• Welcome to The Forum for Gauge 3 Model Trains.
 
The Gauge 3 Society       2.1/2 inch Gauge Association       Cookies and privacy HOW TO JOIN: to request forum membership please click here

Gauge 3 Society members must be logged in to view the Society section
  G3 Clubroom

Welcome to the G3 Clubroom. This is the friendly online forum where members share ideas and inspiration, suggestions and advice, modelling tips, pictures and drawings, and general chat about our fine hobby of Gauge 3 railway modelling. A warm welcome, and enjoy your visit here today.

Complicated Software

Started by 753, Mar 30 2021 10:03

« previous - next »

0 Members and 1 Guest are viewing this topic.

John Candy

This axlebox is of the type used on Collett's plate steel (as opposed to pressed steel) bogies.
It should be used with the 7ft bogie shown in previous post.

The other GWR axlebox design, which I provided earlier, is suitable for the "American" bogies.

http://gauge3.info/openscad/GWRCollettplatebogiesaxlebox.scad

John.
My fellow Members, ask not what your Society can do for you, ask what you can do for your Society.

John Candy


Since I have posted no new SCAD items for several days, you may have thought either my machine or I had given up the ghost.
Far from it .... I have been giving the keyboard a good bashing, while the Snapmaker has been pounding away almost non-stop!  The first things to "give out" are likely to be my computers.... but more on that in a moment.

Recent focus has been on the GW milk train project ("Calcutta" needs a purpose) so files have been prepared for Siphons of types O5(6-wheel), "C" (4-wheel) and bogie Siphons "F" and "G" (the outside-framed O11 diagram). The bogie frames from last week will now have a use.

The items printed so far are sufficient vent louvres for a complete Siphon"G" plus a couple of other frame bits, including one door module (8 are required and they take 7+ hours each to print).









Now to the computer issues.... I wondered whether my computer could be speeded up by adding extra memory since the files for the door modules and framework were taking 30 minutes each to render (the desktop PC is Dell Inspiron with Intel Core i3 CPU with 8GB) . I spent nearly 2 hours just rendering the doors for the "C", "F" and "G" versions and each file revision took another 30 minutes. So I spent long periods just waiting for the computer to finish before I could resume work on scripting SCAD files.

I don't like using laptops for this sort of work but I wondered whether my HP laptop with Intel 8th Gen Core i5 and 12GB would be quicker. No, it took almost identical time for the "C" door file.

I checked on the performance and the CPU was working only at 28 percent capacity of which OpenSCAD was using just 25 percent. Memory was running at 57 percent, with OpenSCAD using just 30 percent.
Current consumption was in the red sector, "Very High", and the fan was blowing out very hot air.... the battery dropped from fully charged to 50 percent in just 45 minutes.

Clearly, adding memory or upgrading the CPU on the Dell was not likely to improve matters.

I know there are large amounts of calculation involved in rendering the SCAD files but what is holding back the processing?
The processing log is copied below to provide an insight as to what was involved in rendering the Siphon "C" door file.
One line which may provide a clue is "Volumes : 2".
The CPU has 8 cores and OpenSCAD was using just 25 percent of capacity. Could this mean it was using just 2 of the 8 cores? If that is the case, is there a way of making it use more CPU capacity?

Regards,
John.


Loaded design '//synologyds115j/Freecom Backup/GRESLEY coaches and other 3D parts/GWR Siphon C door.scad'.
Compiling design (CSG Tree generation)...
Compiling design (CSG Products generation)...
Geometries in cache: 66
Geometry cache size in bytes: 1658632
CGAL Polyhedrons in cache: 0
CGAL cache size in bytes: 0
Compiling design (CSG Products normalization)...
Normalized tree has 99 elements!
Compile and preview finished.
Total rendering time: 0:00:00.118

Parsing design (AST generation)...
Compiling design (CSG Tree generation)...
Rendering Polygon Mesh using CGAL...
Geometries in cache: 203
Geometry cache size in bytes: 27938720
CGAL Polyhedrons in cache: 8
CGAL cache size in bytes: 1193248
Total rendering time: 0:26:37.462
   Top level object is a 3D object:
   Simple:        yes
   Vertices:    90406
   Halfedges:  505996
   Edges:      252998
   Halffacets: 325256
   Facets:     162628
   Volumes:         2
Rendering finished.
My fellow Members, ask not what your Society can do for you, ask what you can do for your Society.

cabbage

The problem may not be the computer but rather the operating system that it is using... I presume you are using Windows 10 of some version?

The time may have come to investigate Debian Linux...

Regards

Ralph

John Candy

Ralph,
Yes, both the Dell and HP are running the same (latest update) Windows10.

The Dell has just set a new personal best for rendering a SCAD file ....1hr 22mins 50.935secs!!
That is almost three times the previous record (30 minutes for the door module).

The file was the end for the "Siphon G" .... which the Snapmaker is telling me will take 18+ hours to print.
By the time I have printed all the parts for the sides and ends of one Siphon G, the printer will have been working for something in the region of 170 hours!












Regards,
John.
My fellow Members, ask not what your Society can do for you, ask what you can do for your Society.

John Candy

The Siphon G end has finished printing and the second end is under way (with 16 hours to go).

The completed end is shown below (the British Standard gangway connector, of the "scissors" pattern, will be a separate item to follow).






