Marc Brooker's Other Blog

Building My Own Watch: About Face!

2025-09-06T00:00:00+00:00

Building My Own Watch: About Face!

Dialing it in.

The next step I wanted to tackle in my watch making journey is making a dial: the face of the watch that goes behind the hands and movement. Dials are where watch makers show off their skill and artistry, with brands like JM Shapiro and RGM famous for intricate metalwork, especially the art of Guilloché. I don’t nearly have the skill (or equipment) to compete there.

But I do have a CNC machine, Rust, and a benchtop power supply. That’s the same as decades of skill and a 19th century rose engine¹, right?

The dial I wanted to make has an intricate design, that reflects the light in a way that it looks different from different angles. Something to catch the eye.

I played with a lot of different patterns: starbursts, spirals, weaves, weaves-within-weaves, waves, and waves-within-waves. Then I thought: you know who’s under-represented in the world of watch design? Q*Bert, the greatest of 1980s arcade heroes.

The idea was to engrave a world of Q*Bert-style cubes, with the goal of them looking 3D when they catch the light at the right angle. The first step was to write some Rust code that generates the GCode that the machine will follow when engraving.

Now, GCode is a full programming language, and there’s nothing stopping you from writing all of this directly in GCode. Except sanity.

GCode in-hand, we need to decide which material to use.

The easiest metal to engrave is brass. Bronze is OK. Stainless is surprisingly OK, especially compared to how much of a pain it is to do other work on. Precious metals can be great, but are a little out of budget for a first prototype.

That leaves the flashiest, SR-71est, strength-to-weightest, controversial food additivest, star of the show: titanium.

Now, my watch is not engineered for extreme temperatures. Instead, I wanted to use titanium for another reason: it’s super easy to make it go all kinds of fun colors.

Sidenote: Electricity

Before I write about anodizing titanium, I need to talk a little about safety. The technique combines electricity with water. Water intentionally made conductive. Mixing water and electricity has a reputation for killing hobbyists. Most of these accidents are from fractal wood burning, where folks build janky devices harnessing thousands of volts to burn patterns into wood. Wikipedia says (citing Richardson and Johnson):

A 2020 review noted that the mortality rate of fractal wood burning cases was “significant” and “exceedingly high”.

Fractal wood burning kills and hurts folks partially because electricity above about 1000V is weird. Rules of thumb like “plastic is an insulator” and “electricity isn’t just going to jump out and kill you” don’t apply. It’s no doubt possible to fractal wood burn safely, but I strongly recommend you don’t try²

What I’m talking about here isn’t that. Instead of thousands of volts, we have (at most) 25V. Instead of amps, we have milliamps. Still, precautions and care are always important in the shop.

Anodizing Titanium

Ok, back to my buddy Ti.

Titanium is the easiest metal to anodize in the home shop. As with aluminium, anodizing titanium coats the metal in a thin layer of metal oxides. In Titanium, this non-conductive oxide layer is super thin. As in single wavelength of light thin. This is where the magic happens: if you grow the layer just right, choosing just the right voltage, you can make the titanium look all kinds of cool colors by causing the light that bounces off the metal and the light that bounces off the oxide layer to cancel out.

This is similar to the structural coloration that makes jays blue.

The result is stunning.

The only hard part—and you can find a ton of tutorials online—is cleaning up really nicely. Even tiny traces of oil and dust wreck the finish. I do a wash with degreaser, a wash with acetone, and then a wash with isopropyl.

Engraving

The engraving was done on my little Tormach, with a 0.01” (250μm) ball tip engraver. I’m still not happy with the speeds and feeds for Ti. In brass, 200μm depth of cut, 300mm/minute, and 10,000 RPM produce nearly perfect results.

In Titanium, the same setting raises a really nasty burr. I don’t yet have a media blaster, but was able to clean up most of the burr with a brass brush and some elbow grease. You can tell in the photo above that it’s still not perfect, so I’m still working on better results right off the machine.

One thing that I have learned about engraving is that I get the best results breaking one of the golden rules of milling: get the damn chips out of the way. Eighty percent of the time hobbyist have a bad time milling (especially aluminium), it’s because they’re recutting chips. Here, though, I don’t flood or blast the chips away. Instead, I coat the whole part with anchorlube and let the chips hang out where they may.

