That’s my kitchen table. I’ve gotten the frame of my Prusa Mendel built, and I’m just waiting on some LM8UU linear bearings to arrive so I can move on. You might wonder why one of the parts is red… yeah, there’s a story behind that. I’ve opted for a few deviations from the “standard” Prusa—if you could call anything open-source a standard.
Mainly, I’ve decided to use LM8UU linear bearings on all the smooth rods rather than the standard printed bearings. I’ve heard stories of the printed bearings wearing out over time, and especially for the build platform and extruder carriage, replacing the bearings will mean recalibrating for the new Z-height every time. That seems like a pain in the neck, and I managed to order 12 LM8UU bearings from some Chinese company called “CoolCheapWorld” on eBAY for only $20! The downside is slow shipping; I ordered them over three weeks ago and they’re still not here yet. Guess that’s what I get for ordering stuff halfway across the world.
In order to use the bearings, I had to print different X-ends that were just recently created by Josef Prusa himself. They’re a very clever new design. I also printed a copy of Greg Frost’s neat LM8UU-compatible X-carriage.
All in all, I’ve spent $487 so far, which I consider to be a fantastic deal considering that I bought my Thing-O-Matic for $1,300, and it’s got a smaller build platform and can’t print even half as fast as the Prusa will be able to.
I finally got around to making Jetguy’s Thing-O-Matic extruder mod to make it print in PLA better. The primary motivation was a frustrating experience with a multi-hour PLA print that was beautiful until a bulge jammed in the barrel when it was 93% done.
All you need to do is cut some slits in the barrel below the heat sink to retard the upward flow of heat. It turned out that a jeweler’s saw is unnecessary and in fact insufficient for cutting through the stainless steel, but if you have a vise and a hacksaw, you can do this in about 15 minutes. I probably spent more time removing and reattaching the hot end than I did cutting the barrel.
The result? It helped! …a little. Not as much as I would have liked, unfortunately. The mod did retard the flow of heat, but not enough that I could touch the top of the barrel without feeling like I was gonna burn myself. The tube really needs active cooling. Unlike most Thingiversians and Reprappers, I don’t have a seemingly endless supply of randomly-sized DC computer fans lying around, available for bolting onto electronics projects, so I took the only fan I had and and aimed it at the bot:
It’s a shotgun solution, and it’s not perfect because it cools the heater while it’s trying to heat itself, but it sure keeps the tube nice and frosty! and I’ve been able to print for hours without any jams. You just really need to cool that tube to have any hope of printing anything large in PLA on your Thing-o-Matic.
The awesome thing about a 3D printer is that you can use it to mock up a new design at close to zero marginal cost. In designing my new super-small NEMA 8 extruder, I’ve been able to come up with an idea, model it on my computer, and then bring it into the world. Here’s my quick-and-dirty first draft of a bracket to connect the motor with a worm drive, printed in a festive blue ABS:
It’s a pretty magical experience. It also lets me quickly and easily identify design problems and correct them. You can see the progression:
Alas, the screw holes are unfilled because nobody local has any M2 screws. Also, I didn’t actually design the worm drive’s gears; I grabbed ‘em off Thingiverse because of coursesomeone has already made a printable worm drive! It’s a 16:1 gear ratio, which is probably too much for my needs: the motor’s maximum torque of 40 mNm works out to 640 mNm when geared down 16:1. That’s definitely overkill, and it will also limit the speeds I can rotate the drive shaft that feeds the filament.
Now that I have an adequate mounting solution, I need to design a drive for myself that’s probably somewhere between 4:1 and 6:1, and I need the circular gear to firmly hold an axle that will either hold a toothed pulley for feeding the filament, or will be hobbed to itself feed the filament. I’m super-happy with the toothed pulley on my Thing-O-Matic’s extruder, but Makerbot is charging $12 for it, and I’d still need a 6 mm axle for it to sit on. Add up the cost of the motor ($24), the pulley ($12), the four LM8UU bearings I’m planning to use for motion ($8), and a few extra bucks for the fasteners, and the filament drive part of the extruder would approach $50. Of course, as a point of comparison, that’s the price Makerbot charges for just the motor on their extruder.
It might be more cost-effective to just buy an appropriately sized bolt and somehow hob it myself…
Even though I haven’t actually started building my Prusa Mendel (half of the parts are still in transit), I’m already thinking about what’s next. It seems to me that one of the biggest challenges in FDM 3D printing is the issue of how to make the extruder move faster. The current trend is to make the print bed stay in one place (or just move in the Z-axis) while the extrusion head zips around. Of course, it can’t do that very well if it’s really big and heavy, and the current solution involves separating the big heavy motor that pushes the filament from the hot nozzle that melts it, connecting the two with a flexible cable that the filament travels down. This is known as a Bowden extruder system. The Ultimaker uses one, and you can see it pretty clearly:
*Drool* Oh how I want one…
But the Bowden extruders have a major problem: hysteresis. That is, when the motor does something with the filament on one end, it can take a bit for that change to reach the nozzle, especially with springy or compressible plastics. That reduces the motor’s ability to control how much filament is extruded with any degree of precision, depending on the length of the Bowden cable and how well-tuned the mechanism is.
