A terribly framed picture of the final hardware. You pull on the two eyebolts at the top of the photo, and the inside elbow of the U stretches. The two little brown tabs are the strain gauges. The one in the elbow actually measures the stretch, and the other on the side experiences no stretch and thus acts as a baseline for correcting temperature change. The cable attaching the gauges to the circuitry is held in place with a zip-tie to prevent the strain gauges from getting ripped off accidentally.
The hardware of the sensor consists of a “U” shaped block of aluminum, with two eyebolts attached to either side of the U. You clip onto these two eyebolts and use them to pull or push on either side of the U. As you do this, the inner part of the elbow of the U gets either squished or stretched, depending on which way you are applying the force. Glued onto the inner surface of the elbow is a tiny sensor called a strain gauge. It is a wafer thin foil tab about the size of a pinky nail, with electrically conductive wires drawn onto it. As the aluminum stretches, the strain gauge stretches too, which changes the electrical resistance of the wires etched onto its surface. Using a circuit, we can measure this change in electrical resistance to determine how much the aluminum U was getting stretched, and thus how much force was applied.
In this post I’ll talk about the process that I followed to machine the aluminum U and glue on the strain gauge (actually two strain gauges, as you’ll see).
In this post I’ll talk about the process that I followed to machine the aluminum U and glue on the strain gauge (actually two strain gauges, as you’ll see).
How big should it be?
It was difficult to figure out how big exactly the U should be. If I made it really big, then it might not bend enough for the strain gauge to be able to measure anything. If the U was too small, then it might break under the loads that I wanted to measure, which as you could imagine would be not super duper safe. Somehow I needed to pick the right size and beefiness of the aluminum U so that it would bend some, but not too much, under the load I wanted to measure of a few thousand pounds.
At first, I tried to figure out how large the U needed to be using software designed for modeling materials called Pycalculix. I drew the shape of the aluminum U, added some forces on either side, and it would calculate a diagram of the stresses and strains throughout the object. By tweaking the dimensions of the U I was hoping to get the stresses and strains to the correct levels. However, I don’t think I was using the software correctly, since I was getting numbers that were totally unreasonable, and I didn’t want to sink a bunch of time into learning the software. Therefore, this wouldn’t work for figuring out how big to make the U. This process was useful though, even if the absolute numerical results were useless, since the plots confirmed that the stresses would be nice and smooth around the inner curve of the elbow, and this would be the location with the highest strains in the entire piece, so it would be a good location to get the maximum range out of the strain gauge.
After the failure of pycalculix, I resorted to finding the correct size of U by using the age-old problem-solving technique of guess-and-check. I would start with a U block of aluminum that was too big and for sure strong enough, and gradually make it less strong and more sensitive. There were two ways I could do this. First, I could make the elbow of the U skinnier by shaving off the back of it using the mill. Second, I could change the amount of leverage that the load would actually exert on the elbow by making the eyebolts movable along tracks on either side of the U: The further the eyebolts were from the elbow, the greater the force at the elbow, and the closer to the elbow, the smaller the leverage and the resulting force. Both of these techniques were attractive because they could be done without having to remove the strain gauge once it was attached, so it would be easy to make modifications if the readings from the strain gauge weren’t big enough.
I also decided to make a second U with an even thicker elbow, in case the first one turned out to be too weak to start with.
At first, I tried to figure out how large the U needed to be using software designed for modeling materials called Pycalculix. I drew the shape of the aluminum U, added some forces on either side, and it would calculate a diagram of the stresses and strains throughout the object. By tweaking the dimensions of the U I was hoping to get the stresses and strains to the correct levels. However, I don’t think I was using the software correctly, since I was getting numbers that were totally unreasonable, and I didn’t want to sink a bunch of time into learning the software. Therefore, this wouldn’t work for figuring out how big to make the U. This process was useful though, even if the absolute numerical results were useless, since the plots confirmed that the stresses would be nice and smooth around the inner curve of the elbow, and this would be the location with the highest strains in the entire piece, so it would be a good location to get the maximum range out of the strain gauge.