Under all those chips, the live preview on the CNC control comes in super useful.

Work holding here is the old standby: superglue. It works great, most of the time, but is a pain to remove. I’m working on a design for a little vacuum fixture to hold dials. Once I have that, iteration should be a good bit faster.

Assembly

After engraving, deburring, anodizing, and cleaning, its time to assemble. The dial goes on the movement, the hands go on, and then the whole thing goes into the case.

Under the harsh light of my assembly area, it looks terrible!

But the final result, once assembled, is pretty good.

I’m very happy with that as a first attempt. I want to improve the legibility of the tick marks, add some color contrast, clean up the surface finish, and get the color more consistent across the surface. A long way to go, but happy with this as a first one.

Footnotes

Rose engines are among the coolest machine tools, maybe only beaten out by the pantograph.
Unless your argument is “I do this for a living and work with another professional engineer who will review my design and safety protocols”, you shouldn’t mess with thousands of volts.

Building My Own Watch: Threading

2025-08-25T00:00:00+00:00

Building My Own Watch: Threading

Tiny tool time.

In previous steps, you’ll have noticed that my watch case has two holes¹: one for the glass, and one for the mechanism to go in, closed with a case back. Nearly all watches work this way. The case back needs to be held in some way, and in this case² it’s threaded.

Specifically an m35x0.5 thread. So 35mm outer diameter, with threads every 0.5mm. That’s a very big, and very fine thread. This is such a weird thread that even the mighty 3000 page Machinery’s Handbook denies all knowledge of it.

The typical way to cut an internal thread is with a tap, a kind a cross between a drill bit and a screw. For this weird thread, tapping isn’t ideal. First, not even the mighty McMaster-Carr stocks an M35x0.5 tap, and even if they did I’d expect it to run over $500. Second, given the geometry, it’d need to be bottoming tap, likely to be even harder to find. Third, aligning a tap like that would be impossible.

So we need another option. One option is to cut it on the lathe. This is a technique I’m familiar with, and the whole point of my silly hybrid electronic lathe conversion was to make it easier. But I wanted to try something new here: thread milling.

Here’s the basic idea: we take a tiny milling tool with a 60° corner, and spiral it down along the thread profile. Thread milling is fast, theoretically very accurate, flexible, and apparently easy.

But I’ve never tried it before.

And the tool for this 0.5mm pitch is, no kidding, very tiny indeed.

So, here’s the plan: we’re going to spin this wee guy around at 10,000 RPM, spiral down at 400mm a minute, and nibble out the thread 0.1mm at a time. Six passes and we’re done. Less than five minutes.

The problem with a tiny tool like this is that there’s no margin for error. Its 1.25mm carbide shaft is strong, but brittle. Even the tiniest mistake and it’ll snap like a twig. No pressure.

To get this right we have to nail two things: dialing in the tool so it has very little runout, and getting the work piece in exactly the right place. This is important because each tooth of this tool is going to be biting out only 13μm of metal, so if the runout or location is wrong by even 10μm it’ll nearly double the chip load. The second was easy: do the bores, then thread without resetting the machine or moving the work. The latter was a little harder.

The first collet I tried I got nearly 30μm of runout at the tip! A brand new collet, very clean collet holder, and a couple goes getting the tightening torque right, and I was within 5μm. Not all bad.

And, in the end, it all worked out fine. First time, which is honestly kind of amazing. Thanks to NYC CNC’s tutorial, even the dimensions came out perfect, and the stock case back screwed right it. When I get around to making the case back, I’ll make the external thread on that. In stainless, which is likely to be even less fun than in bronze.

Footnotes

Or one, if you’re a topologist.
And this case.

Building My Own Watch: First Case Steps

2025-08-15T00:00:00+00:00

Building My Own Watch: First Case Steps

I can't bear to watch.

In my last post, I talked through some of the choices I’d make for the first step of my watch making project. This post is the first step of the first step: an initial version of the watch case. A beta.

First, the unfinished product. The first prototype.

You’ll notice that this has a stock Seiko dial, hands, and crown. But the case is mine! Not the final one, but a first fit of something that works. In this post, I’ll go through the first couple steps in making it.

