MakerBot Industries

NYCDesigner's PET automated build platform conveyor belt

Sometimes there is just no replacement for experimentation.  I’ve been using my Automated Build Platform for about two weeks now¹ , but have found its utility varies with the material I’m using.  Here’s what I’ve tried along with a few notes:

• The mylar belt works much better with ABS than with PLA. ²
• Sanding the mylar belt slightly did not help the PLA stick any better.
• Heating the build platform did not seem to help the PLA stick any better.
• Using just the mylar belt with a single strip of Kapton tape running around one edge of the belt works much better than wrapping the entire belt in Kapton tape.  I noticed that when the Kapton is doubled over itself it will develop wrinkles once it has been around the conveyor.  These wrinkles then remain causing an uneven build surface.
• NYCDesigner tried out a PET belt, which he found easier to assemble than the included mylar belts, but didn’t notice the PET working any better.
• Anfroholic suggested sanding the build surface in only one direction, to preserve the peaks and valleys on the build surface.
• Feilen suggested a Kapton belt has worked excellently.
• I’ve found that blue painter’s tape works great with PLA.  I haven’t tried wrapping it around my entire conveyor belt, just the print area, but I’m hopeful it would work that way too.

And a bonus list of THINGS TO NOT DO WITH YOUR AUTOMATED BUILD PLATFORM:

• Don’t turn on the motorized conveyor belt while your object is still quite hot – especially if your object is very thin.  My test USB enclosure just became even thinner when the motor pulled it down underneath the actual build platform.
• Don’t wrinkle your PET belt.  That’s going to cause problems with an uneven printing surface and they’re difficult (impossible?) to get out.
• Don’t run your automated build platform backwards – it will make the Kapton tape pull up.

What have you learned with your automated build platform?  What are you using for your conveyor belt?  Are you sanding in any particular way?

1. And loving it! [↩]
2. And, frankly, PLA is my new favorite thermoplastic. [↩]

Printed 608 bearing using BB's by TheRooster

Some days, such as today, I love being wrong. 

The other day I suggested printing a 608 bearing using 4.5mm BB’s would be impractical given the dimensions of a standard bearing.  There were a number of very well thought out responses from Tre3 and TheRooster to my wild¹² accusations.³

Among other excellent points, Tre3 pointed out that with careful purchasing you can find ball bearings as low as $0.05 per bearing. ⁴ That’s pretty dang low.  TheRooster said he was able to pick up 2500 zinc plated steel BB’s for $4 from his local big-box store.  That’s about $1.70 for all 53 bearings required for a RepRap.  Cheap BB’s are all well and good – but you’d still need to find a way to shoe-horn them into a printed 608 sized shell.

And that’s exactly what TheRooster did – he designed, printed, and assembled a 608 sized bearing utilizing those 4.5mm BB’s as the ball bearings.  I cannot wait to try this design out.  I realize the labor costs involved in creating your own ball bearings is probably prohibitive.

I think this misses the central questions:

• Just how far can you push a 3D printer?
• When you can make nearly any arbitrary shape out of plastic, just what are the limits?
• What is possible once your biggest cost is the time required to assemble the parts in front of you?

Tre3: I was going to take you up on your offer to test my plastic bead bearings, but they are officially garbage now that I’ve seen TheRooster’s improvement.  However, I would love to see how TheRooster’s printed ball bearings using BB’s match up to your tests.  Are you guys up for the challenge?

1. Drunken?  Intoxicated? [↩]
2. BUI – blogging under the influence! [↩]
3. If I were really smart, I would claim that this was a bit of slight-of-hand reverse-psychology on my part.  Alas, I am just not so clever.  Or devious.  [↩]
4. Given that each bearing requires roughly 15-20 balls and there are 53 bearings in a RepRap, this comes to $39.75 for the lot. [↩]

Small bead ball bearing

There has been a number of excellent points made in the comments to my last post about a printed ball bearing and the associated draft uploaded to Thingiverse.  Oh, and I’ve decided to dub this the “bead bearing.”  My latest version uses super cheap mostly-spherical plastic beads from a local craft store in place of the normal metal ball bearings one sees in commercial bearings.

@tre3 pointed out printed bearings aren’t a good value proposition

VBX sells a RepRap’s worth (x53) of ball bearings for about $45.00, inclusive of their cheapest shipping option, exclusive of tax.  The plastic beads I purchased were $4.00 for 1700 beads, or $0.04 for 17, which is a little higher than the number required for a single bearing.  The design probably requires about 0.5 cc of plastic per bearing, at a cost of $0.04/cc of plastic.  We’re looking at a materials cost of $3.18 for a full set of 53 bearings. ¹  I believe tre3′s main point is that for the quality you get for the commercially available without any of the printing hassle, the cost saving of printing your own just isn’t warranted.

