DIY machinist level

Failed electronic sensor

This story started by making an experiment with a capacitive sensor. Using a straightened sheet of metal from a tin can to make the plates of a liquid capacitor, arranged as two face-to-face combs (see drawing). However the capillary effect caused problems, and making the plates more spaced to reduce the capillary effect resulted in a smaller capacitance hampering the sensitivity.

So after having gave up the electronic way, I endeavored into the classic bubble, testing various options, including an acrylic tube cut to measure, and small plastic tubes (likely ABS plastic) reclaimed from pens.

Making a "spirit" machinist level

Spirit levels are made using curved vials, so that the bubble inside the vial will stay at the highest point of a semi-circle, of which the tangent represent the line that is perpendicular to the direction of the Earth's gravitational force. The slope is given by the ratio between such line and the actual plane under test, it is in other words the raise over run ratio.

So to make a level, one should make a bended vial in the first place, not an easy thing to do. Although vials can be purchased online, still I was curious to see if it was possible to make one at home.

Plastic vial

Using a plastic tube has the obvious advantage that can be easily bended with precision by placing a shim below the center of the vial, and two shims on the top extreme sides, compressing them up to reach the desired curvature.

This however came trickier than expected. The main problem that I experienced lies in the attraction the plastic exerts to the fluid. Despite using different liquids, including various solutions of water and alcohol, and dishwashing soap as surfactant, I experimented troubles in making the bubble to move along the vial with high enough sensitivity.
The bubble always sported the tendency to stick to the wall of the vial, likely because some kind of electrostatic attraction.

For a sensitivity of something around to tenth over two hundred millimeters (4 thou over 8 inches) per division, these vials are fine, but for greater sensitivities they display some kind of hysteresis that makes accurate reading hard to be achieved.

Glass vials that smell good

So the best material seems to be glass. To test this I though to try to use sample perfume vials.
If I didn't messed up the math, the required curvature for the intended sensitivity, which is something in the hundredth millimeter ballpark (.4 thou), should be just a few microns (.2 thou).
So my idea was to either check the natural irregularity of the vial to exploit it, or to force and bend the vial.

It turned out that these vials are pretty well made, with very little irregularities. This is an advantage as the vial will be fairly accurate, but requires to force the vial.
Well, you'll think glass do not like to be bended, nevertheless glass is somewhat flexible, and should be possible to force a slight curvature.

Making the level

The first operation was to remove the silk print from the vial. This can be easy or hard depending on the type of print made. I used a polishing wheel with a polish cream that worked pretty well ...but some prints are really hard to get rid of.
Then I've built the case that will contain the vial out of one 15x3 and one 15x5 brass flat bars.
The first one becomes the holder of the vial, creating a squared slot large 9 mm, and a 15x1 flat stainless steel, that I glued below the slot with epoxy.
The second one becomes the lid of the case. It sports a slot through which the vial will be visible. Below a round pocket is engraved to make room for the top side of the vial.
Then I made a trunnion with a threaded stud. Through a 2 millimeter hole a steel pin claimed from polished nail is inserted, and this pin allows the trunnion to pivot on one end of the case.
The trunnion will then be attached to the base of the level, and that will allow the case to be adjusted by tilting it up and down.
With a chunk from a wooden dowel beam and a 4 MA screw, the knob is made. This will allow to adjust the tilt.
The base of the level is made out of a 20x20 cold steel bar, long 200 millimeter.

This steel part needs to be made flat, so after having roughly flattened down the bar on a flat stone covered with sandpaper, it then required a fine work by scarping the bar to make it really flat against a surface plate.

The case is finally assembled onto the base, and the vial enclosed inside the case.
A spring assures that the case stays pushed against the knob.
And finally a series of marks spaced 4 millimeter each are engraved on the top of the lid to have some references to read from.

The final touch: I added two tapped holes at the sides of the base. And out of a 18x7 millimeter piece of wood I engraved the middle with a piece of sandpaper wrapped around a tube, and then I used a shellac dye to color it in red and further transparent shellac to protect the wood. This created two nice handles that are screwed onto the sides of the base of the level. Not only these handles offer a better grip but also help to thermally isolate the level from the heat coming from the hand.

Something went wrong

As I said, I ended up bending the vial.
My first attempt, however, didn't went good. Because I needed to take practice, at first I tried to get a moderate sensitivity, in a way that to move the bubble by one division a shim of five hundredth millimeter (1.9 thou) was required. To do this I forced the vial to be bended too much and during the night, you know, the temperature lowered causing the spirit inside the vial to shrink, which in turn increased the tension to the point that the vial cracked.
So I started over again with a new, shorter but larger vial. I adjusted the height of the case gluing two brass wires, and put a shim below to get just the few microns (in the order of .2 thou, in inches), and giving a moderate pressure inside the case by carefully shimming the pocket that holds the vial, to get a slight curvature.

Eventually...

The result was enough. The level is fairly accurate, as it shows symmetry swapping the level by 180 degrees, and it shows a sensitivity of one hundredth of millimeter over 200 millimeter or .4 thou over 7.9 inches, with a decent repeatability.
So I'm pretty pleased with the final results achieved.

And I hope you had some fun reading this article, but for more fun don't miss to watch the video:

RELATED ARTICLES

RECENT ARTICLES