Large Scale 3D Printing

Does anyone here have experience with large format 3D printers like the Cincinatti BAAM at ORNL or the Thermwood LSAM? I've been working with smaller scale 3D printers for years, and have been keeping myself up to date with the desktop market, but I haven't come across many discussions about the super large scale 3D printers, especially anything getting into the more technical aspects of their workings. These machines look pretty awesome from the marketing videos on youtube, but I'm curious what they're really like to program and operate when the cameras are off. I know I've had my share of print failures on a much smaller machine, and there's been a ton of work that's gone into software and hardware to get printing to the level its at now. I'd love to get a conversation going about what it takes to print the kind of large parts these machines are capable of, especially around the use of screw extruders and pellets instead of using filament like all smaller FDM machines are using.

I did a knapkin calculation on BAAM - to keep it supplied with filament, you would have to change the spool about every 90 seconds or so.

Assuming conventional Reprap spools, of course.

One other thing I would say about that device, it would appear to have been made almost entirely from off the shelf components, albeit expensive ones. If that is true, it is not impossible to build a faster printer, using custom made components.

Largely a political stunt rather than being a commercial product. All my opinion.

I can see these printers having a number of commercial applications. Looking at Thermwood, they bundle their printer with a 5 axis CNC router that can mill parts to final size. Being able to use cheap pellets as feed stock would be awesome too. I remember there were a couple projects where people tried to make pellet fed extruders for a rep rap, but I never saw anyone post a successful design. I wonder how these large printers handle retraction and prevent oozing, as it would be tricky to control a high flow nozzle like that without creating lots of back pressure.

Hmmm... the smallest version of LSAM has 150lb/hr extrusion rate! A bit more than a standard 1kg spool a minute. If you buy ABS pellets by the ton, that's about $80 per hour for plastic.

I don't think they care too much about ooze... the philosophy is to 3D-print a blank that is slightly larger than the finished product, then use the CNC milling machine to trim it down to final dimensions. So a bit of extra plastic isn't really a problem.... more important is to make sure there are no voids.I suspect that a final manual fill-and-sand step would often be needed.

I know they like to use the carbon filled ABS, so that could easily add up to $750/hr @ $5 per pound. That's still a hell of a lot cheaper than carbon ABS filament which is anywhere from $50-$100 per spool, plus a machine like this would be making parts that would presumably cost a lot more if one had to machine them from a solid block.

You're right that ooze wouldn't be such a big deal when they're machining to size afterwards, however from what I can see in youtube videos, it still looks like they manage to control it fairly well, and I don't see any evidence of stringing either. For a project like the Shelby Cobra that ORNL printed, it looks like they printed slightly undersize, then used bondo to smooth over the layers, finished with lots of sanding and automotive paint. For a project like this I could see it being beneficial to have a decent quality print to start off with, as it would be incredibly time consuming to run a surfacing program with a ball nose cutter to finish the part. Looking at Thermwoods specs for the LSAM [thermwood.com] they mention a precision metering system to precisely control the bead. Obviously a servo driving the screw would give precise feedback, but I'd be curious to know how they would control hysterisis as there would be a lot of backpressure that could build up in a nozzle that can spit out 150 lbs an hour.

With large nozzle sizes, you could easily make a cap or valve that closes off the nozzle during travel moves.

I've been doing some more research and it was mentioned that in some cases a heated bed is used. Obviously this would require a significant amount of power, but I'm curios if this would be worthwhile. From my experience with a heated bed on a much smaller printer, it helps a huge amount with adhesion and eliminates warping for the first several layers, but when printing large, tall parts there is still a high likelihood of warping if the design of the part leads to thermal imbalances. With custom programming and a thermal camera it could be possible to visualize thermal stress as the part is printing and possibly compensate, but that sounds like a very finicky system at best. It would be a lot more reliable to find the right combination of material and machine settings that would allow printing on an unheated bed without warping.

I also found some more info on how ooze can be dealt with. Apparently instead of making a quick move from the end of one path and sending a retract command like a filament fed machine does, BAAM uses modified G code similar to enabling coasting in simplify 3D, as well as a helical move to break off any stringing before jumping to the next start point.

Heating the bed only serves the purpose of getting the first layer to stick. It has nothing to do with controlling warping. Warping is a function of the material, print temperature, and object geometry. ABS is printed in warm, enclosed printers to prevent delamination and warp due to stresses created by the cooling, shrinking plastic. Without a heated enclosure you can print very small ABS parts or single walled vases, but if the object has any bulk it will delaminate once it gets a few cm tall. A thermal camera will tell you what you already know- the layers away from the bed cool off rapidly. If you want to control warping, enclose the printer and heat the enclosure. I have found 45-50C sufficient to prevent warp and delamination in ABS prints- in my printer the bed heater alone is sufficient to get the enclosure that warm. Stratasys uses 70C in their older printers; I'm not sure about newer ones.

Stratasys prints on unheated plastic bed plates inside a heated chamber. In earlier printers they used an unheated urethane foam bed. The prints didn't warp because the enclosure was heated. If you want an unheated bed for a large printer, try using PIR foam insulation board. It is readily available, very cheap, and it works. Any material you print will stick to it because the extruder pushes the plastic into voids in the foam and forms a mechanical bond.

I suspect other print materials would benefit from a warm enclosure too, because they all shrink as they cool.

---

Ultra MegaMax Dominator 3D printer: [drmrehorst.blogspot.com]

Quote
750magna
I also found some more info on how ooze can be dealt with. Apparently instead of making a quick move from the end of one path and sending a retract command like a filament fed machine does, BAAM uses modified G code similar to enabling coasting in simplify 3D...

Enabling pressure advance in firmware should be even more effective, because instead of just coasting at the end of a decelerating move, it actually retracts before the end of the move if the product of deceleration and pressure advance time is high enough. At least, in RepRapFirmware it does, I don't know about other firmwares.

Quote
750magna
... as well as a helical move to break off any stringing before jumping to the next start point.

That sounds interesting! I wonder how effective it is? It's something we could add as an option in firmware retraction, assuming it hasn't been patented.

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].