Step 1: Workholding

The hardest part about machining is figuring out how to hold the things you’re machining. It’s easy when its a block you only need a access from one side (chuck it in the vise), and gets harder from there. This project was particularly challenging. The case is oddly shaped, needs to be accessed from four sides, and is rather delicate once complete. No existing vise, collet, chuck, or fixture would hold it nicely.

I ended up making a total of three fixtures. For the first two operations, cutting the outside, I made a fixture with two screws (the hold the material down) and two dowel pins (to precisely locate the material when I flipped it).

The material blank is drilled and reamed, and then screwed down onto the fixture for cutting. The dowels take the horizontal milling forces, and the upward force is taken on the screws.

This fixture made it super easy to accurately machine external features from both sides, with no need to re-align, probe, or indicate the part after flipping. Repeatability is within 10μm (about 5 tenths, for you inch folks), which is super respectable. Making that fixture also gave me an excuse to use my home made fly cutter.

Fly cutting uses a single point tool, which makes a great flat surface. Doing this fly cutting in the vise on the machine where the actually milling is going to happen really helps repeatability and accuracy, automatically cancelling out any lack of squareness in the work holding below it.

The second fixture is a pair of soft jaws, machined out of aluminium. These replace the vise jaws, essentially making a custom vise for the watch’s exact shape. Repeatability is good here too, within around 25μm (a thou).

CAD makes creating soft jaws like this super easy.

Step 2: My First Screw Up

Machining the external details went great. In retrospect I should have reduced the step overs on the finishing tool paths to reduce work later, but that’s very minor. The next step was cutting out the precision center bores. That’s where things went pear shaped. Or slightly oval.

My initial plan was to do the boring on the lathe. Making perfect round details is what lathes do. Even the crappiest lathe¹ can bore near-perfect bores of of brass all day long. But work holding is a bear, and so I decided to bore on the CNC mill instead. Basically, this works by having the CNC mill move the tool around in a circle. At the best of times, my little Tormach can only get circles round to within about 25μm. I wasn’t too worried, because I could always fix them with a boring head if needed.

So I held the stock in the soft jaws, and did the first set of bores. Nice.

Flipped over clamped down, and did the second set. Disaster! The precision bore for the crystal, that brittle sapphire I need to press into place, was 150μm out of round. Yuck.

That’s less than 0.5% out of round, but plenty to ruin my press fit.

What happened is obvious in retrospect: the soft jaws squeezed this delicate object from the sides, making it a slight oval. In place, clamped into the vise, it’s round to within 20μm. Release the vise, and it’s 150μm out of round.

Step 3: A Second Mistake to Cancel the First

Fixing an out-of-round bore isn’t hard: find a better way to hold the part, and use a lathe or boring head to machine a circle at the oval’s major diameter. Fixing an out-of-round bore where the major diameter is too big is very hard indeed. Running the lathe in reverse sadly doesn’t add material. My basic lathe, unlike This Old Tony’s, doesn’t time travel.

And this is where I got lucky. Weirdly lucky. I pulled the crystal out of it’s packaging, and found I’d ordered a 32mm crystal rather than the 31mm crystal I’d planned.

In the lathe², gingerly held in the four jaw, I was able to open the bore up to 31.975mm for the press fit on the crystal. Sometimes things work out.

Footnotes

My little lathe isn’t exactly crappy, but it’s not exactly high-end either.
Lathe pictured here before it became a hybrid mechanical/cnc mutt. Maybe I’ll write that up sometime too.

Building My Own Watch: The Intro

2025-08-14T00:00:00+00:00

Building My Own Watch: The Intro

Watch This Space!

I like making things. Sometimes furniture, sometimes machines, sometimes RC cars. Whatever takes my fancy at a given moment. What I don’t like is making tiny fiddly things with extremely tight tolerances.

So I decided to make a watch.

More than a decade ago, I inherited a watch that my grandfather was given when he completed 20 years at Unilever. It’s not worth a lot of dollars, but it worth a ton of memories. At some point it’d been dropped, the crystal was cracked, and the mechanism wasn’t working. I considered sending it away to fix it, but instead decided to get good enough at working at watches that I could fix it myself. That started a bit of a journey.

