Friday 4 April 2014

Thoughts on 3D printing a typewriter.

It has been a while since I've addressed the question of 3D printing typewriter components. As I've so far been the only person to really attempt to tackle it, whenever the topic seems to arise in forums naturally I'll end up getting an email or a message on the subject. 

So I think it is time to discuss the current limitations, as well as what is actually possible. But to do so I'm going to have to take you through some basic engineering and design principals to explain these limitations. 

For those who don't know, during one of my many lives I did a course that gave me a certification that was known as a 'COS in Engineering'. It was an intense 6 month course that effectively crammed 2 years worth of technical training that you would get progressively while doing an apprenticeship. Most of my contemporaries went on to become toolmakers and engineers. I however went on to become a theatrical stage manager after studying Theatre Technology. 

I guess you do what you are passionate about. Anyway, that was another life really. 

So I popped out of the other end of this course having done training in machining, technical measuring and quality control, drafting and planning and CNC programming amongst other things. I was always excellent at the theory and the intellectually complicated work. But was crap when it came to the manual machining stuff. So they pushed my study in the direction of looking at a new field in engineering:- Computer Aided Design and Drafting. 

I was fortunate. I got advanced out of some of my workshop modules and instead ended up being placed with several senior engineers with Toyota who were receiving specialised training on CADD. Back then the workshop engineers didn't think much of this CADD stuff, but it was clearly the direction of the future of manufacturing at the time. And getting a grounding in it is important for those who want to use this kind of 3D technology to produce refined and quality parts. 

Problem 1: Tolerance. 
No, I'm not about to lecture you on peace and good will to all man. But rather about a problem that has faced engineers for hundreds of years. Our current work tools are incredibly accurate. But are only accurate to a point. So when parts are made by engineers, they need to specify how accurately the part needs to be made. 

Basically, it is a way of saying that you want the part to be a specific size, but you can allow it to be made a certain size bigger or smaller for it to still be able to work a specified. 

If a tolerance is 'tight' it means you have very little room to move on the specification. Loose tolerances tend to be wider and less accurate. 

For example: The ball bearings on the Royal 10's carriage rail is exactly 5 millimetres wide. This would be specified as 5.00mm on the plan. However, the tolerance would be about 0.1 of a millimetre, which means that the ball bearings could be made between 4.9mm and 5.1mm in size and still be considered acceptable. 

Most of the components made for typewriters after about 1930 however were to much more exacting tolerances. So you would often have a play of much less in size - say, 0.05 of a millimetre. 

Think about it. That's thinner than most people's hair. We're talking incredibly accurate sizing that was made by guys using non computer guided tools. 

The problem with 3D printing however, is that the technology is currently limited to an accuracy of over 0.2mm. So while this is incredibly small, it isn't small enough to get the results needed for the really refined parts of the typewriter. Specifically the moving parts that rub, shift and slide against each other. 

However, this can be overcome, but requires handing of the parts with some sophisticated hardware afterwards. When I made the handle (displayed above) for the Remington Noiseless, it had to be drilled out post print. This was the only way possible of getting the accurate finish required for the part to work. Other 3D printed parts - if done carefully, could potentially be further machined to produce the accuracy needed. 

Problem 2: Design. 
We are unfortunately in a position where every part that we wish to manufacture requires a considerable amount of reverse engineering. The part may look simple when you look at it, but often there's grooves and intricacies that were machined into it that we often mistake as being ornate. Those grooves are there for a reason that often isn't obvious.

The big problem is getting accurate measurements. To do this there's a plethora of tools that are used. These tools need to be able to measure up to a 100th of a millimetre (some of them a thousandth) and as such they are expensive, and require training on how to use them to get a good result, without damaging them. 

The tool displayed in the middle there is a Micrometer, and it is the most accurate of the three displayed here. I used such a tool to get the measurements for the Remington part, and that took me about an hour to get about 12 sizes. And that was off a part that was incomplete, and needed some guessing. 

Curved surfaces are difficult. Threads are relatively easy (you use a thread gauge or tap set to check) complex shapes are..... f***ing hard. Wonder why I haven't popped out a replacement escapement block for the Royal 1 yet? It is a complex shape that would have been easier to design onto the machine already knowing all the angles and measurements required, than it is to measure out afterwards. It is really hard work to get the right measurements that are within tolerance. 

How about measuring up this guy.

And then there's the gears. 

