Inverted FFF printer

I'm thinking of building a large inverted FFF printer, where the print head points upward and moves in X & Y on a fixed base plate. The build surface (business-side down) with then rise above that. The two reasons I'm thinking of doing this are: 1) All of the high speed components that are liable to cause structural vibration are intimately attached to a very rigid and heavy base plate (I'm thinking of using an 18x18" granite tile) that is then sitting on a fairly rigid work table. Thus, very little vibration will be imparted to the rest of the printer. 2) with a larger diameter nozzles (say, 0.6mm), gravity won't be causing the molten filament to ooze out of the nozzle.

The drawbacks that I see are: 1) sub-par first layer adhesion could cause the print to fall onto the print head, making a mess. 2) air printing would leave all the errant extruded filament falling back onto the nozzle, potentially encasing it in a solid blob of plastic. 3) You can't easily see how your print is progressing. All of these problems could be mitigated with some work, but is it worth the effort?

Does anyone see any other problems or benefits that I'm missing?

Support structures would have to be done differently, you can't count on plastic falling onto them for a looser bond. Gravity will cause semi soft plastic to
sag down towards the print head in addition to the curling effect that can happen with a traditional setup. It would be interesting to try, but I worked through these and a few other downsides before coming to the conclusion it wasn't worth it for me to try.

Quote
Koko76
Support structures would have to be done differently, you can't count on plastic falling onto them for a looser bond. Gravity will cause semi soft plastic to
sag down towards the print head in addition to the curling effect that can happen with a traditional setup. It would be interesting to try, but I worked through these and a few other downsides before coming to the conclusion it wasn't worth it for me to try.

In general, the gravitational effects on printing are relatively small and there are plenty of videos showing successful inverted printing. I'm wondering, though, how many "special" cases there are where things would go horribly wrong. Has anyone tried printing a lot of stuff upside-down?

How can you print support material for prints that need it? If you print it close enough to the print layer that it will work to provide support, you won't be able to separate it from the print.

---

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

Quote
the_digital_dentist
How can you print support material for prints that need it? If you print it close enough to the print layer that it will work to provide support, you won't be able to separate it from the print.

Supports don't really work by gravity - they mostly work by weak adhesion between the support layer and the object layer above it. When you squirt filament out of a 0.4mm nozzle the resulting thread is actually a little wider, maybe 0.55mm dia. Therefore, when you skip a, say, 0.2mm layer above the support layer, the next layer, which is 0.4mm above the support, will still actually contact the support layer. Also, even with gravity pulling in the opposite direction, filament coming out of the nozzle will squirt straight for a fraction of a mm before succumbing to gravity, again causing contact with a more distant support layer. No doubt, the clearance between the supports and object will have to be tweaked, but that can be handled with slicer parameters.

Sounds really interesting. I hope you make it as I'd love to see how it works out. I've seen the video of a printer working upside down (one of Nicholas' designs: [www.youtube.com] ) so I know the basics work, but I'm curious to know where the edge cases might lie. As you said, I'm guessing the issue is that when things go wrong stuff will head towards the hotend instead of away from it. Bridging might be troublesome for that reason.

Edited 1 time(s). Last edit at 05/19/2016 01:08PM by JamesK.

Quote
JamesK
As you said, I'm guessing the issue is that when things go wrong stuff will head towards the hotend instead of away from it. Bridging might be troublesome for that reason.

Oh yeah, I hadn't though much about bridging - this is one area where gravity does predominate. One interesting possibility with an inverted printer, though, is to use the fan to blow the filament upward, approximating zero G. I guess nothing to do but build one and see. Worst case, I can hang the whole thing upside down.

I don't think it's a good idea, did you consider what will happen if a print falls on the extruder? My guess is that it will melt, ruin your extruder, and will catch fire.

Quote
Druwan
I don't think it's a good idea, did you consider what will happen if a print falls on the extruder? My guess is that it will melt, ruin your extruder, and will catch fire.

Yes, the print falling on the extruder would make a mess, but at <250deg.C, nothing will actually be ruined or catch fire (other than the print). I've been mulling over making a shroud/duct to cover up most of the extruder body with just the nozzle tip exposed through a clearance hole. I might also make a simple optical sensor to detect anything falling on the extruder head and wire that into the e-stop.

