Large-Scale FDM - Persistence and Failures Are Standard Features

[C855B]

Nearly everybody here has already seen the Weekend Update photo of the great 3D-rendered UP Kelso station my friend Craig produced for the layout:



Now I knew going in it was a significant challenge in any 3D output method. The files were quite large, with all but one of the original .DXG files so complex they could not be converted to any hobbyist-grade CAD app, only editable in the original software.

It now seems like aeons ago, but it was just this past Thanksgiving weekend (at a local train show) that Craig and I were talking at length about his FDM projects. He was working on a switch machine mount, knowing that I had taken an unusual approach on my layout. That conversation morphed into the Kelso station .STL files, and he asked if he might give them a try. I happily obliged. Two weeks later, he sends me this pic:



Not quite model quality yet, but a surprisingly good proof-of-concept, good enough that I think it's currently living on a mutual friend's layout. He had his eye on a new large-format printer kit (https://folgertech.com/products/ft-mega-i3) and we agreed I would chip-in to help with upgrades and materials to execute the station artwork in a higher resolution.

Next thing I see is this, in late January:



OMG, we can use that! IIRC, he was still working on the colonnade, and briefly mentioned the building core might be tough to do as a single piece. Little did I know just how hard:



The colonnade was no piece of cake, either:



Bottom line, there is a whole bunch of machine time and boxcar-loads' worth of PLA filament in those piles of rejects. Unfortunately, the extreme machine time evident here plays into the large amount of waste. It has to be a "start it and go off and do something else" process, so there is little opportunity to stop the print when it goes awry, you just get to come home to a big mess.

Truthfully, if I had had any idea how much difficulty there was in rendering the building core I would have waved-off on that part of it for execution on my Cameo cutter with styrene sheet. It is simple, amounting to four flats with window cutouts, and the sills could have easily been part of the window inserts now on my platter. As it was, I had to trim on two other pieces for the final fit because the core came out roughly 0.020" too long. No big deal. Colonnade and its roof, however, were only reasonable in 3D, and I'm certainly glad Craig stuck with it.

Rodney @rodsup9000 inquired if Craig was using an oven or insulated box around his open-frame printer, and Craig said no, the heated table provided the base adhesion and further heat management wasn't necessary with PLA filament medium. He did, however, observe that ABS is another ballgame, and that even just a 10°C increase in ambient reduces the effects of ABS' in-process dimensional changes.

[AlwaysSolutions]

Ah yes, the graveyard of failed prints - I'm all too familiar with it from my 2nd generation FDM machine.  It's funny that years ago the YouTubers didn't show you that aspect of the hobby.      Heck, for all I know it's still a problem that nobody talks about even with these seemingly bullet-proof printers like the Prusa i3 MK3?   For this particular station, I don't know if it could be done, but generally if you can break a job into parts you can minimize waste, as a failed print would involve a much smaller component of the whole.  You could maybe play off your idea with the Cameo - print the slabs and assemble.  Spray on some textured paint and perhaps the tell-tale print marks would vanish?  That's a sweet looking model, it's going to be amazing when it's all finished.

Mike

[C855B]

John @Lemosteam and I talked about chopping the project into manageable chunks, but I hesitated on that due to the difficulty of maintaining the contiguous look of the Spanish tile across separate pieces. It would be really tough to make the join invisible given that it's basically a texture and print-to-print tolerances come into play.

But, yeah, doing it again, the core would be better done as flats either with a vinyl cutter or even on the FDM printer.

While I'm thinking about it, my years in the high-resolution 2D business (printing industry) were prescient here. Customers didn't comprehend why we had to charge so much, and they couldn't (or wouldn't) understand that anything greater than simple would need multiple tries for acceptable output. The retry requirements made it a horrible business model, and I quickly burned out on 16-hour days trying to make "no problem for us!" files work on our equipment. They just didn't get that 2400dpi on negative film was an entirely different ballgame than their little 300dpi desktop printer.

When we added an early hi-res 4-color machine to our capabilities, any hopes that I would have a home life went away entirely. Those machines would crash if you looked at 'em cross-eyed.

[rodsup9000]

 

  I know about reject piles. I've wasted more rolls of PLA than I care to admit. I need to find more about what firmware and slicer Craig is using. When I built my first printer, I designed it to use imperial acme threaded rod, instead the usual all thread. Then when I enlarged the design to 12" cube, I changed to precision metric acme threaded rod. It was about this time that the GT-2 belting and pulleys were introduced and I changed everything over to them from MXL belting. I'm still running Marlin 1.0 firmware and the original slic3r slicing software. I'm afraid to do any upgrades in fear I will not be able to figure out all the correct settings. I guess I could try it on one of the printers that's in the shed that I haven't used in a few years.     