After the failure of pycalculix, I resorted to finding the correct size of U by using the age-old problem-solving technique of guess-and-check. I would start with a U block of aluminum that was too big and for sure strong enough, and gradually make it less strong and more sensitive. There were two ways I could do this. First, I could make the elbow of the U skinnier by shaving off the back of it using the mill. Second, I could change the amount of leverage that the load would actually exert on the elbow by making the eyebolts movable along tracks on either side of the U: The further the eyebolts were from the elbow, the greater the force at the elbow, and the closer to the elbow, the smaller the leverage and the resulting force. Both of these techniques were attractive because they could be done without having to remove the strain gauge once it was attached, so it would be easy to make modifications if the readings from the strain gauge weren’t big enough.
I also decided to make a second U with an even thicker elbow, in case the first one turned out to be too weak to start with.
Shaping the block
I started with a raw block of 6061 alloy aluminum that was about half the size of a textbook. I went with aluminum because it is fairly strong, and it has the property that it stretches linearly with the amount of force applied. If you read the post on theory I talk more about this stress-strain relationship and why it is important.
Then, I turned the block into a U shape. First, I used the bandsaw to cut the raw stock into a rectangular block of approximately the right size. Then, I drilled out a hole in the middle of the block. This would become the smooth, round, inner elbow of the sensor. I couldn’t make inner corners of the elbow be sharp right angles for a few reasons. First, the entire surface needs to stretch uniformly so that we get an accurate measurement with the strain gauge. If there were abrupt angles, different parts of the surface might stretch different amounts. Secondly, a smooth curve is a lot stronger, since stress becomes concentrated around sharp corners. It could be really bad if the sensor breaks when we don’t expect it.
Next, I took this “donut” and used the bandsaw to cut out a slice of the donut so that I was left with a U shape. Then I used the mill to smooth up all of cuts so far on the inner surface. At this point, the general shape was there, but now I had to attach the eyebolts on either side of the U. I milled out tracks for them so they could be slid along the length of either arm, and then fixed in place by tightening some nuts.
Then, I turned the block into a U shape. First, I used the bandsaw to cut the raw stock into a rectangular block of approximately the right size. Then, I drilled out a hole in the middle of the block. This would become the smooth, round, inner elbow of the sensor. I couldn’t make inner corners of the elbow be sharp right angles for a few reasons. First, the entire surface needs to stretch uniformly so that we get an accurate measurement with the strain gauge. If there were abrupt angles, different parts of the surface might stretch different amounts. Secondly, a smooth curve is a lot stronger, since stress becomes concentrated around sharp corners. It could be really bad if the sensor breaks when we don’t expect it.
Next, I took this “donut” and used the bandsaw to cut out a slice of the donut so that I was left with a U shape. Then I used the mill to smooth up all of cuts so far on the inner surface. At this point, the general shape was there, but now I had to attach the eyebolts on either side of the U. I milled out tracks for them so they could be slid along the length of either arm, and then fixed in place by tightening some nuts.
ATTACHING THE STRAIN GAUGE
The first step to attaching the strain gauge is to prep the surface. We want the gauge to stretch exactly in sync with the aluminum underneath it, so the bond between them has to be perfect. I sanded the surface of the aluminum with fine sandpaper so that it is smooth, but with a tiny bit of texture that the glue will adhere to. Then I used acetone (nasty stuff that’s like paint thinner or nail polish remover) to clean any oils or other chemicals from the surface.
At this point I put on latex gloves to minimize chance that my skin oils would contaminate anything, and used tweezers to take the tiny strain gauge out of its sterile packaging. Now I had to glue the gauge onto the inner surface of the aluminum U’s elbow so that it was perfectly centered and alligned. This would be really hard to do freehand, so I used a cool trick. First I used a piece of scotch tape to pick up the gauge from the back, and now I had the two really good handles of the tape to place the gauge exactly. With this technique it was pretty simple to get the gauge right where I wanted, and the tape held it exactly in its final position. Now all I had to do was peel up one edge of the tape so that the gauge-aluminum interface was exposed, put a dot of glue onto the gauge, and then roll the tape back down so the gauge returned to its exact original position.