Getting back to the original post, what sort of problems do vibrations cause in uninverted printers, and is it gravity that causes extruders to drool or expansion of heating/melting filament? Can either problem be fixed by inversion without introducing, new, possibly more severe problems?

---

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

Quote
LoboCNC
Yes, the print falling on the extruder would make a mess, but at <250deg.C, nothing will actually be ruined or catch fire (other than the print). I've been mulling over making a shroud/duct to cover up most of the extruder body with just the nozzle tip exposed through a clearance hole. I might also make a simple optical sensor to detect anything falling on the extruder head and wire that into the e-stop.

A shroud sounds like a good idea. Not sure how practical a sensor would be.

Quote
DD
what sort of problems do vibrations cause in uninverted printers, and is it gravity that causes extruders to drool or expansion of heating/melting filament?

I suspect the first part largely depends on how rigid the frame is. My gut feeling is that the moving print-head is less of an issue than the moving bed in I3 style printers for height related problems. Ooze I suspect has a significant gravity component, particularly for less viscous plastics like pla. Wet filament is sometimes driven out of the nozzle by steam, but clearly the solution there is not to let your filament get wet! I doubt if expansion is a major cause because it would be readily solved by small amounts of retraction. I suspect that ooze isn't enough of a problem to justify an inverted build, but it's still a tremendous learning opportunity.

Quote
the_digital_dentist
Getting back to the original post, what sort of problems do vibrations cause in uninverted printers, and is it gravity that causes extruders to drool or expansion of heating/melting filament? Can either problem be fixed by inversion without introducing, new, possibly more severe problems?

Good questions which I'm not sure I have the answers to. (That won't keep me from making stuff up, tho.)

On the vibration issue, you typically have the moving print head up in the air at the top of your frame moving back & forth quickly, wanting to shake the frame. Thus, you need a moderately beefy structure to keep magnitude of the frame flex to an acceptable level. On a Prusa-style printer, the head only moves in X and the frame is pretty stiff in this direction so it's not a problem. For a large format printer, though, moving the table in Y becomes cumbersome, so you're better off moving the head in X and Y, requiring everything to be stiff in the Y direction as well. The inverted layout I'm thinking of will have the X-Y printhead mechanism stuck to a heavy plate (I'm thinking an 18x18" granite tile) which is then attached to your even heavier table. With all that mass, the XY motions of the head will induce very little vibration in the base. Thus, the frame required for the Z mechanism lifting the build plate can be pretty light.

With the drooling, I'm sure some is due to expansion, but some is also due to gravity. In theory you can tweak your retraction to compensate for the thermal expansion, but there's no good way to deal with gravitational drool.

As for the new problems introduced by printing upside down, I'm sure I'll run into some. If I actually build this, I'll find out...

It seems like an awful lot of trouble to avoid building a rigid frame and in the end the printer won't be any lighter weight, cheaper, or more portable.

---

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

Quote
the_digital_dentist
It seems like an awful lot of trouble to avoid building a rigid frame and in the end the printer won't be any lighter weight, cheaper, or more portable.

I figure a 350x350mm printer is never going to be light or portable. What I'm aiming for is accurate and cheap. I've attached a couple of diagrams that describe more fully what I'm thinking of. The first is the rough overall design. The X-Y mechanism uses a modified H-bot configuration which is shown in the second diagram. (It was too hard to try including the belts in the first diagram.) I'm calling this an "inside-H" drive, because unlike a standard H-bot which runs the belts around the outside, this design runs the belts up through the inside, thus avoiding the cocking tendency of the standard H-bot drive. Of course, because the belt crosses itself, I need to cheat a little by twisting the belt in places so it can run under itself. (You can see that the motors and fixed idler rollers are both at angles.) I'm twisting 60 degrees over about 50mm (worst case), which is not a big deal for 6mm wide belts. I'm going to use the same right-angle bowden extruder I used for my Delta-T printer (see https://youtu.be/J6X3oLVhHF4), so the X-axis mass is only about 100g and the overall Y-axis mass (with the moving rail plus the extruder) should be about 450g.

The Z-axis will be suspended from #150 test spectra lines wrapped over a differential capstans (one on each side). The lines aren't shown but it works by unwinding line from one diameter, looping under the post on the Z-axis carriage, and then wrapping it back up onto a slightly bigger diameter. This can give you very large reduction ratios. I'm also planning on putting an asymmetrical band-type clutch (essentially a string and a rubber band) on one end of the capstan rod to prevent the table from falling when unpowered.

