Vintage typewriter

I’ve just completed the first “phase” in a, for me, ambitious reconstruction of a 1915 Underwood 5.
Although I’m pleased with the outcome, I am once again faced with this conundrum: The model looks a bit more massive than the original. I work from pictures and direct measurements from callipers.
Since this mechanism was designed and built in the US, I went with the “woodworking, inches” template.
My calliper’s precision being limited to 1/64", I used a 64:1 scale, saving me the constant typing in of fractions.
Could this be the origin of the loss of slenderness that happened during the build, as the cumulative effect of the rounding out that occurs?
I’ve had this slight problem arise before, and I’d like your collective input.
Although I’ve used as few segments as possible when creating curves, arcs and circles, I wound up with a rather hefty file, so that I can only post a picture of the rendition here. Kindly go to 3DWarehouse under “glistenUp” for the .skp.

Thanks for your eventual contributions.


Nice! I really enjoy seeing complex mechanical models of actual objects.

Regarding the dimensional accuracy, for the future I suggest you invest in a caliper that reads in decimals (either inches or mm) rather than fractions, as a way to improve your sanity. :slight_smile: Electronic calipers will have various display unit options; dial calipers will read in a fixed unit. I have a nice Starrett 6" dial caliper and a Folwer 12" electronic caliper (plus various other fun measuring tools) for my modeling re-creations.

Errors can accumulate depending on how you model progressive pieces of geometry. It’s often wise to measure related points from a common origin and create the geometry from that same origin. For example, consider a stack of sheet metal laminations, where each sheet is “about” 3/64 inch thick. If you settle on exactly 3/64 for the model and stack up ten of them (to yield 30/64 inch overall), you may have a moderate error if the actual stack is, say 28/64 inch thick. In this case, you might decide to make each sheet be one tenth of 28/64 thick i.e., 2.8/64 inch.

More generally, measure the progressive dimensions from a common origin rather than approximating each object individually.


My callipers offer displays of inches in fractions (to 1/64th) and decimals (to 1/100 of a mm or 1/10,000 of an inch(!)). But I dare not switch from one to another mid-build so as not to create misalignments and mismatches as the project evolves.

The accuracy principle you bring up actually helped me solve the problem I had with the fact that my callipers only open to 6", when the cradle is over a foot long.
In order to measure the rack’s (as in rack-and-pinion) teeth, I counted out 50 of them, measured that span and divided it to get each cog’s dimensions. By counting ALL the cogs(!), I got the accurate overall measurement that had eluded me till then.


I was under the impression that components were useful mainly with groups that had many iterations.
Nearly all of the objects assembled here are one-of-a-kind. How should I use the component’s functionalities in this instance?

Granted about the screws, although there are a lot of different faces and proportions of these, but the only mirrored parts in this build are these two support arms.

Most of the components I did use during the build were exploded when making the final objects into groups.

That’s why I was inquiring about any OTHER useful properties of components.

Why? Why do you feel they need to be groups? and not left as components?

The final assembly of any particular part can contain geometry from a component and some other data, grouped and/or not. I’ll explode all of the desired geometry, group it and get rid of any superfluous edges or faces that I can spot.

One example would be that your Rack could be made up of 50 tooth components all in a group/component. This would allow you to adjust one tooth to adjust them all. There are many opportunities in modelling to use components to reduce the amount of work you need to do.
Making one half or a quarter or even a twenty-fourth when it could be part of a default circle and arraying, mirroring or flipping cuts down on the work needed and reduces the file size. If done well you can have an easily editable model with the option to convert multiple components into single solids if and when the time comes.

1 Like

They (the teeth) were exactly that while I was building the rack. When I was done, I needed to get rid of the abutting component’s edges to obtain smooth surfaces. You’re saying this can be done while each tooth remains a component?

of what?

You’ve lost me here too…

Remove or hide the face and hide the edges.

That depends on the model, I haven’t look at yours to see if there is anything specific, but something like a detailed chair leg, you may only need to draw half or a quarter of it.

You can make highly complex shapes using a radial or linear array of a single component.


Here’s a quick sample gif, you can see that I have hidden the connecting faces and edge, but the component is still a solid.
The radial array can then all be selected and you have 24 solid components that if you ever actually need it can be made into one single solid with either Union or Outershell.
Obviously this shape isn’t directly relevant to your typewriter, it’s only meant as a general modelling tip.


Wow! What an eye-opener!.. I’d never thought of trying that or heard it mentioned before.
This changes everything, thanks…
I’d click “solved”, except that it has nothing to do with my OP.


@eneroth3 even has a commercial plugin to automate the process.

1 Like

Sorry, but exactly what process are we talking about, and what’s that plug-in’s name?

Well, you quoted the part about hiding edges and faces so I assumed that was what you got excited about.

1 Like

I was, and am, but I’d’ve thought of hiding edges as a simple operation, not a process justifying the writing of a plug-in…

It allows you to select multiple groups and components and hide all the connecting edges in one go. For many people that would be a worthwhile time saver.


Got it, thanks again…

Pretty much everything in SketchUp can be done with the line tool alone, and pretty much anything can be automated. It all depends on how much time you would save, and what that time cost you. One user can save hours on automatic something while it would only be a waste of time to even install the plugin for someone else. It all depends on your workflow.


Another major reason to use components instead of groups is to be able to easily employ the “Dave Method” for working on short edges (to avoid SketchUp’s tendency to delete and combine them). Make a temporary copy of the component, scale it up by 100x or so, and edit that copy. When done, delete the copy. An normal-sized instances of the component will inherit the edits done to the large instance, and SketchUp will not clobber the small geometry (because it wasn’t so small when created in the context of the large instance).