To this, I would respond:

• This was a proof of concept to see if it was viable.  I think it is for low load usages.
• For use in a RepRap or MakerBot replacement part it would only need to operate smoothly at speed, without much load.
• The price differential is significant.  A $42 differential could actually be a deal breaker for some people.
• At 1/10th of the cost, the longevity of the parts probably isn’t as big a deal.
• Assuming the result of wear and tear turns the plastic bearings to plastic dust, the dust could be recycled into new parts!
• I believe the benefits of low weight, ease of assembly, and cost outweigh the potential problems with this design

@TerryB suggests slicing up a steel rod for circular rollers

This is the same essential theory behind jrombousky’s printed bearings – and it’s a good idea.  He uses slices of ABS filament instead of steel rod slices.  However, his bearing is too tall to use as a drop-in replacement for a 608 bearing.  I think it would be really cool to design a printable drop-in replacement for the typical 606 and 608 bearings used in a MakerBot.  As for cutting a steel rod – I was looking for a quicker easier fix with cheap off the shelf parts.

@Foxdewayne and @Whosawhatsis suggest soft BB pellets, @pandelume, @beak90, and @alansblue suggest metal ball bearings

Whosawhatsis had also suggested those soft BB’s, but they’re 6mm in diameter which is too large for this particular design.  A 608 bearing is 7mm tall, so there’s just not enough room in the design of a 608 replacement. ²  Others have pointed out  I’ve found the plastic beads to be spherical enough to work pretty well.

However, I’m open to trying BB’s too.  After a cursory search I found 3mm (1/8″) bearings, 100 for less than $3.  However, this will bring the minimum cost to $0.51 per bearing assembly.  At that price, you really might as well use a commercial bearing. ³

@Whosawhatsis says: “On the lubrication issue, Adrian Bowyer mentioned a while ago that he would only trust silicone grease lubricant on ABS/PLA. This was in the discussion of printed gears, but the same should hold true for bearings”

Hey, who am I to argue with Adrian. 

@beak90 says: “Why not just make it one piece and add an operator stop to the gcode before it fills up the top layer to seal the balls in? That would make things pretty easy. The code is M01 (with the text to display in parenthesis). And as far as the balls for the bearing, why not use actual metal ball bearing balls? They aren’t too expensive and it would make a perfect bearing.”

The M01 operator stop is a great GCode trick for pausing a print while you do something.  It’s a fantastic way to embed objects or materials within a printed object.  I did not use that trick in my initial designs because I wanted to see if it would work with beads at all before investing the time in tweaking the GCode.  With the small circles I’m printing to achieve this design, I’ve been printing from the SD card to allow for faster transmission and processing of the instructions  by the MakerBot.  However, a higher quality SD card print comes at a cost – it ignores M01 pause codes!  By printing the two sections as separate pieces I have less concerns about strings, ooze, or other molten plastic issues.  And, lastly, this way I don’t have to worry about the extruded plastic trying to fuse to the bead bearings themselves – which was the main reason for using non-printed bearings in the first place.

@Buzz suggests “Junk Jewelery” (lots of glass and plastic beads, the small thread hole should not matter) or the removing the heads off glass head sewing pins with pliers.

Both of those are excellent suggestions.  Once we start to move away from thinking of bearings as only printed or commercial, I think the possibilities really open up.  There should be TONS of possible replacement parts.  If you operate your machine in sub-zero temperatures⁴ , you might even get away with using frozen peas!

***

Thoughts:

If you haven’t checked out the Gada “RepRap” prize, it’s worth a look.  Basically the goal is a mostly-autonomous printer that can make most of it’s own parts.  Two of the most stringent goals are that (a) materials and parts cost under $200 and (b) it be 90% printed by volume.  When you’ve got a total budget of $200.00, being able to have all of your bearings for under $4.00 is kind of a big deal.  While the ideal winning printer is supposed to be able to generate all of the printed parts for a duplicate copy within 7, there are no explicit longevity or precision requirements.  I suspect these parts might just be good enough for consideration in someone else’s designs.  If these tests are successful, I hope that I would have contributed towards someone’s success in this prize.

And, as a side note, one of the benefits to having this particular bead bearing sized design is that you could always build a RepRap using these parts and “upgrade” to better commercial parts later one as time or money allow.  They could also work as interim repairs.