That journey got me here, wanting to make my own watch.

Eventually, that might mean making my own movement (the thing in the watch that actually does watch stuff). But, to start, I decided to make just the outside parts: the case, crown, hands, dial, and strap. The bits the movement goes in.

The first custom watch I built was assembled from eBay parts: case, movement, crystal, crown, stem, strap, dial, and hands.

It came out pretty well! Not perfect, but along the lines of what I wanted.

Choices: The Guts

The first choice is what style of watch to build. There are divers and chronographs, field watches and pilot’s watches, tool watches and dress watches, wrist watches and pocket watches, even the Ankh-Morpork night watch. I tend to like the simplistic end of the scale: the field watch with it’s basic readable face, three hands, and few features.

Second is the type of movement. Broadly speaking, there are two kinds of modern watch movements: mechanical ones with gears and wheels and springs, and quartz ones with a tiny crystal and an even tinier computer³. Quartz movements are better in every quantitative measure: they’re lighter, they’re cheaper, they’re more robust, they keep better time, they don’t need constant winding or wearing. Some can even listen to WWVB. Mechanical watches, on the other hand¹, are more fun. They have cool moving bits you can watch. A little bit of special magic. They have link to the past work of folks like John Harrison (no, not that other John Harrison).

The mechanical movement I chose is Seiko’s NH35. It’s relatively cheap (incredibly cheap for what it is), widely available, robust, and solid quality. The NH35 is a kind of mechanical movement called an automatic, because it winds itself up over time as its worn, by stealing picojoules of the wearer’s precious life force throughout the day.

The Seiko NH series is pretty accurate, only around 200,000,000,000 times worse than an atomic reference.

Choices: Materials

The next question is what to make the watch out of.

If you live in a city, you can go out today and buy a watch with a case made of stainless steel, gold, wood, plastic, brass, silver, ceramic, platinum, titanium, glass, and probably a handful of other things. It’s a small box, without a ton of constraints.

Glass is a pain to work, as are ceramic and titanium, so those are out. Gold, silver, and platinum seemed a little rich for a first attempt. Stainless steel is cool, but boring. As is aluminium. Plastic and wood seemed a bit goofy. So I picked brass.

Or, rather, bronze. 954 Aluminium Bronze to be exact. Aluminium bronzes are cool: they look nice, are easy to work, don’t contain any lead², and resist corrosion by covering themselves in a nano layer of strong aluminium oxide. Brass’s density makes for a nice weight (I find aluminium and titanium cases to be weirdly light). But, mostly, it’s easy to machine. I’ll likely make a stainless version later.

The second choice is the crystal, that little glass window at the front. It may surprise you to learn that a lot of high end watches have a plastic crystal. Acrylic is tough, clear, easy to buff to a high shine, and cheap to make into all kinds of shapes. Objectively, like quartz, it may be the best choice. The most boring choice is plain old glass. Like the stuff you’d pull a pint into. Glass is kind of in the middle. At the other end from plastic is sapphire. Yes, sapphire. Turns out you can grow really big sapphires for cheap and make watch crystals out of them and you have a crystal that’s extremely hard. Only a handful of things like diamonds and SiC are harder than sapphires.

The downside to sapphire is that its brittle. And my plan is to interference fit the crystal. As in make a hole slightly smaller than this thin slice of crystal, and press fit it into place, stretching the metal to keep it in. Fun!

The dial I have in mind is going to be titanium, with some custom CNC ‘guilloche’. More on that in a future post.

Choices: The Looks

I wanted to design something that was my own, but also with a classic look. The closest existing watch to the profile I wanted is probably the Tissot Gentleman. But I wanted chunkier, and less dressy. Something classic.

The dial portion is quite complex: a bore for the crystal, a bore for the dial to sit on, a bore to hole the movement in place, and a bore that the back threads onto. By comparison, the case profile is very simple: 22mm strap size, 2mm wide lugs, and a simple tangent picking up the round from the lug.

I like tangents.

Most of the external parts of this design are entirely aesthetic, but the internals need to be held to a fairly high precision. Mating surfaces for the crystal, the crown gasket, and the back gasket need to be especially tightly toleranced to keep water and dust out.