When the final few parts have been printed (still a couple more doors to do at 7 hours each) and the body has been assembled and proven, then the entire set of files will be made available for you to "print your own".

Regards,
John.
My fellow Members, ask not what your Society can do for you, ask what you can do for your Society.

John Candy

Printing of the parts for both sides and ends of a Siphon G is now complete.

The components for one side are shown as separately printed and then loosely placed together to form a complete side (the sides will fit between the ends, the corner posts being part of the end castings).

The printed components will be glued to an internal plywood frame to provide strength and integrity and handrails and other minor detailing then added.

John.



My fellow Members, ask not what your Society can do for you, ask what you can do for your Society.

MikeWilliams

Impressive John.  Really nice.

Mike

753

John

Excellent work, should make an impressive model.

I don't think I would have your patience to wait all those hours to print.

Mike

John Candy

Quote
I don't think I would have your patience to wait all those hours to print.

Mike, The good thing is that I can be doing other jobs while the Snapmaker is churning out parts .... 24 hours every day!
The programs for the Siphons C and F were scripted while parts for the "G" were being printed.
If I decide I want more than one of each, I now only have to press a button.

The side for the "F" has been printed and is shown in photos below. To simplify assembly, I have combined the louvres with the framing, something I wish I had done with the "G".

The parts for the Siphon C sides are now printing, the only difference from the "F" being the width of the outer end framing (the doors and inner framing are same dimensions as for the "F").

Regards,
John




My fellow Members, ask not what your Society can do for you, ask what you can do for your Society.

John Candy


The 4-wheeled Siphon C is the final louvred Siphon I shall be building (the Siphon H being left out, at least for the present) and the remainder of the milk train will be composed of the 6-wheel types with open-slatted sides.

There were differences between early and later-built Lots of the Siphon C. Mine will be an early one with 8ft width (later Lots were 8ft 6ins wide) so I can use the same end as I am scripting for the 8ft wide Siphon F.

Incidentally, for anyone intending to construct the Siphon G, the sides I have produced are for the later Lots which had 9 inch solebars and angle iron trusses. Early Lots had 10 inch solebars with queen posts and bars or rods for trusses and the doors were different, being fitted within a bottom step plate (the later Lots had doors which opened down to the base, omitting the bottom plate).

John.



My fellow Members, ask not what your Society can do for you, ask what you can do for your Society.

John Candy


My machine has now clocked up several hundreds of hours, working through most nights.

Today, I had a string of events which caused the machine to abort prints and display "completed" message when only part way through the print. At first I thought it was a software glitch but after three re-boots the same file aborted at different points in the printing process, the final being 4 hours into the print (the previous two had aborted within tens of minutes of the start).

The consensus on the Snapmaker forum is that this is caused by the over-heat protection cutting in to prevent a fire, should the print head get too hot and ignite the filament. The thermistor is thought to be the cause, so I have fitted a new "hot end" and the machine is currently plodding along.

If it was a thermal runaway situation, then I at least know that the fail safe works.

The removed hot end is shown in attached photos together with a Siphon F door which aborted after 8 percent and another after 46 percent (latter interesting as it shows the internal structure of the print).

Is it time to have a dedicated 3D printing board?

John.





My fellow Members, ask not what your Society can do for you, ask what you can do for your Society.

cabbage

I have been "printing" since 2015. The machine is very reliable and typically a thermistor will last me about 3 months before "Abort Error 2" appears on the display.

I consider them to be "consumables" like the TC index tips on the cutters of my lathes. I have a store of them and I can swap one out in a few seconds after the head has cooled.

3D printing is becoming important in our hobby as it now produces what *I* need and the files can be used by others to make what *they* need.

A "makers" forum would be useful. But, mostly I would like a forum on lathes and mills. The only lessons I ever got on them were from my father in Bulawayo Shed on the Sunday Shift. Transferring the mechanics and operation of machines made by Cincinnati Milicron to those made by Sieg...

Regards

Ralph

John Candy

Two more computer-crippling files completed : First ,the end for the Diagram O6 Siphon (the last of the 6-wheeled designs) and the other, with outside framing, is for the Diagram O7 Siphon F and Diagram O8 (8ft wide)Siphon C.

The open-slatted sides for the six-wheeled Siphons of Diagrams O1/O2/O3/O4/O5 and O6 are being die-cut on the "Cricut Maker" with ends,doors and side framework being 3D printed.

The final photo shows a die-cut side for the O5 and O6 Siphons. It is cut from thick card, which has been soaked in Ronseal wood hardener which leaves it like plastic card (wood hardener is styrene in a solvent).

John.







My fellow Members, ask not what your Society can do for you, ask what you can do for your Society.

753

John

The detail on the end prints is impressive also a smooth finish, is this down to your particular printer?

The reason I ask I have seen prints with lines where the layers have been printed, some people use acetone to smooth prints.

Mike

Doddy

Quote from: 753 on Jun 07 2021 09:42
John

The detail on the end prints is impressive also a smooth finish, is this down to your particular printer?

The reason I ask I have seen prints with lines where the layers have been printed, some people use acetone to smooth prints.

Mike

I'll second the request and also ask that you post higher resolution photos of your work, these small ones are too small for my ageing eyesight.

Robert
"You don't know what you don't know"