Testing:

Overall, I have a feeling printed bead ball bearings can work as a possible replacement part in a Cupcake 3D printer or a RepRap.  However, I intend to test out a newer, possibly more resilient version, this weekend.  The new version has a two-part snap-together internal ring assembly for inserting the beads, is sized to the specifications of a 608 bearing⁵ , and is specifically designed for these 3mm beads rather than the larger plastic pellets.  One test will whether the internal ring can snap together properly.  One test will be how the part operates at high speeds.  I’ll probably put it in an electric drill and see how it goes.  If all goes well, I’ll try installing it into Bender‘s Z-axis and see how it goes.  I will be monitoring it for smoothness of operation, sound, vibration, how well it holds up to speed.

Do you have any other suggestions on tests or criteria?

1. I realize the VBX pack includes two different kinds, but let’s pretend it all evens out. [↩]
2. Dimensions of a608 bearing are 22mm OD, 8mm ID, 7mm tall [↩]
3. Besides price, metal also increases the weight. [↩]
4. Which would make warping a huge pain! [↩]
5. Dimensions above [↩]

And you thought Sketchup was easy to use!  This article from CreativeApplications.net showcases the phenomenal work of Karl D.D. Willis:

Created by Karl D.D. Willis, Beautiful Modeler is an iPad and Desktop openFrameworks application for gestural sculpting using iPad as a multi-touch controller. Each finger is used to control a single touch point in the model, with multiple layers working to build up 3D volume. As the controller is connected over the wireless network, it can be moved freely to change the viewing angle of the model using iPad’s accelerometer.

The model itself is presented on the main display rather than on the controller itself; this prevents occlusion of the model when sculpting with the whole hand. The controller screen does not need to be viewed while sculpting, meaning the controller can be rotated or flipped to sculpt from a range of angles. Currently the model is constructed using metaballs (thanks to Golan‘s code), but this is just one approach for gestural input to be transformed into geometry.

Because both Beautiful Modeler and the Beautiful Controller were created using openFrameworks, the finished mesh can be exported as an STL file (thanks to ofxSTL), meaning the sculpted form can be fabricated immediately. In the video above, the positive mesh has been post-processed to create a negative form for fabrication with a plaster-based 3D printer.

The importance of a good user interface cannot be over emphasized.  A good UI is something where you see someone use it and think, “Hey, yeah, I get it.  Let me try!”  And Beautiful Modeler looks like a great user interface.

Printed bearing, nearly 608 size

This weekend I tried printing Rayraywashere’s ball bearing with mixed success.  The plastic spheres inside we pretty well fused to the sides of the bearing, which made for a difficult cleanup.  Although it got better with time, it was a laborious process.  Ultimately, printed bearings that rely on printed balls may not be the way to go for everyone.  Even if you can print it without fusing the balls to the bearing, there’s no guarantee the balls would be sufficiently spherical to work properly.

That ball bearing design got me thinking – if I could find a reasonably ubiquitous and cheap alternative to small printed spheres, I could make the entire design much smaller and probably significantly more reliable.  The photo above is rough draft/proof of concept for a printed bearing only slightly larger than a traditional 608 bearing.  Rather than printed spheres, it uses plastic pellets of the sort typically used as stuffing in craft projects.  I sorted through a lot of these and used only the most nearly spherical ones.  However, there was still a lot of variation that lead to the bearing getting jammed.

Later I emptied the plastic pellets and tried out small spherical 3mm plastic beads.  These have worked really well in this printed design.  To improve upon this design I intend to move the “bead filling gap” to the interior ring or change the ring system so that two rings will snap together and cover the filling gap.  Overall, I am very happy with this result.  I’m looking forward to actually installing a slightly smaller version into my Cupcake 3D printer in place of a 608 bearing to see how well it works.

I spent $4.00 for 1700 identical plastic beads, silver in color, 3mm in diameter.  It takes about a dozen of these beads to fill the bearing.  The potential savings are pretty self-evident.  The per-unit cost for each bearing is probably only about $0.05 or so in plastic and beads.  That’s pretty good compared to $2+ for commercial bearings.  The real test will be how smoothly they work, how well they work at high speed, how quickly they might wear out.  However, I think I may be on the right track here.

Any suggestions?

Dave Durant makes Skeinforge less spooky

Dave Durant has continued his series on how to configure Skeinforge for your 3D printer. ¹  Although Dave prefaces his comments by saying these are not necessarily recommended settings, he (again!) does a great job of going through each setting available in Skeinforge 033 and explaining why he is using that particular setting – along with suggestions of when those parameters may not be appropriate for your ‘bot.

