Sometimes I wonder if I should change the blog’s name to “The Electromechanical Mercenary”, mostly because I keep doing things like this:



Those were developed in just under one and a half weeks, so they should count only as prototype. I am preparing a fancier version with hollow axles on all gears, better pneumatic mechanical advantage and overall higher quality in design, mostly because I can, but also to show-off at Eurosteamcon 2016


Some more shots of the beautiful model that wore it:


This is my wallpaper for the time being:


Lastly, the blue in the wings is high quality automotive reflective tape, so you must avoid taking pictures with flash, or this will happen:




Fairy’s anatomy.

Because I forgot to show the final mechanical wing engine:


Wire guide close up.
Interestingly, it was easier to mechanize the circular wire guides than the square ones.


Aaaand that’s it, for the moment.

Spineless II: Poor man’s ball joints.

Today I’m going to show you how the spinal prop ball joints where made:



  1. Threaded balls.
  2. Polycaprolactone, a.k.a. friendly plastic.
  3. Access to a lathe, or a friend that has one for a small turning job.
  4. 3 lip chamfer tool (and power drill/lathe)
  5. Clamp.
  6. Scissors.
  7. Pliers.
  8. Ball’s appropriate allen key.
  9. (optional) Plastic dye.

First of all, you must understand what you want to do. You want to encase the ball in some material wich allows it to swivel freely, more or less something with this profile:

Ball joints 014

Ideally, that would have the thickness of the support, but that presents a bit of a problem when the ball is already odd sized:

Ball joints 013

I don’t have a 5.4mm drill bit for the anvils to encase the ball. Also, this kind of arrangement won’t provide any ball centering, so it IS going to be off to one side or another.

The solution? make a flange on the anvils so they fit snuggly in the hole of the support:

Ball joints 015

That, of course, comes with it’s own set of problems, first of all, the odd sized holes:

Ball joints 016

Again, I don’t have a 5.6/5.7 drill bit at hand (I usually only have x.9 bits to use prior reaming an even sized hole). You could always modify the thickness of the support so it gives you a better hole size, but any variation will mess up that fit. Nothing that a bit of ingenuity can’t solve.

For now you can build the anvils, taking care of drillng the holes for the ball slightly smaller than the contact points (I could have used 5.5 mm in this case, but went with 5mm because I didn’t want to take any chances). Also, since you’re at it, drill and tap one of the anvils so you can screw the ball in there, and doesn’t move on further operations.

They should look like this:

Ball joints 018

Now comes the trick. Using a chamfer tool to eat away the corners, slowly fit both anvils until you can’t rotate the support around them (so their spacing when resting against the ball, matches your support thicknes, in my case, 4mm)

Ball joints 017

And they will look more or less like this:

Ball joints 006

Ball joints 007

Also, altough the drawings show a smooth bore, you must provide a means for the ball brace to stay in place, otherwise it will slip out of the support.
I just drilled a hole from the free end of the support and 2mm onto the cylinder’s body itself:

Ball joints 003

And now it’s time to assemble some ball joints!

First, melt some friendly plastic (I dyed mine black for aesthetic purposes only, be warned, it’s a mess, use gloves!) then loosely press fit it in the hole you are going to use:

Ball joints 004

Trim the excess:

Ball joints 005

Now, before everyting cools, submerge both the support with the plastic and the ball holder anvil in very hot water to ensure that the plastic doesn’t cool quickly and flows around everything:

Ball joints 008
(I couldn’t hold both and the camera at the same time ^^U )

Now, do both these steps QUICKLY!

Assemble the support with the hot plastic and the COLD (room temp.) anvil:

Ball joints 009

Then press fit all, use the clamp to ensure everything stays in it’s place:

Ball joints 010
Note how the plastic oozes from the hole, that’s good, it means it probably filled everything as supposed to.

Wait for it to cool a bit, then, grabbing only the support…

Ball joints 011

Pull the cold anvil from it:

Ball joints 012

If you look carefully, you can see the metal from the ball, as there is almost no material left between it and the oozing plastic from the cold anvil:

Ball joints 019

Using the pliers, just rip off the leftover. If you didn’t wait long enough, the soft plastic will deform and mess up the joint. If so, just melt everything up and start again, friendly plastic doesn’t mind it:

Ball joints 020

Assuming all went well, unscrew the ball with the allen key, but DO NOT, I repeat, DO NOT, wiggle the ball yet. Unless you waited a lot, the plastic in contact with the ball is still soft and will grab the ball and deform if you move anything. Leave it to one side to cool down and repeat the process. I was able to do three joints with a very hot glass of water, (didn’t had a thermometer to monitorize water temperature).

