Using A Dial Gauge To Measure The Shape Of A Surface

Recently I've been trying to measure the shape of a vintage plastic cupboard handle.  For those that aren't up to speed, check out my previous post on how to use a radius gauge, where I explain the motivation for the project.  I was able to determine the basic shape of the profile on the face of the handle, but wanted to quantify it a little more.  After a lot of thinking I figured out a way to do this using a dial gauge with a magnetic base attached to the carriage of a lathe.  It takes patience but gives great results.

Measurement Setup

The first thing to do is rigidly mount the object you want to measure in the toolpost of a lathe. In this case I've fitted the handle to a thick sturdy bracket.  The screws in the toolpost weren't long enough to reach it so I've used a lathe tool between the bracket and the centre screw to space it out.

Handle Mounting

The next step is to place the magnetic base and dial indicator in a position on the lathe carriage so that they won't get hit if you move the topslide.  Do some tests to make sure that the tip of the dial indicator can reach all parts of the surface you want to measure.

Probing The Handle

Then comes the tedious part.  Position the handle so that the indicator tip is just to the side of the surface you want to measure.  You then raise the indicator tip, move the topslide a set amount by turning the dial, lower the indicator tip, and then take a reading and repeat.

In my case I decided to take a measurement every 0.1 mm.  In the image below, this means that I had to turn the dial 5 divisions between every reading.  Make sure that you always travel in the same direction when doing this, otherwise the backlash in the topslide will make the measurements worthless.  It's also important to lift the indicator tip when moving the topslide.  If you don't, the surface will push the indicator tip to the side, once again making the readings meaningless.

Cross Slide Dial

When taking readings off the dial, estimate the position of the needle if it's between marks.  It may not be exact but it's better than rounding.

Dial Indicator

Depending on what dial indicator you have, you may need a new tip for it.  The one I was able to use had a tip that was very round and wasn't sharp enough to access all the areas of the surface I wanted to measure.

Dial Indicator

You can see the difference between the tip I used in the above process and the one that comes with the indicator.

Contact Points

I was able to get a set of dial indicator contact points on eBay for 20 dollars Australian.  They aren't the highest quality things I've ever seen, but they do exactly what I want and that's all that matters.  If you're looking for a set yourself, search for "indicator contact tips" or "indicator contact points".  Searching for 4-48 UNF threads also leads you to these as it seems to be pretty much the only thing that thread is used for.  Some dial indicators use an M2.5 thread, so you may need an adapter.

Contact Point Set

The point of the tip I used looks to be round, but using radius gauges I was able to tell that it had a radius of less than 0.3mm.  I'm fine with that.

Sharp Contact Point

When performing measurements I wrote down my results as I went.  I then went way back to the beginning started again and took a measurement every 2.5 mm to make sure that the dial indicator didn't move during the process.  This was just to verify the setup and are the circled data points.

Data

The data was then plotted to reveal the shape of the handle.  When looking at the shape it's important to remember the limitations of the tip geometry.  The three sharp crevasses will appear round because the tip is slightly round and can't get all the way to the bottom of them.  As the tip is conical it will also have trouble measuring the vertical part to the left and right of the central bumps, so these sections will also be slightly distorted.

The red dots in the image below are the verification data points.  This makes me confident that the setup didn't move during the measurement process.

Profile with verification points

Another thing to look out for is that you might not have mounted the object perfectly flat.  Because my object is symmetrical. I can mirror the data, overlay it and see that I do indeed have a small misalignment.

Profile with its mirror image

By lining the two up, I was able to determine that the part was off flat by about 0.5 degrees.  I can now compensate for that later.

Profile with its mirror image rotated 1 degree

Another thing that may happen is that the handle is not mounted perpendicular to the motion of the top-slide.  This would have the effect of stretching the shape side to side.

To be honest this turned out a lot better than I expected.  I need to do more work to clean up the data, but I now have a good reference to work from.

Using A Radius Gauge

I recently managed to buy unused original 1950's stock of a handle that was used on my grandparents kitchen cupboard and I want to be able to reproduce it in the future.  The first step in this process is measure measure measure!

Replicating the exact mechanical operation of the handle isn't that important, what I'm mainly interested in is being able to reproduce it's aesthetics.  The first part I'm measuring is the radius of the beading on the outside of the handle.  One problem, I had no idea how I was going to do it.

Acetex 44L Handle

My go to tool for something like this would normally be a caliper.  Unfortunately in this case there isn't really anywhere to get the jaws of the calipers around the curve of the beading.  Not even 180 degrees of the beading is available to measure.  So to measure the curve you need to compare it to a set of reference curves.  It turn's out that the tool to use is a radius gauge.  After a longer than expected wait for $10 gauge from eBay, I was able to get started.

Radius Gauge

The gauge comes with external and internal feelers with radii ranging from 0.3mm to 1.5mm in 0.1mm increments.  The concave gauges can measure radii where around 70 degrees or more of the total circle is available.

Concave Gauges

The convex gauges can measure radii where 180 or less of the circle is available.

Convex Gauges

Ideally when using these gauges you hold them against the surface you want to measure, hold everything up to the light and look for light leakage around the edges.  To take photos I've just put them on the table and placed the gauge against the handle.  You can see that the 1.2mm gauge is slightly too small.

1.2mm Gauge

It's hard to tell from the below photo, but the 1.3mm gauge is almost perfect.  A little to big. but close.  This now gives a range for the radius of 1.2-1.3 mm.

1.3mm Gauge

It should be said that I haven't verified these gauges against a standard, but I plan to take a few measurements in different ways and use them to build up confidence in the profile I measure.  From what I've been able to tell there are 4 beads that make up the central reeding.  They are evenly spaced and there is 7mm between the centres of the outer ones.  The radii of the beads is as seen before about 1.25mm.  The shoulders of the profile are made of two sections.  Coming from the beading is a flat section that I've eyeballed to be about 5 degrees less than horizontal.  I've tested the flatness of this section by placing a razor blade against it and checking for light leakage.  After this section is an unknown curve that is tangent to the other section but stops abruptly at the edge.

Handle Profile

Still a lot of measuring to do. No idea how to do, but I'll figure it out.

Restrictive Capacitive Touch Screen Guide

Recently at work we received MC40 mobile computers to perform inventory tasks.  Just one problem.  The interface was designed by Lucifer himself.  The image below shows the problem.  The on screen keyboard is located in the red area at the bottom of the screen and the scan button to read bar codes is shown as well.  Repeatedly moving your thumb back and forth between the two areas causes pain in the first joint of the thumb.  Ideally the keyboard should be at the top of the screen in the green area.  That way the thumb can move across without having to bend.

All of that wouldn't be too bad if it wasn't for the other issue, the custom on screen keyboard can be swiped to the left or right to get a full keyboard.  So when typing with your thumb bent back in an awkward position, if you move even slightly to the left or right while contacting the screen the keyboard starts to move left or right and the key press isn't registered.  The units the MC40 replaced were old, but they had physical buttons with audio feedback.  This made it easy type without looking at the keyboard.  The MC40 forces you to touch type like someone who's never seen a keyboard before.  If there was only one lesson to learn from technology over the last 10 years it's what Apple taught us, study how users interact with the product.  Put more than 10 seconds of thought into things.  I've partly already solved the problem by creating bar codes that automate common function that I used to perform on the keyboard.  At an estimate I've cut my key presses by about 95%.  Can we do more though?

MC40 layout

What's needed is a physical object to give tactile feedback of the button locations and limit the movement of the thumb when it touches the screen.  It's a capacitive touch screen so as long as I use a non conductive material I should be fine.  So what I plan to do is 3D print a guide that goes over the screen and is held in place by rubber bands.  It's a prototype just to prove a concept,  I could have had a PCB with holes made to do the same thing, but I want to try 3D printing.  It also allows me to add guides to hold the rubber bands. and round some of the edges.  As this part is designed to be touched, it can't have edges that will wear away at the skin.

On screen short cut keyboard

I quickly realised that Sketchup wasn't up to the job.  I settled on FreeCad and I'm happy with it.  There was a week of swearing and watching video tutorials but I got up to speed after I figured out how parametric modelling works.  (Another thing to add to resume)  The result is below.

Touch-screen guide

When placed over the screen you can see how it's meant to work.  Two rubber bands that run between the guides at the top and bottom of the part hold it in place.  It allows you to feel where the buttons are without looking and it restricts the movement of the thumb so the keyboard cant start swiping to the left or right.  Imagine the hole in the end of a ruler, when pressing it with a thumb the skin protrudes the most in the middle.  This will be the contact point.  I'm not sure that I have the geometry right though.  If the holes are too big the structure will become weak, if they're too small the thumb won't fit through.  If the part is too thin it will break, if it's too thick the thumb won't touch the screen.  You can see that it will be sensitive to geometry.  Consider this a minimum viable product.  Once I have it I'll be able to make some informed iterations.

Guide in place over screen

To see how the 3D printing would be completed I installed Repetier Host and sliced the model as a test.  I don't have a printer, a friend does, so I want to make sure I sort out any problems to make things run smoothly.

Sliced model

I then came up with another crazy idea.  What if I were to make a negative of the design and use it as a mould?  I could make rubber version of the design to prevent it slipping around the screen.  I'm no polymer expert, but some silicone RTV for gaskets from a local automotive store should do the trick for a prototype.  Just a thought.

Mould negative

I may appear to be down on the MC40, but the scan engine in that thing is a beast.  It's a newer (2012, that's new compared to most of our equipment) camera based design and it can read bar codes so fast you're not even sure you fully pressed the button.  The images are processed so quickly I'm 99% sure the algorithms it uses are in custom silicon.  I'd love to see how it works.

This isn't a project that needs to be done and 3D printing may not be the way to go, but at least I'll have some experience with 3D modelling and printing.  Excited.

In Plane Magnetic Field of a Current Loop

Well, this one is a failure (for now).  In the comments on my post about the "Off Axis Magnetic Field of a Circular Current Loop" I was asked why I'd chosen to solve the problem the way I did and why I didn't calculate the Magnetic field using the more familiar version of the Biot Savart law.  I don't know.  I just happened to pick this method and it worked.  I was also asked about calculating the in plane magnetic field of a circular current loop and my response was use the 3 dimensional solution and set z to 0.

I did want to try and calculate it just using the Biot Savart law.  After all, the field shouldn't be that hard to work out.  Due to the geometry of the problem there can only be a magnetic field perpendicular to the plane of the current loop when z=0, and due to rotational symmetry you only need to calculate the magnetic filed along the x axis for example and then revolve it to generate the entire field.  Easy, except I just can't get it.  I'll show my attempt below.

Magnetic field of a current loop derived from Biot Savart

That's where I hit a wall.  I've made some progress, but I just can't get the equation into a form that makes the use of elliptic integrals easy.  I took the vector cross product of the infinitesimal current element and the vector pointing from the current element to the point that the field is to be calculated at and divided that by the distance between these points squared.  This is then integrated around the current loop.

Geometry of the situation

I've performed numerical integration using maxima at various steps of my calculation to make sure I haven't made any mistakes.  Although it's no guarantee I'm right, all the answers match.

Current Loop In Plane Field Working

Looking at the equation below from my original blog post you can see the equation I'm aiming for if you set z to equal 0.  Plugging numbers into this also yields an answer equal to what my equation above gave.

Desired Equation

I think this is one of those cases where it's all a mater of finding the right identity or change of variable.  I might let my subconscious work on it for a while and come back to it refreshed.


Update - 4 December 2015

I solved it.  It was relatively easy in the end. I just needed to take a break and look at it differently.

In plane magnetic field of a circular current loop

Flaring the End of A PVC Pipe

Recently I've been building a chicken coop for my sister and I've finally come to the end of the project.  One finishing touch that still needs to be done is for a downpipe to be fitted to the guttering.  To keep costs down I'm trying to use materials that we already had, and it turns out my sister had a piece of 40mm PVC pressure pipe from another project.  As it turns out it wasn't quite the right size, so I thought I'd document how to make the pipe bigger.

Original PVC Pipe

Finding a fitting to connect the pipe to the guttering isn't easy as it's 40 mm pressure pipe.  40 mm isn't a standard size for guttering downpipe, so I found a fitting that was almost the right size and decided to flare the end of the pipe to fit it.  I think the fitting was for a floor drain, but it'll do what I want.

Flaring the PVC Pipe

Flaring the end of the pipe is easy.  You need something round and tapered, and a heat gun.  The pipe is evenly heated at the end by rolling it backward and forward under the opening of the heat gun until it becomes pliable.  It doesn't need to be too soft, just soft enough to insert the fitting, but before that can be done you need to flare the pipe as the fitting initially won't fit into it.

The only tapered thing I had that was the right size was a coke bottle.  After filling the bottle with water so it wouldn't compress, it was pushed against the pipe to expand the soft end.

PVC Pipe and Fitting

Once the pipe is slightly expanded, the fitting can be inserted.  I left the fitting in the pipe until it cooled naturally.  I figured cooling the pipe too fast might not be a good idea.  Better to be safe than sorry.

PVC Pipe with Fitting in place

PVC Pipe with Fitting in place

It's a really tight fit but it still needs a fastener to hold it in place.  A pop rivet is ideal but a screw will also do the job.  After putting a couple of screws into the sides it was then attached to the gutter with 2 more screws.  Pop rivets would have been better but you do the best with what you've got.

To save an extra couple of bucks I put a bend in the bottom end of the pipe with the heat gun instead of buying elbows. Below is a shot of the finished chicken coop with the downpipe fitted.  There are a few things I wish I'd done differently, but as it is I'm pretty damned happy with it.

Down Pipe Fitted, Chicken Coop Finished