Starting at ReplicatorG version 0020, the folks at MakerBot are also including a recent version of Skeinforge. w00T!!

I’ve been advocating moving up to more recent versions of skeinforge for months now, mostly because I think I can sorta figure out how to configure them. The older versions just make me go cross-eyed and confuse me. YMMV.

Since my last blog post was a bit (or more than a bit.. or even a LOT more than a bit) long winded and preachy, I’ll try to keep this one short. Below’s my notes on going through all the modules in skeinforge 33, the latest as of today. Note that this isn’t really “Dave’s recommended settings.” It’s more like, “where to start things, if you’re moving up from an old version.”

Bottom: This is new, as of version 32 or so. I’m not sure what it does but it seems to be the “bottom” parameters that used to be in the Raft module so I’m leaving it enabled and just taking the default values.

I think the “bottom” params in raft used to control how high the nozzle was at the very start of the print. Since I always tweak a Z pulley when I start printing, I don’t care too much about this one

Carve: Absolutely enable this. This one contains two of my Big 4 (or 5) parameters, Layer Thickness and Perimeter Width Over Thickness. Those two I care about, the rest I just leave as they are.

Chamber: Disabled. I think this is for heated build chamber. Like, if your bot was all enclosed in a temperature controlled environment, you’d want to enable this and figure out some reasonable values to plug in here.

Clip: Enabled. Clip is sortofa method that looks for parts of the print that are really close together and clips bits off, to keep too much plastic from being put down.

I’m very sure that the values you plug in here could be tweaked to get a little more performance out of your bot but I suspect it’s a 2% thing – good to learn if you really want the absolute best prints you can get but not really worth it for most people. I take the default values.

Comb: Enabled. I love comb. Comb rules.

Comb basically tells skeinforge to not let the extruder move outside the perimeter if it can avoid it. This makes a HUGE difference (in a good way) in cleanup times but does add to the total print time. Unless you’re really, REALLY interested in the quickest print you can get, just leave it enabled.

Cool: Disabled. This is an interesting module but I think DC extruders (which virtually all Makerbot people have) don’t play well with it.

Cool lets you tell skeinforge the minimum amount of time it should spend on each layer. If a layer is going to take less than that amount of time, skeinforge will add gcode to orbit around – aka: waste time – until that minimum is reached.

This is a great idea but unless you really have ooze under control, you probably don’t want to use it. If you don’t have ooze under tight control and enable Cool, it’s going to make a huge mess.

Also, if you do want to enable it but don’t have a stepper extruder, don’t use the Slow Down option. Slow Down tells it to drop the flow and feed rates down instead of orbiting – with a DC motor, the extruder will stall and you’ll get no plastic coming out.. Bad.

Dimension: Disabled. This is for “5D” stuff that’s not supported (yet??) on Makerbots. The 5D stuff, if I understand it correctly, are extensions to gcode that lets the extruder move faster on diagonals than it does on straight-X or straight-Y lines. Nice feature but not something Makerbots can use.

Export: Enabled. Not 100% sure what this does…

Some types of machines process the gcode in the firmware and one thing Export allows you to do is strip all the comments (stuff that’s user readable but ignored by the machine) out of the gcode. In theory, if you have lots of comments in the gcode and the machine isn’t very fast, stripping out comments will help prevent problems.

On a Makerbot, the gcode is processed by your PC (or Mac or whatever) so this isn’t much use for Makerbots. Leave it enabled, tell it to strip comments out or not – I leave them in but it doesn’t really mattter.

Fill: Enabled!!! I spend more time in Fill than I do anywhere else. More on this module later (next post?) but once you get your profile all nice and dialed in, this is pretty much the only place you need to be when you want to print something.

Fillet: Disabled. This is another interesting module that I don’t use very often. Fillet sorta rounds off sharp corners in the object, which can help if you’re suffereing from quality problems due to high feed rates.

Picture printing a perfectly square cube. There’s lots of “full power to X. Stop X! Full power to Y. Stop Y! Full reverse on X!” over and over. This can encourage belt-related issues like backlash. Fillet lessens these issues by rounding things off of a little.

In general, I’d say disable it for prints that require bolts and bearings and things – stuff that’s been measured out and has bits that fit together – and enable it for more organic-type object that don’t require precision.

Home: Disabled. This allows you to add a bit of custom gcode to the start of every layer. I’m sure there are good reasons why you might want to do this but I don’t think any apply to Makerbots..