A gear might just look like a disk with teeth, but oh my... there's a hell of a lot of measurements involved. 

There are engineers that specialise in designing and manufacturing gears. I don't know if I can emphasise enough how hard gears are. 

At the moment 3D scanning technology is in its infancy. The tolerances on it's scanned information are about the same as what the 3D printer outputs. So as you could imagine, a combination of the two could blow your tolerances out by a huge amount - rendering the effort pointless. 

There however is a very simple way to bypass this whole process:- work off the original drafts. 

The patent drafts don't usually contain enough information, but the original drafts for each and every part must exist somewhere. Somewhere..... Anyone got a basement full of designs? 

Problem 3: Materials.
The 'stainless steel' that I constructed that Remington part out of is strong. But it is what we would call a 'hard' material. What this means is that the material - while strong, is likely to be brittle when stressed and simply crack and crumble when put to its limits. This is a different definition to hardness than what we are talking about when we are looking at platens. There, we are actually talking 'more or less tough'. 

A tough material isn't brittle and has give, and flexes instead of just simply breaking. Plastics are tough. Even metals are more tough than hard - depending on how they have been treated. 

A typewriter needs both kinds of materials. However, at this stage 3D printing produces only 1 when it comes to metals - hard materials. 

But 3D printing isn't the only weapon we have at out disposal. CNC milling, lathing, shaping and cutting tools are also available to produce objects out of other materials. These methods are also far more accurate and able to cater for our more refined manufacturing needs. But they have a disadvantage - they aren't actually all that flexible. 3D printers are able to produce complex shapes easily. But CNC gear requires a cutting tool's route (the part that shapes the metal) to be planned carefully. This makes machining these parts a lot more time consuming and expensive. But once done, you can produce hundreds of thousands of identical parts. Which is exactly what this kind of technology is typically used for - mass production of highly accurate components. 

A 3D printer however does some pretty good work with making tough plastics. Even your home based 3D printer can make some great plastics. But because of the layering technique used to print a part, it tends to produce an interesting texture to the item's surface. As such this isn't exactly ideal - especially when you need the surface to be smooth for smooth operation. 

But then we have the problem of type-bars. 

Image from

Type-bars are unique in engineering. They have complex shapes (the gears and the slightly turned type-head) that require a lot of accuracy. But they also have an unusual metal structure. At different points on the bar the metal goes from tough to hard. For example: The head that the type stub is soldered onto is hard. This is done by heating that section till it is glowing hot, then cooling it very rapidly. 

The middle of the type-bar is typically forged. This is effectively a tough section of metal that is pressed under sudden and immense pressure - this compresses the metal's atomic structure closer together so that it is even more rigid. This makes extremely tough, while being very hard as well. 

Around the gear at the other end of the type-bar needs to be tough. It has to be flexible enough to handle the bounce and the whip as the head hits the platen, but it also needs to be machined very smoothly so that it doesn't break down over time with friction with enough hardness to resist wear. 

In short: There isn't a machine in existence that will produce a product quickly and easily to replace a type-bar. But with a fair bit of post machining and handling you can get close. Sadly, you won't be able to forge anything easily. 

So is it possible to make a typewriter? 
Not including the potential hundreds of hours of work required to measure up and design the parts in CADD software which could be useful in the future - the answer is 'No'.

Well... sort of. Actually, there is a typewriter design that could potentially be a great candidate for 3D printing with a lot of work. But it will require assembling the parts afterwards and a lot of adjustment.

What is this typewriter you may ask? 

Photo - Oztypewriter

The (not so) humble Blickensderfer. Anyone up for taking on a 'Project Blickensderfer' to see how far we can get? 

There's other issues involved, and really I have only glossed over the issues that I have discussed. There is a whole lot more involved that may need a fair bit of discussion. But all in time. It is about to hit midnight here, and my eyes are drooping. 



  1. Scott -- awesome write up! I might also throw in that you can overcome tolerance issues with things such as precision bushings pressed into less precision parts, but again, to your point, it is labor intensive and expensive.
    "Project Blickensderfer" - man, kinda makes your skin tingle thinking about it! :-)
    I wonder how much of a blending of metal and plastic 3D printing could be done here - obviously components like the keys and body shell could be printed in an ABS material. I really hope this gets some teeth and you can make this dream a reality, it would be awesome.

    1. Well, it is one of those things that will take time. I think the idea of re-creating a Blick is possible because it can be done collaboratively. People can design parts separate to each other, and the data then can be used to build more units.

      One thing that is immediately possible is that type-wheels (balls?) can be created on a 3D printer - so it is possible to add new typefaces to what is essentially a very old machine. In fact, this is something I would love to pursue.

  2. It's another tool in the restoration toolkit, but indeed not as easy as it may sound at first glance.
    Myself tried 3D printing about a year ago to get proper fitting new feet for my Remington portable. Made a couple of test prints to get the size right, but the shape takes quite a lot of material so a bit costly. The basic materials from the print-shops are rigid, a mate with a few printers and more experience tried printing in a flexible plastic. These feet in Arnitel Eco actually have a proper rubbery Shore value, but are very slick. With extra finishing with a couple of layers of plasti-dip sprayed on they were good 'rubber'-feet. In the end bought a set of new real rubber feet from TTS as just cheaper and easier.

    Parts that were originally cast or machined are probably best candidates for 3D printing. Stamped parts less so. Also the printed part is more of a start than the finished product; you'll want to do finishing on the part you get back in the mail for fit and accuracy or even just a proper appearance.

    Am (coincidence!) right now this weekend in the process of re-creating a winding-hole escutcheon (?) for the gramophone. Will post about how that comes out...

    1. AH! You know, I have thought a fair bit about the feet and at an early stage tried to design a similar thing myself back when Shapeways produced 3D printed rubber. But it got me thinking though... and I recently had a brilliant idea about the rubber feet which I'll talk about soon.

    2. Intriguing! Very curious what you are thinking of.
      (Shapeways now have a very squishy translucent material, by the way. More suitable for squeaky toys than feet probably though...)

  3. An even easier candidate for re-creation might be the early 1900's Bennett Pocket model:

    Very few parts, and the tolerances aren't all that tight. (:

    1. That's a good point. I never think about those little machines! And there is actually an awful lot of data out there on the measurement and sizes.

  4. 3-d printed daisywheels are the first thing that come to mind when I think of impact damage on the typing parts. Make a plastic daisy wheel, and when it cracks or loses a petal, grind it up and feed it back into the printer for its replacement.

    Also: now I want one of Ted's Bennets.

    1. That's entirely feasible. Actually, the plastics used in 3D printers today are often better than the ones used in those 80's machines. It is entirely do-able.

  5. I'm impressed by the "theatrical stage manager" part! :D

    1. Ha ha ha. Should I mention I ran my own theatre company (albeit briefly).

  6. Daunting ... thanks for giving us some good technical information and helping to create a realistic idea of what is currently feasible.

    The idea of new Blick typewheels is exciting. Techno Blick, anyone? Emojis?

  7. This comment has been removed by the author.

    1. Hey! Where did this reply go. It was really good!

    2. Argh! Yeah, I was feeling so inspired, too...

      I think I said something about how recreating typewheels for original Blicks might be a good idea. But, if you were going to create a whole, new typewriter via stereo lithography, it would seem more practical to design one that utilizes IBM Selectric golf balls. Those things are are relatively easy to find and they come in a dizzying variety of fonts.

      The thing to remember about rapid-prototyped castings is that they may need some finish machine work. If you go in expecting precision tolerances, any casting process can be disappointing. Stereo lithography just speeds up the process from design to casting.

      Something else I can mention is the ultimate in sintered metalurgy: Way back in the dark ages, when I was getting my degree in manufacturing, it was possible to rapid-prototype with powdered titanium. It can't be investment cast like bronze or aluminum, but it can be initially formed by a similar process. Then, with heat and pressure, the particles are fused into a semi-porous solid.

      How about a 3-pound, titanium neo-Blickensderfer that uses IBM golf balls? Anybody with unlimited funds want to take on this fantasy project? :)

  8. I really appreciate you posting this, I found it in my searches for information on whether or not anyone had ever 3D printed a typewriter before. I recently started collecting and doing basic repairs on typewriters, and I am a college student who has to do a capstone later this year/into next year. My roommate suggested 3D printing a typewriter because we have ample access to 3D printers and unlimited filament. There are also quite a few people in my group that have experience with CADD, so it seems like it might ever so slightly be possible to attempt. I don't have much information on it as of present, but I just wanted to say that I appreciated the resource, and if you are interested I can keep you posted and potentially field questions to you. Hope to hear back soon, and thanks in advance.
    -Erica Ryan