Ball joints 021

If you applied enough pressure, that bit of flash should come off easily:

Ball joints 022

Now you only need to break-in the joint. Move the axle to one extreme and then all around the range of the joint. That will loose it enough to move smoothly but still have very little play:

Ball joints 024

And that’s it! now you wave fully functional ball joints for normal temperature conditions. I suppose you could thread the exit hole from the support and thread a nozzle from a 3D printer to inject ABS plastic. But that’s delving into high temperatures and performances I don’t need at the moment.

And now, let’s watch it one more time in this glorious shot:

I bet you didn’t mind the vertical video. XD!

Also, remember I said you needed to provide some sort of anchorage for the ball brace to hang into? Here’s what happens if you don’t:

Ball joints 023



Unlike Alanis Morisette, my model does not abide to any man. (ahem, ahem) still, the name was fitting.

So, apart from the winged mechanism, this is what I have been up to:


Fully articulated spine “reinforcement” (wich didn’t reinforce a thing, it’s just a fashion accesory) for a Steampunk Fair. (Eurosteamcon 2015)

Apart from the M3 balls and the axles, everything was machined or cut @ home. 28.5 hours of machining in it. I even developed a technique to make room temperature ball joints at home.

See it in it’s full range of movement, it’s hipnotic!

A very collaborative model allowed the pieces to be adhered directly to her skin:

attachment 001

A different angle:

attachment 002

Unfortunately, I ended up with a schedule so tight I just had time to machine, no photos of the process or step by step or anything. Altough it’s basic machining and reaming. The interesting bit, (the ball joints) will have it’s own post.

Here’s how cool it looked:


Another shot with my secondary model:


Also, one with both the spinal prop and the fairy wing mechanism:


Even more busy!

But I can show you the prototype:


Details from the hardpoints for the wings:



Here’s how it works:

Basically, the first disc acts as hardpoint for the brass screw. The second disc gets actuated by the piston and moves the first wing (the one wich travels further). After 45º, the third disc is engaged by the second and starts traveling sinchronously, lifting the second wing to the midpoint. Finally, after 45º (for a total of 90º) the third wing (fourth disc) is engaged, and everything travels a further 30º.
As you can see, everything rolls on ball bearings, even the engaging pivot has two per disc. Mostly because I can. XD
The other wing pivot is just a mirror image from this.

And here it’s moving under power: (@3.5Kg pressure) The wing doesn’t actually move all the way up because the cable was too long, but it’s adjustable so there should be no problem in there, just some fiddling around with lengths.

Some stats:

Machining of the discs, base design, and ball bearing whire pusher took 26 hours of mill and lathe. (includes some pneumatic adaptors).

Solenoid assumption fail.

So, a while ago I was doing this steampunk clockwork mechanism for eurosteamcon 2013


The thing is, as it was all lasercut, I didn’t want to fiddle with a real escapement mechanism. I had not the time, nor willingness, to spend weeks tweaking such a mechanism.


I decided to make a simple solenoid driven escapement. I always save miniature solenoids from everywhere (who does not? XD ) So I had the ideal one around my bit box, from a cheapass single-use analog camera:


I would just use a 2N3904 and a 555 to make it tick.

The actuator stops a simple lever in the last reduction stage, allowing for a complete revolution in every tick. (reduction was 60/10 – 30/10 – 30/10 – 30 (162 to 1)

Did I say simple?, that was the idea, yes…

Some of you might already have notticed what went wrong. For those who don’t, look at the whire gauge. I completely failed to nottice that that coil was going to take a lot of oomph, so imagine my face when I had everything set up, and I could not make it go.

When I did realize, I made a simple measurement, and saw that at 1,5V, the coil required 1,36A peak current to energize (less than that, and the plunger didn’t had any kind of authority, so it would not move under load), way beyond the 200mA of the 2N3904.

And I didn’t had any fet’s around, or a power transistor…nothing, nada…I didn’t even had smaller whire to re-reel the coil for less consumption, nor the time (I had to deliver the mechanism in less than 2 hours!)

What did I do in the end? Just a simple magnetic flux brake:

FINAL 001That copper disc you see, has a neodimium magnet underneath. As the force induced is directly proportional to the speed at what the disk wants to rotate, when it tries to go fast, magnetic induction brakes it, and when the mechanism is nearly unwound and rotating slowly, it has no effect.

I might have to give it a new go someday.

For the curious ones, here are some other shots of it:



The mechanism accepted about 20 full rotations, with an endurance of about four minutes. Enough for photo and video shots.

Below, highlight of the different reduction stage supports in descending order: