Gyroid Infill Crashes V5

[jfkansas]

I setup a simple block that size on a stock cr10max profile and got the same crashes. Fwiw the v4 honeycomb did the same thing.

[wirlybird]

Ok, actually good to know.
I am running a very high end system so I wonder if this is GPU or an issue with S3D?

[DrD]

Sinusoidal also crashes it when set above 15%.
There are no odd settings or anything complex. simple straight forward print.
The crash takes place while it is "parsing build file".
The green progress bar gets to the end and it acts like it is about to display the resulting slice and the program closes.
I tried "wiggle" at 45% and it finishes but it does not display the resulting slice. The data is there but no image of the sliced part. The bed is empty.
It also crashes V5. I am running 501.

Here is a factory with a simulated part that represents the actual part.

I had a look at your .factory file and played around with the settings.

Keeping the infill @ 45% and keeping the Gyroid infill created a GCODE file that was over 2GB!!!! S3D v5 probably encountered
some kind of unhandled exception related to a buffer overflow of some sort... and that's why the visualizer "chokes" on the file
and S3D just closes (classic behaviour for unhandled exceptions in code).

From what I've seen in the GCODE file, Gyroid infill doesn't create arcs for the toolpaths but rather a series of short straight segments
that approximate the arcs of the Gyroid infill (hence the massive GCODE file).

Changing the infill pattern to Octahedral slices the model just fine (GCODE file is about 57MB) and the visualizer displays the
sliced model as well (business as usual). The print time however sits about about 185 hours of machine time (not sure if that's
what you're aiming for... however, it does seem to be a touch on the excessive side).

Not sure if you'd ventured down the path that I've described or not... Perhaps some of my "diatribe" will give you an idea or
two as to developing a strategy to get your model printed.

D.

[wirlybird]

Sinusoidal also crashes it when set above 15%.
There are no odd settings or anything complex. simple straight forward print.
The crash takes place while it is "parsing build file".
The green progress bar gets to the end and it acts like it is about to display the resulting slice and the program closes.
I tried "wiggle" at 45% and it finishes but it does not display the resulting slice. The data is there but no image of the sliced part. The bed is empty.
It also crashes V5. I am running 501.

Here is a factory with a simulated part that represents the actual part.

I had a look at your .factory file and played around with the settings.

Keeping the infill @ 45% and keeping the Gyroid infill created a GCODE file that was over 2GB!!!! S3D v5 probably encountered
some kind of unhandled exception related to a buffer overflow of some sort... and that's why the visualizer "chokes" on the file
and S3D just closes (classic behaviour for unhandled exceptions in code).

From what I've seen in the GCODE file, Gyroid infill doesn't create arcs for the toolpaths but rather a series of short straight segments
that approximate the arcs of the Gyroid infill (hence the massive GCODE file).

Changing the infill pattern to Octahedral slices the model just fine (GCODE file is about 57MB) and the visualizer displays the
sliced model as well (business as usual). The print time however sits about about 185 hours of machine time (not sure if that's
what you're aiming for... however, it does seem to be a touch on the excessive side).

Not sure if you'd ventured down the path that I've described or not... Perhaps some of my "diatribe" will give you an idea or
two as to developing a strategy to get your model printed.

D.

Thanks for the input.
One reason I am picking the infill types I am is they are not "straight line" infills. I have a theory that the straight line infills such as rectilinear create a great deal of stress or force on the walls of the print as the infill cools and contracts and it causes the corners and edges of the print to lift.
The part printed with 15% rectilinear does ok but increase to 70% and it lifts from the bed.

My most recent print was with Gyroid at 60% and it actually did ok and didn't lift from the bed. Had to use Cura since S3D would create the gcode. These are very long and expensive prints so reducing failures is key!

I know that the infill percents are excessive for most applications but the customer has given me very specific settings to use.
These are one-off parts for testing they are doing for a much bigger part of a machine.
Production parts will be molded Urethane. But when just one mold costs 40K we look to alternative ways to test designs.

[DrD]

Sinusoidal also crashes it when set above 15%.
There are no odd settings or anything complex. simple straight forward print.
The crash takes place while it is "parsing build file".
The green progress bar gets to the end and it acts like it is about to display the resulting slice and the program closes.
I tried "wiggle" at 45% and it finishes but it does not display the resulting slice. The data is there but no image of the sliced part. The bed is empty.
It also crashes V5. I am running 501.

Here is a factory with a simulated part that represents the actual part.

I had a look at your .factory file and played around with the settings.

Keeping the infill @ 45% and keeping the Gyroid infill created a GCODE file that was over 2GB!!!! S3D v5 probably encountered
some kind of unhandled exception related to a buffer overflow of some sort... and that's why the visualizer "chokes" on the file
and S3D just closes (classic behaviour for unhandled exceptions in code).

From what I've seen in the GCODE file, Gyroid infill doesn't create arcs for the toolpaths but rather a series of short straight segments
that approximate the arcs of the Gyroid infill (hence the massive GCODE file).

Changing the infill pattern to Octahedral slices the model just fine (GCODE file is about 57MB) and the visualizer displays the
sliced model as well (business as usual). The print time however sits about about 185 hours of machine time (not sure if that's
what you're aiming for... however, it does seem to be a touch on the excessive side).

Not sure if you'd ventured down the path that I've described or not... Perhaps some of my "diatribe" will give you an idea or
two as to developing a strategy to get your model printed.

D.

Thanks for the input.
One reason I am picking the infill types I am is they are not "straight line" infills. I have a theory that the straight line infills such as rectilinear create a great deal of stress or force on the walls of the print as the infill cools and contracts and it causes the corners and edges of the print to lift.
The part printed with 15% rectilinear does ok but increase to 70% and it lifts from the bed.

My most recent print was with Gyroid at 60% and it actually did ok and didn't lift from the bed. Had to use Cura since S3D would create the gcode. These are very long and expensive prints so reducing failures is key!

I know that the infill percents are excessive for most applications but the customer has given me very specific settings to use.
These are one-off parts for testing they are doing for a much bigger part of a machine.
Production parts will be molded Urethane. But when just one mold costs 40K we look to alternative ways to test designs.

Having worked in Electronics Design and Manufacturing for almost 30 years now, I totally understand the stringent
customer requirements and wanting a cost-effective methodology for prototyping prior to "jumping into" more costly
manufacturing methods.

On the flipside, have you tried to use a bed adhesive to assist in holding down the high-infill prints? One bed adhesive
that I've used with success is Dimafix. With your print bed being cold, you apply the solution from the bottle/dabber combo
then once dry (only takes a minute or so) heat up your printer as usual and print. Once the print bed has cooled down
(to about 30-35 degrees C), the adhesive auto-releases the part.

Cleaning the print bed afterwards is a snap as the adhesive is water soluble and you simply wipe it off when not needed.

D.

[wirlybird]

Yes, have tried PEI beds, Buildtak, and glass. I find glass with old fashioned glue works the best. Tried Magigoo also.
I also have the printer in a room where I can control the environment so no drafts and constant temp.
The material is TPU 98A which compounds the issue since the sides can flex more as the infill cools and that adds to the "peeling" effect at corners and sides!
Pain in the butt but for what I am changing I will do it and babysit 350+ hour prints which use about 3600g of filament.

[DrD]

Yes, have tried PEI beds, Buildtak, and glass. I find glass with old fashioned glue works the best. Tried Magigoo also.
I also have the printer in a room where I can control the environment so no drafts and constant temp.
The material is TPU 98A which compounds the issue since the sides can flex more as the infill cools and that adds to the "peeling" effect at corners and sides!
Pain in the butt but for what I am changing I will do it and babysit 350+ hour prints which use about 3600g of filament.