The glue of choice for attaching strain gauges is actually just common super-glue. It’s really strong, it sets quickly, and unlike some other glues such as epoxy, it is also really stiff, so the gauge’s movement is totally coupled to the aluminum’s. After adding the glue and rolling the gauge down, I pressed on it with my thumb to ensure a good tight bond. After five minutes the glue was supposed to be set, but on the first two attempts it was still totally liquidy when I pulled away my thumb. I was getting nervous that my glue was shot, the aluminum was weird, and that I was going to run out of strain gauges, since I only had four total. But after some research I discovered that super-glue is activated by moisture in the air, so the low humidity of Colorado was messing things up (#coloradoproblems). So I tried a third time, but this time held the dot of super-glue over a bowl of steaming hot water before pressing it down, and this time the glue stuck. Boom.
I did the same procedure for a second strain gauge on the unstressed edge of the U, since as you will see in the circuit section, I wanted a second gauge to act as a baseline. This made it so there would be some constant against which I could measure the primary strain gauge, and any temperature effects on the primary gauge would be canceled out by an equal temperature change in the secondary.
I soldered some beefy cable to the leads of the strain gauges and then zip-tied this cable around the U so that any jerks on the cable wouldn’t damage the gauges themselves. Finally, I put scotch tape over the actual gauges so that they would be a bit better protected. At this point everything looked pretty good, but it was impossible to tell if I had messed something up. It was time to build some circuitry and see!
At this point I put on latex gloves to minimize chance that my skin oils would contaminate anything, and used tweezers to take the tiny strain gauge out of its sterile packaging. Now I had to glue the gauge onto the inner surface of the aluminum U’s elbow so that it was perfectly centered and alligned. This would be really hard to do freehand, so I used a cool trick. First I used a piece of scotch tape to pick up the gauge from the back, and now I had the two really good handles of the tape to place the gauge exactly. With this technique it was pretty simple to get the gauge right where I wanted, and the tape held it exactly in its final position. Now all I had to do was peel up one edge of the tape so that the gauge-aluminum interface was exposed, put a dot of glue onto the gauge, and then roll the tape back down so the gauge returned to its exact original position.
The glue of choice for attaching strain gauges is actually just common super-glue. It’s really strong, it sets quickly, and unlike some other glues such as epoxy, it is also really stiff, so the gauge’s movement is totally coupled to the aluminum’s. After adding the glue and rolling the gauge down, I pressed on it with my thumb to ensure a good tight bond. After five minutes the glue was supposed to be set, but on the first two attempts it was still totally liquidy when I pulled away my thumb. I was getting nervous that my glue was shot, the aluminum was weird, and that I was going to run out of strain gauges, since I only had four total. But after some research I discovered that super-glue is activated by moisture in the air, so the low humidity of Colorado was messing things up (#coloradoproblems). So I tried a third time, but this time held the dot of super-glue over a bowl of steaming hot water before pressing it down, and this time the glue stuck. Boom.
I did the same procedure for a second strain gauge on the unstressed edge of the U, since as you will see in the circuit section, I wanted a second gauge to act as a baseline. This made it so there would be some constant against which I could measure the primary strain gauge, and any temperature effects on the primary gauge would be canceled out by an equal temperature change in the secondary.
I soldered some beefy cable to the leads of the strain gauges and then zip-tied this cable around the U so that any jerks on the cable wouldn’t damage the gauges themselves. Finally, I put scotch tape over the actual gauges so that they would be a bit better protected. At this point everything looked pretty good, but it was impossible to tell if I had messed something up. It was time to build some circuitry and see!
Cool! You've learned about the hardware of the force sensor. Click here to continue on to the circuitry portion of the project!