[C855B]

I've sent Craig a link to this thread and hopefully he can directly answer your software question.

[rodsup9000]

I've sent Craig a link to this thread and hopefully he can directly answer your software question.

 Thanks, Mike

[johnb]

Oooohhhh......these arches.....are they N and do you want them?

[C855B]

(Hang on, guys. Craig's waiting for account approval and he will reply once he gets the okey-dokey from Tom and the crew.)

[CRudeME]

Hello! Finally got approved!

@rodsup9000 - I am currently using whatever Marlin firmware shipped with my printer with Simplify 3D. Save your known settings/firmware versions and do not be afraid to try newer slicers! I moved from Cura to S3D and tested Cura again because I was curious. I was impressed with the updates! When it comes down to it, the only thing that matters is getting a model printed the way you are happy with.

Both of FDM machines have been rebuild many times. 5 times for my larger printer and I am on round 7 for my smaller printer. Sometimes, that requires a firmware/software change. The more familiar you get with these tweaks, the easier it will be to adjust settings for your printer.

@johnb yes, the arches are N, and I am uncertain what would be best suited for them

[rodsup9000]

I am currently using whatever Marlin firmware shipped with my printer with Simplify 3D. Save your known settings/firmware versions and do not be afraid to try newer slicers! I moved from Cura to S3D and tested Cura again because I was curious. I was impressed with the updates! When it comes down to it, the only thing that matters is getting a model printed the way you are happy with.

  First, I have to say you did a hell of a job on the station. It look awesome.

 Thanks Craig, I think I'll load the latest Marlin and try it and hope I can get everything in the settings. The one area I might have problem with is the display settings.
I'll give S3D a try and see how well it works for me.

Both of FDM machines have been rebuild many times. 5 times for my larger printer and I am on round 7 for my smaller printer. Sometimes, that requires a firmware/software change. The more familiar you get with these tweaks, the easier it will be to adjust settings for your printer.

 What are you doing on the rebuilds???  After building over 25 from scratch without changing much from the first, I don't see where I can improve on mine at all. The 12" that I'm using, has 2 to 3 thousand hours (guessing in the 7 years since I built it) It all aluminum frame with delrin wheels.

 The first 2 photos was when I first built the first 12" and it's the one I still use the most now. I built 14 of 12's when I was building them and sold 10 of them.



     












This is one of the first things I printed after calibrating it.









 


 I have a 8" that has a .02mm nozzle that I use for things like N scale windows. I can print cobweb thin with it.

[johnb]

@johnb yes, the arches are N, and I am uncertain what would be best suited for them

I was thinking about using the scraps for a structure....

[CRudeME]

@rodsup9000 My apologies for taking so long to get to answering your question - the 24 x 42ft N layout move took over everything. @C855B started a thread which shows some photos from the project.

As for you question on what I've done on the rebuilds:
-The entire hotend assembly was replaced with an E3D Titan Aero. I've got a bunch of parts coming to do a rebuild on both hotends due to an "oops" moment of snapping the heat break off and noticing my replacement was bent.
-Belt tensioners were designed/added. I loathed having to remove and reattach belts because I always found myself snapping a printed part off of the carriage, or never being able to get the belt in the +1 tighter spot since loosening. An M3 screw and a captive nut fixed that issue!
-Z axis accuracy was tweaked and is also still developing. The vertical supports were held in place in a similar fashion to yours, however there was still much opportunity for play in the y axis. I ended up using a test print tower to figure out how perpendicular the prints were being executed. Ultimately, I will be having some custom side plates CNC cut to remove any guesswork with the vertical axis.
-Both printer build surfaces are the Buildtak Flexplate system. Each has two spring steel sheets, so I can do a hot-swap when I'm bulk printing items that print great on their own, but don't play well with others. Also, it is always handy to have a sheet with a fresh surface ready to go whenever you have a catastrophic failure, or a sheet is warn too much.
-Larger bolts/higher tension bed springs. On the i3 Mega, the CNC aluminum heated bed is designed for M3 mounting hardware with thumb screws and tiny, low K springs. As the bed moved and printer shook, I would find these little thumb screws all over my workbench, and sometimes even the floor! To fix this issue, I drilled out the holes, added just a regular nut and M5 (I believe) cap screw. Adjusting the bed is a little trickier now, but the much higher tension on that nut keeps it from wavering much, if at all.

However, now I have a BLTouch that has been added to the i3 Mega printer. After a bit of Marlin calibration, I was able to get the printer to a much more repeatable. For rough calibration of the build plate, I put together a spreadsheet which takes the BLTouch mesh data and tells me what screw to loosen and tighten to get everything nice and happy. Just before my printer failed, I was able to get the probe repeatability down to about 0.01-0.005mm.

Once I get photo posting access, I will surely post the rig I used to print the station, as well as field any other questions!