Wow... you've really "gone down the rabbit hole" on this one, for sure.

What about using a 0.6mm nozzle? Based on the sample .factory file you'd provided the print time dropped by almost 50%.

D.

[wirlybird]

Yes, have tried PEI beds, Buildtak, and glass. I find glass with old fashioned glue works the best. Tried Magigoo also.
I also have the printer in a room where I can control the environment so no drafts and constant temp.
The material is TPU 98A which compounds the issue since the sides can flex more as the infill cools and that adds to the "peeling" effect at corners and sides!
Pain in the butt but for what I am changing I will do it and babysit 350+ hour prints which use about 3600g of filament.

Wow... you've really "gone down the rabbit hole" on this one, for sure.

What about using a 0.6mm nozzle? Based on the sample .factory file you'd provided the print time dropped by almost 50%.

D.

The printer I use for this job I have just switched to a 0.6mm nozzle and running another PLA job on it to get settings dialed in.
I am hoping it may cut the time some. We'll see if I can ever get it to slice! Really don't care for Cura.

[DrD]

Yes, have tried PEI beds, Buildtak, and glass. I find glass with old fashioned glue works the best. Tried Magigoo also.
I also have the printer in a room where I can control the environment so no drafts and constant temp.
The material is TPU 98A which compounds the issue since the sides can flex more as the infill cools and that adds to the "peeling" effect at corners and sides!
Pain in the butt but for what I am changing I will do it and babysit 350+ hour prints which use about 3600g of filament.

Wow... you've really "gone down the rabbit hole" on this one, for sure.

What about using a 0.6mm nozzle? Based on the sample .factory file you'd provided the print time dropped by almost 50%.

D.

The printer I use for this job I have just switched to a 0.6mm nozzle and running another PLA job on it to get settings dialed in.
I am hoping it may cut the time some. We'll see if I can ever get it to slice! Really don't care for Cura.

Same here with regards to using Cura... klunky, disparate user interface. Settings buried everywhere. Ugh.

As a side note, I took the "sliced" GCODE file that S3D had created for your sample and tried to open it in Prusa's GCODE
viewer with 0 success. My RAM was quickly maxxed out (I have 16GB worth) and then the viewer just hung and stopped
responding. So no luck on that front, either.

Seems like everything that we try or come up with just fails outright. hehehe.

Hopefully the S3D devs are following this thread and figure it out...

D.

[jfkansas]

Sinusoidal also crashes it when set above 15%.
There are no odd settings or anything complex. simple straight forward print.
The crash takes place while it is "parsing build file".
The green progress bar gets to the end and it acts like it is about to display the resulting slice and the program closes.
I tried "wiggle" at 45% and it finishes but it does not display the resulting slice. The data is there but no image of the sliced part. The bed is empty.
It also crashes V5. I am running 501.

Here is a factory with a simulated part that represents the actual part.

I had a look at your .factory file and played around with the settings.

Keeping the infill @ 45% and keeping the Gyroid infill created a GCODE file that was over 2GB!!!! S3D v5 probably encountered
some kind of unhandled exception related to a buffer overflow of some sort... and that's why the visualizer "chokes" on the file
and S3D just closes (classic behaviour for unhandled exceptions in code).

From what I've seen in the GCODE file, Gyroid infill doesn't create arcs for the toolpaths but rather a series of short straight segments
that approximate the arcs of the Gyroid infill (hence the massive GCODE file).

Changing the infill pattern to Octahedral slices the model just fine (GCODE file is about 57MB) and the visualizer displays the
sliced model as well (business as usual). The print time however sits about about 185 hours of machine time (not sure if that's
what you're aiming for... however, it does seem to be a touch on the excessive side).

Not sure if you'd ventured down the path that I've described or not... Perhaps some of my "diatribe" will give you an idea or
two as to developing a strategy to get your model printed.

D.

Speaking of file size, sounds like it is hitting a file system size limit...