Hop: Disabled. Not really sure what this does but I think it sorta tells skeinforge to add more into the Z increase at the end of a layer then drop back down for the start of the next layer.

Inset: Enabled. Another one I’m not to sure on but I think it controls tweaks on how to remove overlapping bits of the print that will likely cause blobs. I just leave this at the default values.

Jitter: Disabled. You know that extra little blob of plastic you sometimes get at the point where the Z goes up to the next layer? This module causes Z to move up in a different place on each layer, which spreads out those blobs.

Personally, I don’t get those blobs too much any more and they don’t really bother me anyway. I leave this module disabled but feel free to enable it, if you want.

Lash: Disabled because I haven’t gotten to mess with it yet. This seems to be about controlling backlash – that bit of lag you get when the X or Y steppers quickly start/stop/reverse. I suspect it won’t work well on a Makerbot but haven’t tried it yet.

Limit: Disabled but this is near the top of my list of things to mess with. Yet another module I’m fuzzy on but it seems to control the maximum feed rates of the gcode.

In particular, I’m eyeing the Maximum Z Feed Rate. If upping this value actually makes the Z stage move faster, it will go on my list of things to always enable – it should (might) help a lot with those blobs you get when Z moves up to the next layer.

If you want to enable this and see if you can make your Z blobs disappear, make sure to disable Jitter first.

Multiply: Enabled. Multiple is another one that rules. Enable it, set both Number of Rows and Number of Columns to 1 and skeinforge will automatically make sure your object is centered and on the platform. Very helpful.

If you want to print multiply copies of an object at once, you can mess with the rows and columns values – this is useful on small objects, since they tend to print too quickly and have heat problems. (which Cool would also help, if we had stepper extruders or well-controlled ooze)

Oozebane: Disabled. Oozebane tries to limit ooze (the extra strings you have to clean up post-print) by shutting the extruder off a little early. It can also turn the extruder on a little early if you have lag between when the machine is supposed to start and when it actually starts.

Probably very useful but I think it’s tricky to configure correctly and don’t use it.

Preface: There’s no enable/disable on this one. Take the defaults.

Raft: Enabled, even if you don’t want to print rafts. Actually, I’m not sure this always needs to be enabled now, since the temperature settings moved to a different module – they used to be in raft.

I just leave it enabled anyway. If you don’t want to print a raft, just set Base Layers and Interface Layers to 0 instead of disabling it.

Speed: Enabled. Two more of the Big 4 (or 5) settings live here: flow rate and feed rate.

Splooge: Disabled. Yet another module that I’m not to sure on. Seems related to oozebane.

Temperature: Enabled. This is where you can tell skeinforge to use different temperatures for different parts of the object. Note the Cooling and Heating rates in this module – if you use different temperatures for different parts of the object, these control how long skeinforge will orbit between the different parts. (I use the same temp for everything because I don’t want it to orbit)

Tower: Disabled. Another good (or at least interesting) module that I don’t use. Say you want to print ”I I” standing up. Tower tells it to print multiple layers of one leg then drop back down and print multple layers of the other leg.

This is useful because you’ll have a lot less ooze between legs but beware using it on objects that have small legs – using tower on objects like that will encourage heat-related problems which, IMO, are worse than ooze.

Unpause: Disabled. I think this tries to bump up your feed rate a bit to compensate for processing delays on complicated objects.

If you have problems printing things like bolt holes or other feature that have lots and lots of little turns in them, you can try enabling this. Beware having your feed rate times the Unpause “Maximum Speed (ratio)” being higher than your maximum useful feed rate, though – if you speed things up too much, your print will end up worse than before.

Widen: Disabled. Yet another module I’m not sure of.

Wipe: Disabled. This one’s good if your ‘bot has a toothbrush. This lets you send the bot to some particular position at the end of every layer so the extruder nozzle can get cleaned off.

If you set the Wipe Period to some big number, it should do this only at the very start of the print, before it does anything else. If you’ve got some sort of brush mounted in your bot, this is useful for auto-cleaning up the test extrusion.

That’s it!

Next up: Creating a new profile…

Thanks Dave!

1. Photo courtesy of Great Beyond [↩]

Rocketboomer Ellie stopped by the MakerBot Botcave to chat about innovation. It’s so great to see Rocketboom still going strong, they inspired me to get into video in 2004!
Two of my friends, fashion designer Diana Eng and artist Marius Watz are also in this video! Awesome!