I got to thinking: what if you could design an extruder with a motor that’s light enough that it could be mounted on the moving mass with the hot end? This would of course be the much sought-after small motor with high speed and high torque. In the real world, generally you have to settle for only one of those characteristics, or maybe two if you’re willing to pay a lot.
I just picked up a cute little NEMA 8 motor. Here it is next to the much larger NEMA 17 motor that drives my Thing-O-Matic’s Z-axis:
It’s a quarter the size! And with that reduction in size comes a reduction in torque. The little NEMA 8 is only capable of outputting 40 mNm (that’s milli-Newton-meters) compared to 240 or so for the NEMA 17. But just how much torque do you actually need to push 3mm filament sandwiched between a toothed pulley and a ball bearing?
…It turns out that nobody’s really sure how much torque it takes. There are ballpark estimates, of course. Ed Nisley estimates that it’s between 100 and 700 mNm, with maybe 300 mNm being a good place to start.
300 mNm is a lot of torque to ask from a motor that can only put out 40. I’d have to slow it down with a 7.5:1 gear ratio. That’s certainly possible, but it limits how fast the feed pulley can go if the motor itself has to be spinning 7.5 times faster. The motor on my Thing-O-Matic’s extruder can go a maximum of 50 RPM, so I’ll use that as a theoretical maximum for my little NEMA 8. That means the feed pulley would be able to turn at a maximum of 6.66 RPM (50 RPM / 7.5). Is that enough? That’s what experiments are for!
I’m still tweaking my PLA setup, but it’s getting better. Here’s a 70 mm tall Yoda I just printed with a nice small layer height of 0.15 mm:
There’s still work to be done around the face and ears, but I know what the problem is. I was extruding this PLA at 225ºC, but I think that’s too hot. It was taking too long to cool, so by the time the next layer was being laid down in areas with overhangs—such as the chin and face— the earlier layer was starting to droop. I’ve since dropped my extrusion temperature to 215º and the drooping has been much better. I’m also going to try directing a desk fan at the in-progress print until I can manage to hook a computer fan up to the extruder assembly.
You can get a sense of the detail my bot is physically capable of by looking at the rear of his cloak:
It’s like a real high-resolution rapid prototyping machine!
I’ve finally, finally managed to print with PLA. And it’s every bit as wonderful as I’d hoped. Here are my initial observations:
It’s sticky. This stuff really liked to bind with itself. I haven’t had any problems at all with warping or curling, even in situations that would have destroyed an ABS print. I opened a nearby window and the breeze cooled it better rather than causing cracks. Marvelous!
It smells good! They weren’t lying. It’s got a nice sweetness to it, unlike ABS’s harsh chemical smell.
It likes active cooling. In the few prints I’ve done so far, extruded PLA stays soft for longer than ABS does. That means that you need to either slow down or cool the in-progress print with a fan or something. Even blowing on it has helped. I’ve turned off my heated build platform as it’s obviously counterproductive in this use.
Reversal doesn’t work yet. I think I know why, though. Reversal works by rapidly reversing the extruder motor to suck the molten filament back up in order to make plastic stop coming out of the nozzle while the head is moving from place to place. But in order for that to happen, there has to be a mechanical link between the plastic that the motor touches and the plastic coming out of the nozzle. That’s not the case with my extruder because the plastic bulge is still there in the barrel.
Now, increasing the tension has allowed my powerful extruder motor to smash through that expanded plastic and force it down, but that doesn’t change the fact that there still is a bulged bit of plastic in the barrel. The motor can push it down, but when it pulls up, it’s pulling up on the bulge, rather than the plastic coming out of the nozzle. If I want to get reversal working, I think I’ll need to make this mod, or the more advanced one described in the comments.
My bot is capable of much more detail than I thought. Until I saw this picture, I had no idea how important layer height was in determining the final detail of a model. I thought a smaller nozzle was the key; it turned out to be thinner layers. My 0.4 mm nozzle is just fine for producing prints that are an order of magnitude more detailed than what I’d been doing before. Example:
Just look at the surface detail on his forehead and robe! That was printed with a 0.15 mm layer height. It’s still double the resolution that Martijn was able to achieve with his Yoda head.
Brute-forcing the blockage through the barrel isn’t a long-term solution. I’ve still gotten PLA jammed in the barrel due to that stupid bulge a few times, and it’s a real pain in the neck to clear. Here’s an example:
No amount of brute force is gonna push that through. I’m definitely gonna have to do that mod to reduce the heat level of the middle and top end of the barrel. The problem is that PLA expands at a lower temperature than it melts at, so when the temperature at the top of the barrel is hot enough to cause it to expand, but not hot enough to melt it entirely, it’s a recipe for jamming.