Lots and lots of unknowns with this design - that's what makes it fun.

Edited 1 time(s). Last edit at 05/22/2016 07:11PM by LoboCNC.

The two advantages that I can see with an inverted printer is first heat from hotend will not affect the cold side of it, so melting zone and heat break could be smaller, and second ooze will not occur.

The only disadvantage that I can see is that if nozzle distance to bed isn't perfectly calibrated, the filament will curl into the nozzle.

Sounds interesting so keep us updated.

Why not just turn your current printer upside down, then you can see what the problems are and then design the new printer.

Quote
etfrench
Why not just turn your current printer upside down, then you can see what the problems are and then design the new printer.

Probably a good thing to try but unfortunately none of my printers can be used upside down, at least not without some modifications.

Upsidedown printing back in the day at Bath Uni - [vimeo.com]

Plus Fisher works quite well upsidedown [goo.gl]

Another disadvantage: heated plate. If you're not using a fully enclosed printer the heated bed would be totally useless as it won't help keeping higher layers warm. Even with 5 sides closed printers that's a problem, and fully enclosing a printer has other disadvantages. I'm pretty interested on what the outcome would be on parts that require supports, altough i think we would have the same problem. also i'm thinking about thin features with high angles that usually bend and are struck by the extruder. If they bend up while being affected by gravity, what happens when the gravity itself helps the process?

ps: sorry for my english, still trying to improve

Quote
Druwan
Another disadvantage: heated plate. If you're not using a fully enclosed printer the heated bed would be totally useless as it won't help keeping higher layers warm. Even with 5 sides closed printers that's a problem, and fully enclosing a printer has other disadvantages. I'm pretty interested on what the outcome would be on parts that require supports, altough i think we would have the same problem. also i'm thinking about thin features with high angles that usually bend and are struck by the extruder. If they bend up while being affected by gravity, what happens when the gravity itself helps the process?

ps: sorry for my english, still trying to improve

Interesting point about the heated bed (and your English is fine). I'm not sure, though, if you need to keep the upper layers warm. My impression is that the heated bed is primarily to ease the transition between the very bottom layers which can't shrink because they are stuck to the bed, and the upper layers which shrink normally. Even with an unheated bed, it is only the bottom layers of a print that will warp - the upper layers print without deformation.

(Edit: I should point out that most of my experience is with PLA where layer-to-layer adhesion is not much of a problem. With ABS, keeping the upper layers warm can help with layer-to-layer adhesion.)

Also, you could argue that an inverted bed does a better job of trapping a pocket of warm air around the base of the print vs. a non-inverted bed where the heat is free to rise across the entire surface. You could even build a lip arount the edge of the inverted bed (maybe 15mm tall?) that would enhance the trapping of warm air. Without an impossibly complex heat transfer and fluid flow analysis, though, it'd be pretty difficult to make any predictions.

Regarding the curling of overhangs being accentuated by gravity - that for me is the big question. There are plenty of videos of upside-down printers printing modest overhangs without any problem, but when you get to more extremes overhangs - who knows?

Edited 1 time(s). Last edit at 05/25/2016 05:28PM by LoboCNC.

I've got some initial printing results! Having watched miles of filament print over the last couple of years, so far, printing upside-down looks just like printing right-side up. The treefrog print, which has some steep overhangs, printed the same as on my conventional printers. Even my bridging test - ranging from 4mm to 34mm - came out pretty much the same. I'm sure longer bridges would cause trouble, but I'd never really want an unsupported bridge longer than 20mm anyway. So far, the 2 main drawbacks are: 1) poor visibility of what you are trying to print and 2) you have to manually wipe the extruder's nose when doing any sort of priming. The payoff, though, is that I've got a rock-solid printer (sorry about the pun) with very little structure. I've got accelerations set to 3000mm/s/s and the jerk set to 20, and I'm getting very smooth prints even though the heated bed is suspended from fairly spindly 12mm rods.

