, which certainly wouldn't be hard for them to do.

Re: Cura estimated print time inaccuracy

2015-04-02T02:20:44Z

In my experience, Cura can be notoriously bad in predicting times. As in 2x-4x the estimate to actual time. With that being said, two things. 1. Supposedly going from 13-14 or 14-15 they rewrote their algorithm and it was supposed to improved. 2. The two times I have used it with Repetier Host it was almost spot on to the minute.

Re: Cura estimated print time inaccuracy

2015-04-01T21:48:13Z

I'm a graduate student so I can work from home when I want to - but if I had a print ever give me that long of an estimation, I'd probably throw on my Volcano. I've put a lot of work into upgrading things so I can print accurately at higher speeds, too (thanks to wardjr for all of his help on that ) Without bearings on the back rod I print at 125mm/s reliably with exceptionally good repeatability. When I (some day) get bearings back there, 300mm/s travel and 150mm/s print speeds should be achievable if I can melt the plastic fast enough, but I have some issues with this already at 125mm/s solid fill. I also tend to print more functional parts so things like ringing aren't really a concern for me as aesthetics don't matter.

You'd be amazed at the speed improvements you can achieve with just a larger toolpath. My 0.6 nozzle substantially reduces print time without increasing speeds.

Re: Cura estimated print time inaccuracy

2015-04-01T21:26:32Z

jagowilson wrote:

It's an extremely important feature if you take full-time attended printing seriously. I think the last time I used Simplify3D is when it estimated a 5 hour+ job at 3 hours. I wasn't too happy having to cancel that print when I had to leave...

You actually babysit all your prints? Have you ever tried a long print, I mean something in excess of 10 hours?

I routinely do 12-15 hour prints. The longest print I did once, ended up at 33 hours.

Re: Slic3r, it's accurate within a few seconds to a few minutes (depending on print time). The longer the print is, the more it deviates. But even it needs a fudge factor of +10%, which seems to be the RH default anyway.

Re: Cura estimated print time inaccuracy

2015-04-01T20:28:00Z

That's strange. I have mine all set at 0 in Slic3r (1000 XY acceleration in the firmware) and it is accurate to within +/- 3 seconds regardless of fill type, fill density, layer height, first layer speed, nozzle diameter, ...

pirvan wrote:

So, do you use Cura to slice? If so, how close are your estimates to the actual completion time?
Also what value do you have in RH Printer Settings/Printer Tab/Add to comp. Printing Time?
I now have 42%, but it still varies, just not as bad as before, but it still can be off by 10-20 minutes depending on total print time.

I don't use Cura, but my suspicion is still that Cura's estimation does not properly account for acceleration. I had this issue with Simplify3D - the print time estimates were way off, in both directions, and there was no way to adjust its estimation. This might have changed, but I stopped using the software in favor of Slic3r 1.1.7. Slic3r is just plain good at estimating, and I'm surprised others haven't just copied their algorithm for time estimation. It's an extremely important feature if you take full-time attended printing seriously. I think the last time I used Simplify3D is when it estimated a 5 hour+ job at 3 hours. I wasn't too happy having to cancel that print when I had to leave...

Re: Cura estimated print time inaccuracy

2015-04-01T14:14:27Z

jagowilson wrote:

I've never attempted to alter slic3rs acceleration values before. I checked mine out and they're all 0 - I'll have to play with that and see how the emitted GCode changes and whether or not the firmware obeys them.
Your acceleration values are the same as mine - all default.

I was seeing actual print time 2X predicted when I had mine all 0 using honeycomb infill so I am thinking the fw does obey them. Now I have them all at 1000 except perimeter which I reduced recently to eliminate bounce artifacts. I can tell, watching & listening to the print that it is working as advertised (perimeters have a distinct Weeeee-ewwwww, Weeeee-ewwwww sound I don't hear on the other components).

Re: Cura estimated print time inaccuracy

2015-04-01T04:11:55Z

So, do you use Cura to slice? If so, how close are your estimates to the actual completion time?

Also what value do you have in RH Printer Settings/Printer Tab/Add to comp. Printing Time?

I now have 42%, but it still varies, just not as bad as before, but it still can be off by 10-20 minutes depending on total print time.

Re: Cura estimated print time inaccuracy

2015-04-01T02:47:06Z

I've never attempted to alter slic3rs acceleration values before. I checked mine out and they're all 0 - I'll have to play with that and see how the emitted GCode changes and whether or not the firmware obeys them.

Your acceleration values are the same as mine - all default.

Re: Cura estimated print time inaccuracy

2015-04-01T02:42:34Z

jagowilson wrote:

The slicer can't take into account the firmware's values unless the host reads the eeprom and sends them via command line input - I'm fairly certain this doesn't happen. What are you using for acceleration in EEPROM? The default XY acceleration in Marlin is 1000 mm/s^2. I'd check to make sure yours isn't any lower than this, as this is quite slow.

Here are the EEPROM settings:

19:27:23.805 : echo:Maximum Acceleration (mm/s2):
19:27:23.807 : echo: M201 X3000 Y3000 Z100 E3000
19:27:23.807 : echo:Acceleration: S=acceleration, T=retract acceleration
19:27:23.808 : echo: M204 S1000.00 T2000.00

Slic3r uses it's own acceleration settings:

Perimeters: 1000
Infill: 2000
Bridge: 2000
First Layer: 1000
Default: 2000

Re: Cura estimated print time inaccuracy

2015-03-31T18:54:09Z

The slicer can't take into account the firmware's values unless the host reads the eeprom and sends them via command line input - I'm fairly certain this doesn't happen. What are you using for acceleration in EEPROM? The default XY acceleration in Marlin is 1000 mm/s^2. I'd check to make sure yours isn't any lower than this, as this is quite slow.

Re: Cura estimated print time inaccuracy

2015-03-31T18:24:20Z

TickTock wrote:

That was my problem with RH/Slic3r. I had left the acceleration at 0 (the default). Once I changed it to 1000, I was much closer to the prediction.
http://www.soliforum.com/topic/7799/sol … g-is-slow/

I can't seem to find an option to input acceleration values in Cura, which leads me to believe, they use the values built into the firmware, or they don't use acceleration at all.

Re: Cura estimated print time inaccuracy

2015-03-30T03:35:37Z

That was my problem with RH/Slic3r. I had left the acceleration at 0 (the default). Once I changed it to 1000, I was much closer to the prediction.

http://www.soliforum.com/topic/7799/sol … g-is-slow/

Re: Cura estimated print time inaccuracy

2015-03-29T16:00:46Z

Could it be that their estimation algorithm assumes different (faster) acceleration? Or worse, doesn't consider acceleration at all?

Cura estimated print time inaccuracy

2015-03-29T15:05:10Z

Does anyone else have problems with Cura's estimate print time?

to say it's inaccurate is actually a misnomer. It's so far off sometimes it's not even funny.

For example I just finished printing a model that was estimated at 13:34:48. Actual print time was 17:26:48. That's almost 4 hours difference.

Slic3r was always under estimating, so I added about 10% to the time of completion in the printer settings, and it became a lot more reliable, print time completions were only within a few minutes of estimated time.

The 13:34:48 estimated time that Cura gave in the example above, included the 10% add-on it inherited from Slic3r. I had to change that to 42% to get the ETA close to the actual time.

Does anyone know why the estimations are so far off? Other than the fudge factor in the RH printer settings, is there something that we can tweak inside Cura?