Naked pull-less cord

I had a wildly productive Saturday morning this last weekend. ¹  I tore through my honey-do list as if it were made of paper. ²  For those of you unfamiliar with the concept, a honey-do list is a list of things your spouse gives you to fix, take care of, or generally address.

The way our list works is that my wife can add anything she wants to the list and I can ignore it all week long.  Come the weekend, I put forward a good faith effort to resolve as many of the things on the list as I can.  There’s no need to knock them all out, just make a dent.  My wife likes this system because she knows whatever she puts on the list will be taken care of one day.  I like this system because I know there’s a finite number of things, and I can do whichever ones I want whenever I want.  We both like this system because we can see the progress being made on things that need to be done around the house.

One of the things added to my list last week was to get a new pull for our kitchen mini-blinds.  The cheap plastic pull had broken or come off at some point and my wife wanted a new one.  The only requirements were that it “just work” and not be ugly.  No problemo.

I designed a simple cylinder with a hole in the top and knocked it out in clear PLA in 6 minutes flat.  For those of you who don’t print in PLA, it tends to hold its heat and stay malleable longer.  This is a mixed blessing.  On the one hand, the cylinder walls were so thin that they weren’t fully cooled before the next layer was put down, making the next layer up squish the lower layer slightly.  (This can be mitigated with any number of Skeinforge tricks such as the Cool or Orbit settings or using Multiply to create more than one instance at a time.)  The result was a slightly twisty, sculpture-esque tower.  Although the gloppiness of the molten PLA obscured the hole at the top of the cylinder, the part was still so hot that I easily widened the hole with a pair of pliers before it had a chance to cool.  This made for a very easy installation. ³

Printed PLA pull

I was fine with using this as a proof of concept, to make sure I had the basic design down, but my wife liked the twisty nature of the pull and that’s what we have on our kitchen mini-blinds right now.  One of the coolest things about this fix is that I was able to design and fabricate a fix faster than you could yank a part off a shelf, all without leaving my house.⁴

Bonus real conversation:

MakerBlock: Wow, honey.  You were totally right to insist I purchase a MakerBot last year!

Mrs. MakerBlock: <eye roll and grin>  Yes, I’m very glad I insisted you purchase a MakerBot.

1. At least, wildly productive for me. [↩]
2. Which, in fact, it is. [↩]
3. Step one: Untie knot in mini-blind cord.  Step two: Thread new mini-blind pull onto cord.  Step three: Tie knot in mini-blind cord. [↩]
4. A MakerBot might not necessarily be the best fix every home repair.  However, it is the best way to get me to start working on a home repair.     [↩]

Halloween Hero!

This was a big weekend for a MakerBot dad like myself.  My daughter wanted to be a witch for Halloween and so we got her a costume, complete with small plastic broom.  The broom came in two parts that screwed together.  While letting a kid play around in their costume even when it’s not actually Halloween is part of the fun, it wasn’t long before she managed to break the broom right in the middle.  The plastic screw had broken off the one side, while still stuck in the other.

No problemo!  I measured the broken parts, thought of a fix, and created a workable digital model in less than five minutes.  The part took about 30 minutes to print (in PLA, since that’s what I had loaded in my 3D printer).  It consists of a plastic cylinder with notches where the pins in the broom would fit.  This would keep the part from rotating or sliding out of place.  Fitting the part, wrapping it in duct tape for strength, and then again in black electrical tape to smooth out the wrinkly duct tape and blend in with the color of the broom took less than minutes more.

Sure, I could have fixed that broom with little more than just the duct and electrical tape.  However, I would be virtually guaranteed that it would have been bent at that join before the morning was out.  I’m pretty sure that particular joint is the strongest part of the entire broom at this point.

Super-hero dad tip: Tape glow sticks up and down the broom handle for extra visibility while trick or treating and for a seriously awesome witch’s broom. 

Skeinforge! Go!

Dave Durant has done it again with a second post in his series about configuring Skeinforge for your 3D printer.¹ Skeinforge has been integrated with ReplicatorG versions 0017 onwards.  His guide about the five most critical Skeinforge settings is extremely helpful.  Lock down these five settings and off you go!  Once again, take it away Dave!!!

Unlike the previous post which was sortofa mini Skeinforge dictionary, this one is about some parameters you’ll actually find in Skeinforge. This was written with Skeinforge 31 (“10.09.21″) in mind but anything past version 20 or so should be pretty close. Hopefully future ones will be, too.

Although there are many, many parameters in Skeinforge, I think there are 4 or 5 really important ones that make up the foundation of a profile. A good portion of the other parameters in Skeinforge are ones that reference these parameters (usually with that darn “over” word in there) so if you get any of these 4-5 big ones wrong, it will have a ripple effect. More than any other settings in Skeinforge, these are the ones I try to get right first when making up a new profile.

