Set Screws (and Why You Shouldn’t)

Set screws are one of those fastenings that has its roots in the early industrial revolution. Think of the era when we didn’t have splines to connect shafts; we used keyways and keys. Set screws are a similarly crude design for fastening two things together but really don’t belong in aviation. In the engineering world, set screws are often strictly prohibited in environments exposed to vibration or requiring precision. Ironically, precision doesn’t apply to a lot of aviation, but vibration might as well be our motto.

The problem with set screws is not one of personal taste or preference and isn’t something to get worked up about. Set screws have a design flaw that make them extremely prone to coming loose, and I’ll try to explain. Let’s first look at a typical bolted joint for reference.

The bolt in orange is tightened with the pink nut across two plates (blue and gray). As the nut is tightened, the center section of the bolt deforms like a spring and is stretched in the direction of the arrow. It is this spring action that creates a proper bolted joint, which is resilient to vibrations and temperature changes. Using this same logic, examine the joint with a very short bolt, which has a resulting very short spring. This joint is slightly less resilient as a result.

It is important to note that in aviation, most of the fasteners we are used to incorporate two separate types of fastener retention. This is where lock washers, threadlocker, lock wire, lock plates/tabs, cotter pins, and lock nuts come into play. A bolt with a Nyloc nut would have: 1) spring tension from bolt torque and 2) nylon friction locking. A rod end with a jam nut has: 1) torque of the jam nut and 2) threadlocker.

Now let’s take a look at set screws, which have some distinct differences. First off, unlike the examples above, which create tension in the bolt as it is tightened, set screws use compression of the screw.

Note that the only part of the set screw in compression is the section protruding from the threads. This length is determined by the gap left in the assembly, which is usually intentionally minimized to near zero. The “spring length” of an installed set screw is far less than even the shortest AN bolt grip length. An AN3-3 has a grip of 0.062 inch, while for set screws 0.010 inch would be a typical protrusion and 0.030 inch would be very loose. This is why when you tighten a set screw, there isn’t much give or “feel” to it. It touches the part, and then is immediately tight.

Of course, if you keep cranking on it, you will deform the part beneath it, which improves the joint but is usually undesirable from a functional or serviceability standpoint. Nobody likes to see Bowden wire ends that are zigzag shaped and chewed up at the end. Set screws set you up to make a counterintuitive decision because if you want the most resilient joint, then you should make the tolerances loose and deform the components during assembly. A tight tolerance joint that won’t allow component damage is incredibly more prone to failing with a set screw.

The other fundamental problem with using a compression spring for joint retention is that metals are twice as strong in compression as they are in tension. Strong is good, right? Unfortunately, this means the “spring” will deform half as much. In order to create the same amount of spring deformation—aka joint resilience—for a set screw, it would have to be torqued twice as much as a bolt in tension (assuming the same spring “length”). I dare you to attempt tightening a #8-32 set screw to 70 inch-pounds. The female threads will strip or the tool will fail long before you achieve torque.

Why not use threadlocker? This is the obvious solution, right? Well … it isn’t that simple. Many people operate under the false assumption that threadlocker helps keep your bolted joints tight. The truth is by the time threadlocker is doing its job, the joint has already failed, meaning the joint compression has been lost. Threadlocker should be compared more similarly to lock wire or cotter pins; it keeps the fastener from falling out. This is OK in many applications and is exactly why we use cotter pins and lock wire in many places. However, with a set screw, usually we need that joint to remain tight, clamped, compressed, etc. Think of a solid wire control cable connection. If the set screw comes loose, the wire can slide out of the joint and render your mixture, throttle, or machine guns useless.

A commentary on inferior fasteners would be incomplete without at least consulting AC 43.13-1B. While it does mention the various specs of set screws in Table 7-11, it does not provide general guidelines for torques or use cases. The best source of information would be the applicable AN, MS, or NAS spec document for each of the following types of set screws.

In summary, builders should be wary whenever plans or instructions call for a set screw, especially forward of the firewall. Yes, they may have been working for years in many applications, but I think we should pay attention to the things that cause general aviation accidents and do our part to prevent them. Failing control cables and cable connections are a somewhat regular source of general aviation emergencies. There is always a better method than a set screw, which I will dive into next time.