Category Archives: FAIL

Plutarch 1.0: finished, still not a robot

After this stability test it’s pretty safe to say that Plutarch is 97% ready to make his debut at J’s Halloween party later this year.

So how did we get here? When last we left our hero he was a naked 3d printed shell enclosing a buggy and unstable collection of wires and electronics.

Rubber cementing the feathers to the body was strikingly straightforward, the only caveat being that one needs to layer the plumage and make sure that none of it gets into the spaces between moving parts.

fledging collar

I ran out of feathers during the fledging process so he’s still got three percent’s worth of bald spots to fix during the 138 days before Halloween.

Epoxying the googly eyes is simple enough. Next time I’ll include 3D printed eye socket markers so I can be sure that Plutarch isn’t walleyed.

googly eyes

Here’s the thing about googly eyes. You can’t buy just two. You have to get a whole mess of them.

Epoxy is enlisted once again to affix the wing feathers to the body, as they’re too heavy to attach with rubber cement alone.

fledging wings

A clever designer would have included tail feather mounting holes in the original 3D printed body, but unfortunately no clever designers showed up to work on body design day. So Plutarch got a few aftermarket holes drilled in his rump.

drill

You may feel some slight pressure, Mr. Plutarch. Please try to relax.

OK! So! Forty-five minutes of fledging and butt-drilling hence, Plutarch’s ready for his big reveal to the family.

too sexy

Possibly too sexy for your cat.

You can see an original Pirate Parrot Accessory in the background of the photo above– one of his duplicates was cannibalized to provide most of Plutarch’s feathers. I bought a bag of blue turkey flats to compensate for the feather shortfall and still didn’t have enough.

The kids (and, of course the Lovely and Talented Mrs. Zheng3) have been watching Plutarch’s progress in bits and bites for the last few months, and they’re assembled in the kitchen for Opening Night. Plutarch is perched, powered up, and ready to go. I hit the button on his remote that makes him shake his head and… BAM.

broken

To everyone’s horror (except the cat, who gives approximately zero f*cks) Plutarch torqued himself off my shoulder, broke a foot, and snapped his battery cables.

Gah. I pinned the broken foot with a couple of epoxy-coated finishing nails and set the patient aside.

Another setback. We were so close.

The epoxy cure delay allows for a few hours of self-reflection. Why am I doing this? Is making an animatronic parrot really best way for me to spend my limited time on earth? How do magnets work, anyway?

Magnets. We need moar magnets. Better-positioned ones, too.

The original perch design put the magnets on the underside of a steel can lid, relying on luck to link up with the magnets embedded in Plutarch’s toes. Fearful of another catastrophic and embarrassing fall, I moved the magnets to the top of the lid and made sure they’re aligned as closely as possible with their mates above.

magnets

The video at the top of the post proves that this time Plutarch stayed put.

But. BUT! Even after all this improvement, Plutarch, while ready for primetime performance, is still not a robot. He’s best described as an animatronic parrot. Roboticization– the addition of sensors and the ability to respond to an environment– shall have to wait until Halloween 2017.

So here’s what I’ve got planned for Plutarch’s next year:

  • improved internal accessibility
  • 2-axis head movement
  • articulated beak
  • audio
  • some kind of sensor ability, for crissakes

#staytuned.

Tales of Plutarch’s earlier incarnations can be enjoyed here and here.

Lao Zheng out.

Schrödinger’s Shipping

TL;DR: Some backer rewards are shipping. Others are not. Be sure to set a minimum layer time for spherical prints.

Kickstarter update: The first wave of physical backer rewards has been shipped and is happily trundling towards Strongholds backers. This batch is mostly Dice Plinths and a couple of mini-citadels. Eyries and full-sized citadels are (slightly) delayed, for reasons soon to become clear.

But first! We’ve got some photos of backer prints to share, yes we do. Here’s a FlashForge Creator print of the modified Citadel straight outta Pinshape. Some backers had expressed disappointment that their printers couldn’t handle the citadel’s height, so I used Maya’s Multi-Cut tool to split the mesh at a mortar line.

flashforge

Thanks to backer Nagromic Strongbow for this one.

