My first post on this topic was devoted to the Digital Side of designing a part. It had lots of images, a fairly waffly bit in the middle where I admitted I hadn’t remembered to take images or notes and ultimately stopped before I put the rubber to the road. What a cop out.
In this second part, there is still a bit of digital talk while I go through how I get my digital design from farm to table, but there’s going to be a hell of a lot of table too. There may also be mixed metaphors.
I should also warn you, this is a bit of a long one. On reflection, it should have probably been two posts.
This post contains 5 chapters (or steps, as it were):
In Step 1, I go in to a bit of detail on the digital steps I use to get to physical.
In Step 2, I put all the pieces together.
In Step 3, I give the project some smarts.
In Step 4, I attach it to the machine.
In Step 5, I cover some of the “corrections” I had to make.
As modern makers we’re really only presented with two approaches when turning our digital ideas into a physical reality, Additive manufacturing and Subtractive manufacturing. This project would suit either, I could have 3d printed these parts, CNC milled them or Laser cut them out of stock. All are tools I have available fairly readily at the Connected Community Hackerspace here in Melbourne.
I guess I like the idea of vaporising plastic though, because friken laser beams is where I went without even a thought.
Step 1 – Speaking the laser cutter language.
Every laser cutter I have ever used has spoken a different language. The two at CCHS and my own (broken) unit all run a 3D printer RAMPS board and the Marlin software. Technically they aren’t different languages but it’s at least a different dialog.
Many commercial laser cutter solutions can take a vector image in a .svg file format and use that to produce your design. That wasn’t really mature when I first started on my manufacturing journey and I ended up buying a copy of a piece of software called CAMBam Plus.
Now I’m just addicted to the control. I can’t go back. I WON’T go back. You can’t make me.
It’s a somewhat complicated process. If you’re really interested I made a short video.
At the hackerspace we use Pronterface to interface with the laser cutter. It’s not really made for dealing with 2d shapes, but it does work.
And in no time at all I was lasering.
It took a few tweaks to get my design running smoothly with the laser cutter, so I ended up with 3 versions of the parts. That’s gonna hurt when I have to pony up the dosh!
Step 2 – Auto-Changer! Assemble!
One of the design features I included where small reference holes in order to make sure all the parts aligned as expected while the glue set. There was no reason to use anything fancy glue wise, so I used a brush on super glue for a little more control.
And here we discover the first little hiccup. The gears, she no worky. They kind of mesh, but there’s a barely perceptible slant on the laser and the tolerances are too high to work with this error. As cool as my gears look, I think they have to go.
All these little servos come with a range of arms that are keyed to the splines on the servo shaft. A few extra holes and some tie wire mean I can use these arms as levers to push and pull the thread catchers up and down.
Step 3 – It needs a micro.
“Bring me the brain, Igor!”
Fortunately, I had quite a range to choose from. Over the years I’ve collected pretty much every kind of Arduino, ARM or ESP chip, including quite a few boards I’ve custom designed. In this case I went for a simple Uno compatible board because it meant I could drop an Adafruit Servo Board on top and save myself a little effort with wiring up the servos. This board supports 16 Servos, so I guess that is now some kind of ridiculous stretch goal now? Hmm, maybe some kind of rotating carriage to keep the size down…..
Stay on target John. Stay on target!
Fortunately, turns out code was very simple. I butchered the Adafruit example and had something working in about 10 minutes.
Step 3.5 – Any sundries for your order sir?
Why thank you! Yes, I’ll have 5 orders of buttons, a Opto-sensor and a piezo buzzer.
Damn, I forgot to ask for photos with that. Oh well. I guess you can imagine from my tortured prose or something?
The 5 buttons perform the handy little function of letting Sarah select the order for the little gates to open, effectively programming the controller. Not that she ever does, it pretty much always shuffles along in a 1, 2, 3 pattern, but she could if she wanted to.
The Opto sensor also acts as a button, it detects when the carriage has passed the changer, providing a cue for the changer to switch.
Incidentally, a little more butchery to the code and the servos now move in parallel. That sped things up enormously.
One thing the original KRC-900 includes is a mechanical clicker that lets the knitter know when they can start winding the machine back, hence the Piezo buzzer. It plays a 2 second long “C” note after triggering the colour switch leaving you in no doubt that you’ve successfully changed. It also means the damn knitting machine now makes more beeps and if you’ve ever been near one of these things you’ll be aware that they’re pretty high on the beep quotient already.
Step 4 – Fix it to the machine.
This hacker recommends ACME hot glue and cable ties.
Step 5 – If at first you don’t succeed, bodge.
It’s a prototype, and prototypes are always a little rough.
Gears, you mentioned them already.
I’m not really happy with the lever solution as it means the top plate had to be half removed. I’ll think up something for the next blog post about that.
The carriage jams itself into the Autochanger
Whoops, the clearance required by the carriage is about 2 mm higher than expected (makes a note on schematic, jams spare piece of plastic in to raise height of unit.)
Like, 50% of the time the thread misses the guide rail. Did you mean to do that?
A bit of electrical tape wrapped around a cable tie will “fix” that. Better make another note for version 2 though.
The tab that holds the Opto sensor on the left of the unit is the wrong size, utterly the wrong place and generally causes havoc.
Well, that bit was always on the “suck it and see” list. I hit it with a file to get it good enough then used copious amounts of hot glue to hold the sensor in position.
Uhhhh, so where do these buttons go?
Oh yeah, I never really planned those did I. A piece of thin scrap plastic, a hot air gun to bend it over and even more hot glue will solve that!
Do you plan to just leave the brain hanging like that? Seems dangerous.
You’re probably right (hot glues it to the bottom of the servos and cable ties up any extra slack)
Should you really be asking yourself questions in third person.
No. I’ll give myself a thorough talking to later about that.
Step 6 – But it works right
Yes! Yes it does.
Tune in for another exciting episode in a few weeks time when I detail fixing my design. I’ll try to keep it short. Well, shorter at least.
Once again, when my design is complete I’ll update this post with a link back to my github.
$500 is STILL too much! How much have we spent?
Cash on the barrel-head boys!
$50 – materials for cutting 3 sets prototype parts on the Laser cutter
$30 – Arduino Uno compatible (had this one floating around)
$37 – Adafruit Servo Board (massive overkill, also from my junk box)
$20 – 4 Servos (*sigh* Junk Box).
$10 – Motor Speed Sensor control.
$3 – Piezo buzzer (Junk)
$1 – 5x Buttons (Box)
$?? – Super Glue, Wire, Bolts, Hot Glue, Cable Ties (Carefully ordered boxes)
Total – $151
There’s skin in the game now! Prototyping a design is always a little more expensive than a final product but almost all of these bits were out of my parts bin so this figure is a bit inflated. I favoured convenience over good engineering here and that really does add to the bottom line. Thinking about repeatability and mass production is a worthy topic and I think I’ll cover that in a post all of it’s own.
This Prototype is fully functional and Sarah has already used it to knit a few scarfs. If this was a one off I could stop right now and never look back, but I can do better!