The story continues in the second part.

Footnotes

The left hand, most likely.
My entire workshop is lead-free, or as close as I can get it. Leaded metals, from brass to free-machining steels, are a joy to work with, but environmental lead is something I just don’t want around my household.
The exception is the weird and wonderful Seiko Spring Drive, which is a mechanical movement governed by a quartz oscillator and tiny microcontroller. It’s not quite a quartz watch, and not quite a mechanical watch.

Prusa XL Extruder: What Were They Thinking?

2024-07-07T00:00:00+00:00

Prusa XL Extruder: What Were They Thinking?

A big flaw in a good product.

A few months ago, I got a Prusa XL 3D printer. In general, it’s a good product. Huge build volume, decent speeds, and good quality. Having multiple print heads is a big improvement over other multi-material setups: more speed, less waste, and more flexibility in mixing materials. The single killer app is printing PLA with PETG supports, and vice versa. Having that capability has change my approach from avoiding supports at all costs, to taking advantage of supports to design better parts.

Despite its strengths, the XL has one huge and annoying flaw, which nearly ruins the experience. Let’s look at a nozzle:

Going left-to-right, we have:

The nozzle itself: the 0.4mm orifice that molten plastic comes out of.
A brass thread that threads into the heater block, where the plastic melts.
A very thin ‘waist’, which stops heat from the heater block creeping back up the filament.
A metal tube, which inserts into a cooling block and guides the filament that’s being pushed down to be melted and extruded.

On the right would be the extruder gears and motors, that push the plastic filament hard down into the nozzle. The pressures and forces here are quite high: all the forces for extruding plastic comes from pushing the filament down into the nozzle from the right hand side (the top, as installed in the printer).

Those beefy brass threads should be plenty strong enough to resist the extrusion pressure right? Well, right. But that’s not how this thing is designed. Those nice big threads only exist to hold the light aluminium heater block in place below the print head. The printing forces are resisted by a grub screw tightened against the thin wall of the heat break section of the tube. Yeah, seriously. A grub screw. Onto a tiny thin-walled metal tube.

Nozzle blocked? Nozzle slips down. Wrong temperature? Nozzle slips down? Too much vigor unloading filament? Nozzle slips up.

Which would probably be fine, in a single-material printer. The XL has a snazzy load sensor which allows it to calibrate the Z height by bonking the nozzle on the bed. It’s a nice feature. But it only works for the first material in the print. The rest of the heads use a fixed relative calibration to find the right place in space to print. Nozzle slippage means printing at the wrong height. At best that means poor quality prints. At worst, it means the printer gouges the build plate hard enough to slip the CoreXY belt. Chaos and damage ensue.

I’ve only had one such disaster, but a couple of ruined prints.

But there’s another more day-to-day issue with this design: the nozzle doesn’t have a rigid, positive, engagement with the tool body. Which means that every time you install a nozzle, it installs at a slightly different Z offset. So every time you swap nozzles, you have to re-run the printer’s calibration cycle to allow the printer to figure out exactly where the nozzle tip is now located. For all five nozzles. And they need to be perfectly clean for it to work. And all filament needs to be unloaded.

That means that a simple nozzle swap takes, at best, about 20 minutes. It’s much more of a pain that the old V6-style hot tightening.

Here’s what I think Prusa should have done:

Make the nozzle a slip fit with a collar into the heater block. Maybe keyed if you want to control the heater block angle. It’d need to be a fairly tight fit to ensure good heat transfer, but the heater cartridge itself is already a slip fit, so that would clearly work.
Add a threaded section on the other end of the nozzle, that screws into the print head.
Add a rigid mating surface at the top of the nozzle that’s rigidly connected to the tool changer arms, ensuring the nozzle is the same place in X, Y, and Z relative to those arms (the reference point for the whole design) every time. This is the way automatic tool changers for CNC mills work: clean metal-on-metal contact between the tool holder and the spindle.
Carefully specify overall nozzle length, or add a per-print touch-off procedure to deal with nozzle length variance (it’s already got all the hardware to do this!)