It seems like people have had luck cutting slices out of the barrel near the bottom to retard the movement of heat. If I can do this and keep the top of the barrel nice and cool, it’ll probably stop the jams right in their tracks.
In Replicatorg’s “Motherboard Onboard Preferences” dialog box, increasing the “Z home offset” value puts the extruder nozzle closer to the build surface.
While I was in the middle of printing out the parts for my Reprap, I ran out of ABS plastic. Thankfully, in my last order from Ultimachine, they happened to throw in a few loose coils of colored ABS plastic. How nice! Now it was time to finally use them, mostly out of necessity. So I printed a nice motor mount in red:
Then, just for fun, I tried out a print of a super high resolution Yoda bust with a ridiculously small layer height—0.15—lower than I’d ever tried before. To my surprise, though the next print would need a lot of tweaking, it came out better than I had hoped. And the detail in the areas that came out was phenomenal:
Boy, that stuff sure is shiny, huh? Hmmmm… And it seemed to be a lot stickier than the white ABS I’d been using before. Hmm… On a hunch, I pulled the filament out. Tug tug tug. Resistance! After getting it out, here’s what it looked like:
See that bulge at the end of a long skinny part? That’s what PLA looks like when you pull it out. This was PLA. I’d been successfully printing with PLA, and I didn’t even know it!
I have no idea why this suddenly started working after every single prior attempt to print with PLA resulting in failure. And moreover, it started working without me knowing, and without changing any of the software settings to compensate. There are times when technology really does seem like black magic.
In all seriousness, I do have a theory. I tightened the filament drive idler to Schwarzenegger-like grip strength, and I think that helped it bust through jams caused by the PLA expanding too high in the barrel. Never underestimate the power of a strong grip on the filament. Stepper extruders put out a ridiculous amount of torque, and the stronger they grip the filament, the more force they can push it with. I’m convinced that this was the decisive factor. The biggest problem I had in my last few attempts was that the filament inside the barrel was expanding and causing too much friction. I think allowing the extruder motor to push the filament harder gave it enough oomph to overcome whatever friction was caused by the unwanted expansion.
After a nice refreshing vacation followed by five less nice days of post-vacation flu, I’m ready to print again. And let me tell you, I’m really starting to sour on ABS plastic. The stuff is just nearly impossible to prevent from warping or cracking. It’s not so bad on small prints, but anything that takes more than an hour or so is likely to pull up on the corners, often very badly. In addition, tall prints will often crack between the layers. Here’s an example:
I’m so desperate that I’ve gone back to using a raft, and though it’s basically solved all my problems with the corners pulling up during multi-hour print jobs, it’s also made the bottoms ugly. Ahh well, such is life.
As you can see, I’m also using blue painter’s tape. It’s less finicky than kapton tape, and I found that I wasn’t changing the kapton frequently enough because of how much of a pain it is to handle. The blue tape’s slightly worse surface adhesion properties don’t matter at all if you use a raft when printing using ABS. Also, it’ll allow me to quickly and easily switch to PLA once I finish up my last ABS roll.
Despite the fact that I have yet to successfully print with PLA, I can’t wait to use up the last of my ABS plastic and really put the effort into making it work. The aforementioned issues with warping and cracking are caused by the fact that ABS shrinks as it cools. These are frustrating problems because you won’t see them until you’re printing something large that therefore 1) takes a long time, and 2) you actually care about. When it happens, you’ve lost hours of time and many feet of filament.
By the way, here’s the final product of what I was printing in the video above:
These doohickies are all parts for a Prusa Mendel, my next 3D printer, and a machine designed to print PLA plastic. I’m pretty sure I can build it for under $500.
I managed a first print of my in-development Star Destroyer model. It’s proving to be a tough one!
You’ve probably noticed the lack of a bridge section (of which there is entirely too much information available). That’s because I still haven’t figured out a great way to handle it. Sadly, Skeinforge’s built-in support material feature doesn’t seem to handle the case where you have an overhang above other geometry rather than just the surface of the build platform. So that means I’ll have to build some support material into the model itself or else remove the bridge and have it printed as a separate piece, which comes with pitfalls of its own. I’m still weighing the options.
I’ve been really trying to get the entire model printable with only one file, because I don’t like multi-part models that you have to do several prints of to finish. I laid out the top and bottom pieces beside one another and they fit pretty well:
This is the first model that’s made me a tad disappointed by the small size of my build platform. My Star Destroyer has a lot of detail, but it’s just really hard to get much of it to show up on a 4-inch model made by a printer with a 0.3 mm nozzle. So much just gets lost in the individual steps of the layers. I think it would look a lot better doubled in size to fit the 8-inch bed of a RepRap or an Ultimaker. And at that size, it would look and feel less like a toy.