Layer Thickness: I always think of this as “layer height” though skeinforge never calls it that – it uses the term “layer thickness.” I’m going to stick with layer height because I think that’s a less ambiguous term (“thickness” can mean vertical or horizontal and “height” is really just vertical, so I like “height”).

A MakerBot prints an object by drawing (with plastic) one layer, moving up a little then drawing another layer. Do this enough times and put down enough layers and you’ve got yourself a real, physical object.

Layer height is simply the height of each of these layers. You don’t really get to print an object with different layer heights - you pick one value and that’s the height for every layer in the object.

In a lot of ways, layer height equals resolution. Say you tried to print something like a wine glass but could only do it with a 1-inch layer height. If you think about a wine glass sitting on a bar (or table, bedstand, whatever) and the bottom inch of it, there’s a whole lot of detail in that one inch – there’s a wide base, which tapers tapidly to the stem which goes (mostly) straight up.

If you tried to record the description of a wine glass in 1-inch resolution then play that recording back later, it’d be complete crap – you probably couldn’t even tell it was supposed to be a wine glass because there’s just not enough resolution to accurately describe the object. For something like a big, perfectly square 3-inch cube, that sort of resolution would probably be fine but it just doesn’t work for objects with lots of detail.

As the layer height decreases, the resolution increases.

Flow Rate: This is how fast the extruder motor is turning and is frequently also called PWM, Pulse Width Modulation. PWM is like turning a light switch on & off really quickly to simulate having a dimmer dial – by varying the ‘width’ of the how long it’s on, you can change how bright it appears. Flow Rate PWM is the same thing but with a motor instead of a light.

For DC-motor extruders (which includes just about everybody with a MakerBot), a value of 255 means full speed or about 2 RPM. With a 10mm gear on the end of the motor shaft, that 255 or 2RPM means a length of ~30mm of filament going into the extruder every minute.

The really, really important bit to remember about flow rate is that this value is only about how fast the extruder motor is turning and how fast a given length of plastic is going into the extruder – it’s not the same as how much plastic is coming out of the nozzle. It’s all about the length of plastic going in.

If you’re trying to get really high quality prints out of your makerbot, this is extra important because when you buy “3mm filament” you’re actually getting anything from 2.70mm filament to 3.10mm filament. If you’ve got a skeinforge profile all nicely dialed in with a 255 flow rate on your 2.70 filament then you try using it with 3.10mm filament, you’re going to be pretty disapointed. Why? Because a 2.70mm filament has a ~5.7mm² area (cross section) and a 3.1mm filament has a ~7.5mm² area; it’s +30% more plastic per length of filament and, again, the flow rate is about how fast a given length of plastic is being fed into the extruder.

All that said, it’s important to keep in mind but in reality you’re not likely to end up switching from 2.70mm filament to 3.10mm filament – I’m just using the extreme examples to make the point. If you always buy filament from the same place, they’re probably shipping stuff that’s all pretty close to the same diameter.

One aspect of flow rate being all about the amount of plastic going into the extruder is that layer height has a much smaller impact on build time than you might think. Imagine filling up two identical boxes from a tube of liquid plastic that flows at a constant rate. On one box, you move the tube around really, really quickly so it makes lots and lots of tiny layers. On the other box, you move the tube around slowly and it makes a smaller number of thicker layers. Which one fills up quicker? Trick question! They fill at the same time. This is also true when printing – the difference in build times between low and high layer heights (on the same object with the same filament and PWM, etc) is the difference in the objects non-printing times. Non-printing times are things like travel and orbit. For something easy to print like a solid cube, there’s little non-printing time so the difference in pretty small. For something complicated, it can add up. Either way, it’s the difference in non-printing times.

You can adjust the flow rate from 1 to 255 but with DC extruders (again, that’s most of us) the nature of the beast limits us in the what numbers are actually usable. On my machine, I can get down to about 180 if the machine is nicely warmed up. Anything below that and the motor just stalls and nothing comes out of the nozzle.

If you’re really going after quality or resolution, the minimum flow rate you can use on your machine will be a good value to know. You can test this out in the ReplicatorG control panel by putting values in the ”Motor Speed (PWM)” box and trying to extrude. Note that this minimum-flow rate value may change a bit when you change filament stock, if the diameters are different and may also vary based on the temperature you extrude at and, believe it or not, the color of your filament – these probably won’t be huge differences but if you’re going for the best quality you can get, they’re good to be aware of.