These changes would have made the XL a much better product, done away with the annoying nozzle change process, made changes easy, and probably even reduced costs slightly.

They would have one downside, in requiring a different approach to V6 nozzle compatibility, which today works with a simple adapter.

My approach would require a different heater block and two threaded sections on the nozzle for V6 nozzles. Not a big deal.

Anyway, I still like the XL, but would like it a lot more if the nozzle wasn’t so poorly designed.

Fixing the KitchenAid's Major Flaw

2024-03-24T00:00:00+00:00

Fixing the KitchenAid’s Major Flaw

Maybe just using a file would have been easier.

It’s somewhat surprising that a product as loved and ubiquitous as the KitchenAid tilt-head ships with at least two major design flaws.

The tilt adjustment screw doesn’t have any locking mechanism, and so the tilt setting drifts down over time. Everdime is a neat little product that fixes this.
The tilt head shaft (the one that the upper casting tilts on) isn’t retained properly, and so works loose over time.

Today, we’re going to fix that problem. Here’s what I mean:

That little chrome shaft is the only thing holding the head and foot together. As the head vibrates (especially when kneading with the dough hook), that shaft moves to one side. Left long enough, the head falls off entirely. The problem isn’t a manufacturing problem, but a rather fundamental design flaw. We can see what’s happening by looking underneath.

Here we can see the shaft is retained with a small screw. That’s not a problem in itself. The problem is that the retaining screw goes right into the casting (which, I think, is aluminium), limiting torque. It also pushes straight into the chromed, hardened, shaft. Without tons of torque on the screw, it can’t get enough bite to keep the shaft in place. This can be solved with a small flat spot on the shaft, or a small hole in the shaft, so the screw doesn’t need to hold just by trying to dent the hard chrome surface.

We’ve got a couple options to fix it: file a small flat, drill a small hole, or mill a small flat with a CNC mill. The first two are sensible. The third absolutely overkill. To the mill!

Side Quest: Making a Jack Screw

Somewhere, some how, my machinists’ jacks have gone missing. First, let’s talk about why that’s a problem:

We’re going to hold the shaft in a collet in a collet block while we mill it. The collet will stop the shaft from turning or slipping. Unfortunately it only holds one end, so as the mill pushes down to cut, the shaft will bend, vibrate, or rattle. Best-case that sounds bad. Worst case it breaks the cutting tool, made from brittle carbide. One solution is to hold up the far side of the piece with a jack.

Easy! I have a bunch of 123 blocks, and I can bolt one down to the machine and use a 3/8-16 screw as a jack. Unfortunately, I’m all out of 3/8-16 screws in my (mostly metric) shop, so we’re stuck making one.

First we turn a piece of brass to size in the lathe, then thread the brass to 16 tpi. As is traditional, I screwed up my first attempt at threading. But the part was mostly recoverable, so I ended up with a decent-enough brass jack screw.

Back to the main project

Which brings us back to the main project: milling that flat. With the jack screw in hand, we can set things up in the mill and get milling. On a CNC machine like this, the milling step is mostly telling the machine what shape we want to mill out, and choosing how fast to cut. Rigidity is the name of the game in machining, and I don’t have a lot here, so we’ll cut slowly in several passes.

Finally, a shaft with a nice flat spot.

After a nice clean with some acetone then some dishwashing liquid to get the machine lubricant off, it can be reinstalled in the KitchenAid. A dab of food-safe grease on the moving surfaces also seemed like a good idea. Then the retaining screw can be tightened down onto the flat, and stuck in place with loctite. Problem solved!

Since doing this, the KitchenAid has been way more stable and reliable. I don’t have any idea why they don’t do something similar at the factory, because it’s a tiny cost that makes the whole machine way more stable.

Other folks online have written about gluing the shaft into one of the castings (e.g. with Loctite 620) which would likely work, but would make the machine much harder to disassemble later.

Review: Tormach PCNC 440

2024-02-24T00:00:00+00:00

Review: Tormach PCNC 440

It's a little CNC mill.

Bottom line: It’s a solid little 200kg-ish CNC mill that works like you’d expect a 200kg mill to work. It doesn’t work like a 2000kg mill. It cuts real parts out of real metal.