In other, less enthusiastic gnus: we’ve hit our first official backer reward fulfillment snag.

Booooooo.

Hundreds of hours of nonstop printing with abrasive Glow in the Dark filament have worn out our print nozzle. Worn nozzles lead inevitably to failed and/or low-quality prints, up with which the quality control minotaur at Zheng Labs simply will not put.

failed cap

Enjoy the schadenfreude of failed prints? See more at http://3dprinting.fail

The above print failure is hilariously catastrophic, but these holes and blisters in the Eyrie caps are by far more pernicious:

caps

Here one can see the importance of paying attention to your slicing software; a good slice can make the difference between a mediocre and a solid print. What’s happening here is that the hot plastic doesn’t have a chance to cool before the extruder starts laying down the next layer. There’s some inter-layer fricton going on, and the newly extruded layer is dragging the still-gooey previous layer around with it.

Sometimes this process creates a tiny hole in the top of the cap. Smetimes one can’t tell there’s an issue at all. Must have something to do with ambient air temperature and viscosity thresholds or sunspots or something.

Naturally, I didn’t stumble upon the solution until I’d printed a dozen or so faulty caps. All you’ve got to do to fix an error like this is give each layer a chance to cool. In Cura, the setting’s under the Advanced tab– change your Minimal Layer Time to some value other than 0.

I’d rather delay shipping than send out prints of anything less than the highest quality that I can manage, so production’s been suspended until a replacement part arrives. Once it gets here I’ve got maybe 20 caps to reprint at 3 hours a pop.

While this situation is terribly frustrating, Strongholds is still mostly on schedule and, more importantly, shows no signs of devolving into one of those I-took-backers’-money-and-built-myself-a-house-in-rural-Saskatoon Kickstarter fiascoes.

On the subject of fiascoes, I’ll leave you with this deceptively attractive Ultimaker print from backer Michael Zions:

ultimaker

Looks good, right? A thousand quatloos to the Ultimaker team for designing a 3D printer that can do double duty as a diffuse lightbox.

Unfortunately this print crumbled to the touch as soon as Michael removed it from the bed. If you’re having this kind of trouble with your prints, use the citadel_alt STL file instead of the original.

You’ve Got Fail!

TLDR; I released suspect geometry into the wild. Also 3dprinting.fail is now a thing.

failed citadel

Despite rumors to the contrary, it’s not all mimosas and backrubs here at Zheng Labs. There are times, thankfully few and far between of late, when we fail to make the proper sacrifices to the 3D printing gods and models go kablooie on the print bed.

To wit: these Eyrie caps that bought the farm before they could reach their full potential.

eyrie cap

The bottom surface of the Eyrie cap is a thin circle and occasionally won’t adhere to the build platform. Double woe if one gets ambitious and prints multiple models at a time; one failed print can catch on the extruder nozzle and get dragged into its doppleganger, causing a calamitous cascade of failure.

The printer, being blind, deaf, and completely lacking in agency besides, has no idea that this failure is happening and if left to its own devices will merrily continue extruding hot plastic into thin air.

Fail.

A quick primer on 3D printing for those of us who don’t live and breathe this stuff. One can break the process down into three basic steps, about which one could proceed to write volumes of details.

Step 1: Creation! By some mystic means, a 3D model is created. My weapon of choice in this arena is Autodesk Maya because I use it in my day job, but there are scores if not hundreds of software packages that will export a 3D printable model. Amazingly, even Minecraft can do it with the right mods. Tell your nine-year-old niece.

Step 2: Slicing! Before it can be printed a model must be divided up into a vertical series of horizontal layers. If you’ve accidentally introduced wonky geometry in Step 1 the slicing process will create toolpaths that kinda work, but might result in a less-than-optimal print in the real world. More on this and a mea culpa in a bit.

Step 3: Printing! A tireless bot with a melted plastic-filled hot glue gun draws successive layers on top of each other. The plastic cools and before you can say Bob’s your uncle you’ve got a brand new Dice Citadel. Usually it works, and sometimes this happens:

The reasons for print failure are legion, but I’m guilty of letting an avoidable one slip through my quality-control network with the Classic Citadel. Back to slicing and the aforementioned mea culpa:

Here at the lab our preferred slicing software is Cura. Cura has been churning out perfectly usable G-code for months and I’ve printed dozens of citadels with nary a problem. But here’s the rub, mein grübenses: not everyone out there uses Cura.

Printing problems started cropping up once the Citadel was released into the wild. Users of slicers Simplify3D and slic3r were shocked to find that their printed citadels, walls too thin to withstand an assault even by Marshmallow Mangonels, were crumbling to the touch– see the photo at the top of the post.

Mea maxima culpa, I really should have run the models through several slicers before releasing them. Parallels may easily be drawn between the current state of 3D slicing and the early Web when different browsers would render the same HTML in completely different ways. 1996 was a great year for flannel, but damned if I enjoy the return to crossing one’s fingers and hoping that WYS is truly WYG.

Wizzywig. Now there’s a term I’ve not heard in a long, long time.

But! Thanks to the heroic and dogged troubleshooting efforts of Strongholds backers Chris Yohe and Nate Johnson, the problem’s been fixed as far as I can tell. I’ve uploaded a new Citadel to Pinshape and alles gut. Print, my friends. Print LIKE THE WIND!

Other backers have been busy printing Eyries and plinths a-plenty. If you’ve got a print you’d like to show off to our little tribe of medievalists, send it my way.

backer prints

Unexpected creative output: There’s plenty of downtime to be filled while the printer is producing Citadels and Eyries for physical rewards backers, we’ve been thinking about failure a lot lately, and here at Zheng Labs we’re certainly not ones to let a good domain name go unclaimed.

So I went and registered 3dprinting.fail, polished up my JavaScript-stealing chops, and made a nice slideshow of some of the spectacular messes my printers have created over the years. Tell your friends. Tell your mom. Tell your mom’s friends at the next euchre tournament.

beast fails

(That model’s a Beast Token and you can grab your own at The Forge.)

So! That about covers it for this week. Back to printing backer rewards and obsessing over the Next Thing. Here’s another wee teaser for that project, which I’m hoping to release within a month or so:

wee

Who loves ya, baby?

Lao Zheng out.

11 Reasons A Heartwarming Kickstarter Failed: #2 Will Give You The Feels

Well, that certainly could have gone better.

crash

The funding period for Faire Play 2 has ended and we’ve come up short. Like, way, way, way short. You can still purchase the models for the Faire Play 2 chariot, gladiatrix armor, and Empress Makeover Kit over at The Bazaar, but the Kickstarter’s gone off and joined the choir invisible.

Before we proceed, allow me this big, red, throbbing, baboon-ass of a caveat. Nowhere is it written that one is entitled to a successful Kickstarter. You buys your ticket, and you takes your chances. Zheng Labs rolled the dice and came up with a big old nothing-burger this time. C’est la vie.

But you can’t hit if you don’t swing, right? Build it, as is told in the Book of Costner, and they will come.

And come they did, in droves. Faire Play 2 received a tremendous amount of favorable media coverage in online publications across the spectrum, from The Huffington Post to Newsmax. There are Kickstarters that would kill for that kind of exposure.

There are rumors that the potato salad guy did kill for that kind of exposure. But only rumors, mind you.

And yet, despite a truly inspiring amount of publicity and online good will, Faire Play 2 missed its goal by roughly eighty percent.

Ouch.

But as Grandma Zheng always used to say, if you’re going to fail, you might as well fail spectacularly. So there’s that.

What happened? Let’s explore some possibilities.

#1. In hindsight, launching a Kickstarter contingent on the cooperation of cats might not have been the smartest business plan.

Everybody knows there’s no way you’d ever get a cat to pull a chariot. Ever. And even if you did the cat would find a way to kill you in your sleep. The Venn diagram of people with Barbie dolls, 3D printers, and cats is also vanishingly small, which further limited the number of potential Faire Play 2 backers.

cat reveal

#2. People are used to getting amusing content for free, and converting Kickstarter video views into Kickstarter backing isn’t a viable strategy.

I had a tremendous amount of fun creating the Kickstarter video for this project. My hope was that people would watch the video and then think, “Yes! Imagine what this guy could do with greater resources. Here’s two bucks.”

Alas, in a world where hilarious dashcam vids of road-raging mascots are free for the taking, asking people to selflessly contribute to one’s artistic endeavors is hoplessly naïve.

thumbs down

Still, STILL! Many people did contribute to the project for precisely that reason! I’m thoroughly grateful to the hundred or so idealists who supported Faire Play 2. You folks are the best, and the world would be better off with more people like you.

Also, apparently 3d printed lithopanes are a terrible idea for a backer reward. Who knew?

#3. Nobody plays with dolls anymore.

It hasn’t been a great year for Mattel. Barbie sales crashed 21% in Q3 of 2014 and have been sinking for three years straight, and Mattel’s CEO left for greener pastures.

I’m doing what I can to revitalize Barbie’s image, but I can’t imagine a kid born after 2010 picking up a doll when there’s an iPad nearby. The market for aftermarket Barbie accessories, already quite nichey, gets smaller every day.

Nonetheless, I couldn’t help but model this Barbie-compatible tire iron. In this house, Barbie changes her own tires.

tire iron

It’s free! Download it here. You’ll have to heat-deform it after printing to get that nice bend in the shaft.

My guess is that at this point in the article some folks are wondering why there aren’t 8 more reasons the Kickstarter didn’t succeed. It’s because the 11 in the title of this post is in base-2.

So, what’s next?

Oh, man, I am so excited for the Next Thing.

(I’m always chasing the Next Thing. Character flaw. Greatest strength, greatest weakness, and all that.)

I’ve been working on the Next Thing for the last couple of weeks and hope to announce it soon. Here’s a wee teaser image for you.

teaser

#staytuned and #watchthisspace, my friends. This temporary setback has got me generating all kinds of new ideas. The ride’s only going to get more fun from here on out.

And big, big, BIG thanks to everyone who supported the project. I’s got good peeps.

Newest Way To Fail

Wow. Totally didn’t see this one coming.

So I’ve been furiously prototyping models for Kickstarter #2. The keystone model for this next project is what we in the business call a giant-ass bucket of geometry. The gCode works out to something like 90MB (!) of data.

Models like this take forever to print, so rather than try to maintain a USB connection between the computer and the printer it’s generally best to print stuff like this from an SD card. A little interruption like a monitor going to sleep can kill the USB if one isn’t careful, and then you’ve got yourself a half-finished print.

The model started printing at 9AM yesterday after some very careful calibration and painters’ tape finessing, and everything was going great until about 2:30AM today, when I came downstairs to find this:

glitch

Mother pus bucket.

I’ve never seen this before, so to teh Googles we go. Turns out, at least in the opinion of some electrical engineers, the Replicator1’s Mightyboard generates a lot of electrical interference can cause the bot to fail over long prints.

Live and learn, back to the drawing board, and a host of other try-try-again platitudes must suffice to get me through the day.

Curiasser and curiasser

Faire Play backer rewards are all I’m thinking about these days. The Athena Makeover Kits are done and should be winging their way towards my midrange backers shortly, and now I’m printing a suit of field plate armor for one of my top backers. My printer is working on a lovely pair of sabatons in the background as I type this.

Nothing’s ever completely straightforward at this level of 3D printing, and of course now I’ve got a strange little gremlin that didn’t surface the last time I printed a cuirass: I’m getting a small line up the side of the print where my printer’s print head stops, lifts, and begins a new layer. Take a look at the cuirass on the left.

cuirass lines

I’m doing my best to make this suit as nice as possible and I’m aesthetically offended by artifacts of the printing process like this one.

What’s wrong here? The geometry hasn’t changed since the first print, the filament’s exactly the same, and as far as I know I’m using the same slicing algorithm. I have replaced a worn-out print nozzle, but I can’t imagine that’d be the source of the issue. I’ve changed every setting I can think of and I’m still getting this weird little printing track.

Truth be told, it’s a fairly minor aesthetic problem in the grand scheme of things and now that I look at the original armor I do see some traces of this artifact there, too.

Even if it can’t be completely eliminated, it can at least be put on the back side where it’s less visible. The best way to do this is to just rotate the model 180° around the Z axis before printing. Easy peasy: take a gander at the cuirass on the right.