Feed Rate: This is how fast the build platform moves, in milimeters per second.

A good number to know for your machine is the maximum feed rate you can use without losing quality. The ‘standard’ MakerBot profiles are usually around 30mm/second or so but you can, if you want, run the machine a lot faster – I’ve printed at over 65mm/second.

Though having the build platform zip around at +60mm/s is pretty cool (and a little scary, the first couple times) to watch, it’s probably not what you want to do. For one, it doesn’t really make the print complete any faster – that’s largely about flow rate and the non-printing moves that Skeinforge does.

Another issue is that it usually degades the quality of the print. The reason for this is the way the Cupcake X and Y stages are driven. In particular, the belts that move them around will always have a little backlash (lag when they reverse direction) no matter how tight you make them and the faster you go, the more they will vibrate – both of these things get transfered to the print and degrade quality.

Here’s an example of printing at a feed rate that the machine can do but is too fast for it to do nicely. This side of the object should be all straight, except for 3 holes and a circular indentation (it’s a Brutstruder at around 60mm/s, if you’re curious) in it. Note the waves or echos around the holes and the dent.

Although turning up the feed rate doesn’t really help build times, it’s something you’ll probably need to do if you start trying to increase resolution by doing stuff like turning down the layer height or using a smaller nozzle.

Keeping your belts & bolts tight and your X & Y rods oiled will help raise your maximum usable feed rate. You can also upgrade your X and Y stages to use bearings instead of bushings – not only will upgrades like that make your bot a lot quieter, they will also move more smoothly, which will help lessen the vibrations. If you’re interested in such upgrades, search through Thingiverse for “mendel inspired”.

Feed rate is what I consider to be the number one tweakable thing in a profile. That is, if you’ve got to change one of the settings listed in this blog post to make a profile come out right, this is the one to change if possible. More on that later..

Width Over Thickness, two different kinds: I don’t really care for this term, Width Over Thickness. Math scares me and it always makes me think I have to do math. I think a better term would be “Aspect Ratio”. Yes, it has the mathy word “ratio” in there but aspect ratio is a pretty common term so it’s not as bad. You can also think of these values as “width over height”, which I think would also have been a better term than “width over thickness.”

In short, these values tell Skeinforge how wide the filament being laid down is, in relation to how tall it is. If you have a layer height of 0.25mm and width over thickness values of 2.0, it’s going to generate gcode that works just right if the threads are 0.25mm tall and 0.5mm (0.25mm * 2.0) wide.

Skienforge has a number of parameters labled “width over thickness” but the two big ones are the Perimeter Width Over Thickness in Carve, how wide perimeter lines are vs layer height, and Infill Width Over Thickness in Fill, how wide infill lines are vs layer height.

In general, you should have the same value for both of these parameters. If you’re really tweaking the most out of a profile, you might want to change one just a little bit to make things come out perfectly. I think that if you have a difference more than about 0.2, something is wrong in your profile. I’m tempted to say that if the difference is more than 0.1, be suspicious.

Changing the width of threads has a big impact on two things. One is that higher numbers have more adhesion than lower numbers – they’re better at sticking to the layer below them. Picture two big circles of tape, side by side and touching. They have a puny contact area and won’t stick to each other very well. Now sorta squash them a bit so they are more oval than circular – more contact area, better stick.

The other big impact is on the feed rate. Picture the melted plastic coming out of your nozzle. If the build platform isn’t moving, you’re just going to get a big blob of plastic. If the build platform is moving really quickly, the plastic is going to stretch out and be a thin line, probably thinner than your nozzle size. If the build platform is moving slowly, you’re going to get a fat, hopefully non-blobby line.

If you raise the width/thickness values, you probably need to drop the feed rate down to make the threads come out wider. This is good news, especially if things like tiny layer heights have driven your feed rate up, but there are limits to it. If you raise width/thickness enough, you’ll have to go so slowly that the plastic will start blobbing up around the print head – beware width/thickness and layer height values that make the thread width more than the width (outside diameter) of your nozzle. That hard, brass nozzle doesn’t just spit out plastic – it also helps flatten & smooth out the tops of threads.

I had planned to yak more about the interaction of these big settings but this is getting pretty long. Reeally long. Too much coffee today or something. I’ll save that stuff for the next post..

Next in this series: Configuring Skeinforge: Moving’ on up!

Tune in next time when Dave Durant gives us another primer on the latest version of Skeinforge!

1. Photo courtesy of Jaycoxfilm [↩]