Back in 2020, just before the event¹, I bought a Tormach PCNC 440 CNC mill to add to my shop. After buying it, I learned that these things are super controversial. There’s more than one machining forum that’ll outright ban you for mentioning you own one. You’ll find folks saying that they don’t work at all, that they’re only good for plastic, can’t hold tolerances, or won’t cut steel. After having owned one for about four years, I don’t see what the fuss is about. I like it. It’s a little mill that performs like a little mill. If you expect it to do the things a million-dollar four-tonne machine can do, you’ll be disappointed. But if you want to make parts in aluminium, steel, brass, or even stainless that can be made with 1/4” (6mm) or smaller tooling, you’ll probably be happy.

Pros

I’ve cut wood, plastic, brass, bronze, aluminium, mild steel, tool steel, stainless steel, and even nasty air-hardening stainless tool steel² with mine. With some care given to feeds and speeds, I’ve had no problem with any of these materials.
I’ve been able to hold good tolerances (down to 25μm/0.001”) and get good repeatability. Light finish passes are sometimes needed that wouldn’t be needed on a bigger machine.
PathPilot (the UI) is solid. The conversational programming interface is pretty basic, but does the important stuff and is super easy to use. I like the built-in probing, interactive 3D gcode preview, and simple tool library management.
PathPilot runs on a regular PC running Linux, so it’s super easy and cheap to accessorize (I added a touch screen and wired LAN, for example). It also doesn’t come with the memory or storage limits that some ‘classic’ controllers come with.
The Tormach Tooling System (TTS) is adequate for this size machine, and makes it super easy to make custom tools (without having to perfectly match a machine taper).
The pneumatic power drawbar system is super convenient and easy to use.
Decent 1/4” carbide tooling is pretty cheap (e.g. from Lakeshore Carbide or Shars). Other smallish tools like thin slitting saws and fly cutters work great too.
The spindle goes up to 10,000 RPM and down to a few hundred. This range makes it easy to use small tools, while still being able to drill. There are two physical ranges (with a belt), and within those ranges the control is continuous.
The fogbuster coolant system works well, and makes it easy to choose between mist cooling and just an air blast to clear chips.
I got the fourth axis, and it’s a great little thing. Simultaneous fourth axis machining works well. Cheap CAM options are limited so I often end up writing my own code that generates GCode.
It fits in my shop. I have a low ceiling (2.1m), so I couldn’t have a Haas or Brother or something even if I wanted to. I can’t even fit a decent knee mill in, but that’s a different problem.

Cons

It’s a 200kg machine, and about as rigid as one could expect a 200kg machine to be. You’re not going to be hogging out steel with a 1” endmill on this machine.
The limited rigidity means limited tool diameters (mostly 1/4” or up to 3/8”), which also means limited length of cut, making some parts harder to make.
It sucks at drilling with bigger (>10mm) drills. The whole path (spindle motor, belt, TTS) just don’t have enough torque to do drive big drills at their optimal speeds and feeds. I mostly make up for this by interpolating holes with an endmill and then reaming to final size.
The envelope is quite limited in size, especially in Y. There’s plenty of Z space for the stuff I do, but if I did more castings I’d probably want more.
The material removal rate in steel isn’t great. That’s not a problem for me at all, but would mean it’s a poor choice for a production shop.
The touch probe is a great feature, and works really nicely most of the time, but needs frequent recalibration.
I wish PathPilot was a little more extendable/customizable without having to throw it out (in favor of LinuxCNC or Mach or something) entirely. In particular, I’d like some surface mapping routines, SVG engraving, and automatic support for the “interpolate-then-ream” workflow.

Stuff I’ve Made

This is a small sample of the stuff I’ve made on the Tormach. I’ve made a bunch of other things, including machine parts, a few Hemingway Kits, and a bunch of small gifts on the machine too.

A small prize for somebody at work, in aluminium and walnut.

Star Trek: Strange New Worlds badge keyring (in silicon bronze).

Custom Knurls (for the Hemingway Sensitive Knurler, which I also made on the Tormach), in O1 and A2 tool steels. This is a simultaneous fourth axis operation.

Some “carved” details for furniture, in cherry wood.