I did have a few unexpected challenges:

- I'm using a 350mm x 350mm x 1/8" aluminum plate without any glass for the heated bed. In general, it is sufficiently flat, but I discovered that the 2-3deg. cycling of the heater caused a tiny amount of warping that shows up as banding in the print. I'm using an 800w, 120v silicone heater, which is a kind of overpowered, so I managed to get the temp. swing to less than a degree by connecting it through a heavy-duty dimmer switch.

- To keep the build plate from falling with any unexpected loss of power, I made a simple asymmetric band brake using a piece of fishing line wrapped 1.5 times around a pulley. It provides very little friction when lifting but more than enough to hold the build plate up when trying to lower.

- Keeping the bowden tube and cable from getting hung up on the mechanism was a real headache. It took me a few iterations before I got a fence/guard fashioned that would allow the tube to run over everything smoothly.

Hopefully I'll get a video posted shortly.

I love it. My word there isn't much of it up top is there? I would never have guessed that the bridging test would work so well. The granite (?) base brought a real smile to my face.

Well done!

Quote
LoboCNC
I've got some initial printing results! Having watched miles of filament print over the last couple of years, so far, printing upside-down looks just like printing right-side up. The treefrog print, which has some steep overhangs, printed the same as on my conventional printers. Even my bridging test - ranging from 4mm to 34mm - came out pretty much the same. I'm sure longer bridges would cause trouble, but I'd never really want an unsupported bridge longer than 20mm anyway. So far, the 2 main drawbacks are: 1) poor visibility of what you are trying to print and 2) you have to manually wipe the extruder's nose when doing any sort of priming. The payoff, though, is that I've got a rock-solid printer (sorry about the pun) with very little structure. I've got accelerations set to 3000mm/s/s and the jerk set to 20, and I'm getting very smooth prints even though the heated bed is suspended from fairly spindly 12mm rods.

Hmm, the frog looks nice, but... shouldn't the bridge be bending "upwards" if printed upside down? With the hotend inverted I would assume that any long overhangs would interfere with the nozzle and get flattened out during the next layer pass. We need some videos here.

Also, if the print gets unstuck from the bed, then it crashes down over the X/Y assembly. This could be detected with some weight sensor on the bed and then stop the print.

Quote

This could be detected with some weight sensor on the bed and then stop the print.

That's a neat idea. Tricky at the beginning of a print when there isn't much added weight, but it will get progressively easier as the print progresses. The piezo sensors might even work for this as they respond quite well to sudden changes.

Is this for the Australian Market?

Quote
ipcalit
Hmm, the frog looks nice, but... shouldn't the bridge be bending "upwards" if printed upside down?

Yeah, it is actually kind of peculiar. As it printed the first bridging layer, the filament did sag down a little as you'd expect. But when the next layer went down (er, up), it pushed the bridging layer away so the net result looked very much like conventional bridging. I'll try to get a video of the bridge printing, but it's pretty tricky seeing exactly what's going on thru the small gap under the build plate. I'll try moving the print to the edge for a better view.

Quote
JamesK

Quote

This could be detected with some weight sensor on the bed and then stop the print.

That's a neat idea. Tricky at the beginning of a print when there isn't much added weight, but it will get progressively easier as the print progresses. The piezo sensors might even work for this as they respond quite well to sudden changes.

I'm pondering some sort of optical "spaghetti" sensor on the print head that would detect the inevitable loose bits of extrusion generated with most any serious print failure.

Quote
LoboCNC

Quote
ipcalit
Hmm, the frog looks nice, but... shouldn't the bridge be bending "upwards" if printed upside down?

Yeah, it is actually kind of peculiar. As it printed the first bridging layer, the filament did sag down a little as you'd expect. But when the next layer went down (er, up), it pushed the bridging layer away so the net result looked very much like conventional bridging. I'll try to get a video of the bridge printing, but it's pretty tricky seeing exactly what's going on thru the small gap under the build plate. I'll try moving the print to the edge for a better view.

One trick you could try is to insulate the hotend really well then blast that fan you have there to stop the sagging. On the next pass the hotend will push it into the right position and freeze it there. Theoretically at least You could also remove the nozzle fan and have instead larger fans on the side and benefit from the flat base to achieve some laminar flow.

And here are a couple of videos:

printer walk-around

bridging test

I think that deserves a round of applause

I would never have guessed that the bridging would work so well, and it's completely awesome to see that big print hanging from the bed

Great job!