Pieces for a custom Quarto set for my wife’s birthday, in silicon bronze and 316 stainless (the round ones were turned, but the square ones and all the pocketing are done on the Tormach).

Slotting S90V, a particularly nasty Vanadium-rich stainless.

Footnotes

Remain indoors.
Curse you, S45VN! Great for knives, awful to machine.

Do skis get blunt?

2024-01-23T00:00:00+00:00

Do skis get blunt?

Of course they do.

Last winter, I rode a ski lift alongside a guy who claimed to be a retired Olympic ski tuner. Ski tuning, he said, was all a scam. If you keep your skis dry between days out they’ll stay sharp for their entire lifetime. Don’t get him started on waxing. His specific claim was that skis get blunt from corrosion, and all you have to do is keep them dry.

That didn’t seem right. In fact, it sounded like complete bullshit. Snow and ice are quite abrasive stuff, and even hard steel gets abraded after a while. It’s not unusual for a ski day to include 30km or more of sliding the edges on ice crystals, and so it’s hard to believe that abrasion doesn’t happen.

Turns out it does. Because, luckily, I do keep my skis dry. And I own a very impressive inspection microscope (OK, it’s a $30 one off Amazon, but whatever).

Here’s what one of my ski edges looks like after ~4 days since the last tuning:

If that doesn’t make any sense to you, I don’t blame you. Here’s what you’re looking at:

The ideal ski edge is a metal strip with a ~88° angle between the base and the sidewall of the ski. Here, you can see the base edge, and a “corner” on the edge. I’m not a metallurgist, but this sure looks like abrasion damage to me. Corrosion damage is more pitted, more rounded, and a lot more uneven.

Further down the ski, thing get even worse.

Here, the edge isn’t only rounded over, but small chunks of metal have ripped out to create tiny serrations in the edge. I suspect this effect is why blunt skis can feel “grabby” in some snow conditions.

Luckily, fixing this kind of damage isn’t hard. I use a 3D printed jig with a 500 grit diamond stone. A couple of glides down the edge is enough to completely clear this up. This is what the edge looks like after basic tuning:

Here, the “corner” has completely disappeared, and been replaced with a nice clean sharp angle. The serrations are gone, and the edge is clean. For me, a sharp ski is the difference between fun and terror on an icy day. But maybe I’m in the pocket of Big Ski Tuning.

Knife sharpening nerds will notice that I’ve raised a significant burr here. I believe that race tuners will strop that burr off, but I’ve never noticed a difference (and, let’s be honest, the level of my skiing isn’t exactly “Olympic downhill”).

The Shadows of Trees

2023-09-05T00:00:00+00:00

The Shadows of Trees

Weird optical effects are the best.

I love shadows.

Each shadow tells a rich little optical story about the environment it lives in. A tiny fleeting photograph of the sun, the lights, or the sky, taken through the lens of some complex object. Sometimes, when I’m waiting somewhere I’ll look at shadows on the ground and try work out what the lights above look like. Their color complemented by the shadow’s color. Their shape giving hints of anistropy. Their distance and size controlling the umbra and penumbra. A clear antumbra is a particular treat. It’s especially fun if there’s both sun and lights.

Trees have the best shadows.

The shadows of trees tell the richest stories.

There’s one tiny story I particularly love in the shadow of trees. An opportunistic little pinhole camera where the tree’s own shadow sharpens and bends itself.

Have you ever noticed how the shadow of a tree has some fuzzy shadows and some sharp ones? Normally, the fuzzy ones are from the highest leaves. Their large penumbra measuring the size of our distant sun. The sharp ones are from leaves lower down, their umbra still intact.

But sometimes something magic happens. A tiny gap in the leaves above occludes part of the sun’s disk, forming a type of pinhole. This pinhole shines on the leaves below, creating particularly sharp little shadows even at long distances.

Look at the image above. The center of the image shows a trident of leaf shadows, sharper than the rest. Not from near the ground, but from the second tier. Their sharpness comes from the leaves above forming a pinhole that shrinks the apparent size of the sun’s disk.

Or this one. The same effect, but with significant anisotropy. Or would that be astigmatism?

Have you ever looked at the shadow of a tree during a partial eclipse? You should. It’ll change the way you see the world around you.