This article is a guest-post by one of our team dads, David Beaver.

At the end of VEX matches, the field is usually littered with robot parts . . . mostly nuts and loose bolts.

One could ask: How can we keep our robots from falling apart?

VEX Keps NutYou’d think the toothed washer on the standard VEX “keps” nuts would have some effect on keeping the nuts on, but the field litter shows they’re not as good as we wish.

Our team did some research and experimenting on this, and the conclusions are clear: which nuts or washers you use doesn’t matter, but you need to make your connections TIGHT!

Here’s why and how:

The Science

Bill Eccles, Bolt Science logoAs always, there is real science behind the optimal use of bolts and nuts. For more information than you ever wanted on the subject, see, the web site of Bill Eccles, a Ph.D. specializing in “The Self-Loosening of Threaded Fasteners.” This site may be quite fascinating to the engineer in you (people have struggled with bolts loosening since they were invented), but if you don’t want to browse the site, here are key points from Bill’s research:

  • Any two plates fastened by bolts and nuts WILL, WITHOUT QUESTION, loosen over time. . . . IF those two plates can move against each other: wiggling, sliding, or vibrating.
  • At a micro scale, the stretching of the bolt along with the flexing of the threads in a tightened bolt and nut acts like a spring to keep the two plates in contact.
  • The friction between the two plates caused by the tightness of that “spring” must be greater than the forces that are acting to make the plates move against each other. If the friction force is sufficient, then the plates won’t move and the bolt won’t loosen.
  • A component of this equation is the friction between the face of the bolt and the nut against the two bolted plates. Lock washers, spring washers, and keps nuts actually have less contact friction than a plain nut (because less surface area is in contact), and will actually loosen faster. (Details on Bill’s poster: “Why Nuts and Bolts Can Self-Loosen.”)
  • The friction from the nylon insert in a “nyloc” nut is much less than the contact friction of the faces or the threads, so the nylon doesn’t help (in the way you think it would).
  • Depending on the size, length, material strength, and application of any bolt, there is an ideal tightness (torque) that stretches the bolt and flexes the threads to achieve the right spring force on the plates . . . without permanently deforming the bolt. This force can be calculated, and the right torque value can be specified for the person putting the parts together.

(After learning all this, we had fun looking for bolt-and-nut applications in the real world: buildings, bridges, etc. We didn’t see many keps nuts!)


As good students of the scientific method, we ran an experiment to test our new hypothesis: that the tightness of the connection, not the type of nut, is what matters. We built a simple VEX chassis and attached a vertical plate with some long C-channels extending out horizontally from it; in this setup, bolts are very prone to being loosened from vibration. For added fun, we taped some balled-up paper to the wheels, which shook the chassis madly. And then we drove it up & down the hall many times, trying different configurations: 2, 3, and 4 bolts; every kind of nut (keps, plain, nyloc, washers, and even two plain nuts jammed on top of each other), and various levels of tightness.

Standard Nut

Nylock Nut

VEX Keps Nut

Keps Nut

Our experiment absolutely validated the science and the hypothesis. When the connections were very tight, the C-channels withstood the vibration, and didn’t loosen. The looser the connection, the faster the joints failed (sometimes after just a few feet of bumpy driving!). And connections with plain nuts lasted just as long as connections with any other type of nut, when the nuts were tightened to the same torque.

(Our team is in high school now, but it occurred to us that this would be an excellent middle school science fair experiment!)

Applying This to VEX

After learning these lessons, we bought some better tools and changed the way we assemble our robots, and our 2018 robot was rock solid and never dropped parts. Here are our new “Rules for Building a Solid Robot”:

  • Make every joint as tight as you possibly can.
  • There’s just no way to make a joint tight enough by turning the bolts with the standard VEX hex keys; you’ll quickly strip both the socket head in the bolt, and the end of the hex key.
  • You must tighten the nut side of the connection: use the hex key only to keep the bold head from spinning. Once you get the nut tight enough, you can release the hex key and the bolt head’s friction will keep it from spinning as you keep tightening the nut. (In most cases, just holding the bolt with your fingertip is good enough.)
  • There’s no way to tighten nuts sufficiently with the standard VEX 3-inch wrenches. You need to get a socket head firmly onto the nut.
  • If you do one thing to change the way you put your robots together, buy a bunch of “nut drivers”—screwdriver-like tools for turning nuts. The best price on the Klein Tools brand that we found for the 11/32” size needed for VEX nuts is at Zoro Tools: $7.31 each. You can easily get nuts VERY tight with these. [Ed. note: sells Bondhus brand for $4.49 each.]
  • The next tools to buy are long (8″+) hex keys. They are super-useful for getting to the heads of deeply-buried bolts, and also spinning bolts on or off once they’re loose. You need 5/64″ size for motors and shaft collars, and 3/32″ for all VEX button-head socket bolts. We bought “Eklind” brand, like these at Zoro. (Note: don’t buy the “ball end” wrenches . . . they strip very quickly).
  • Even these good hex keys will strip at the ends with use. Every few weeks we grind down the stripped ends of our wrenches with a Dremel and start over fresh.
  • When connecting metal parts, align them so the largest possible surface area of their faces is in contact (e.g., use the 3-hole sides, not the 1-hole side, of C-channels face-to-face if you have a choice).
  • When putting bolts into parts, feed them so that you can get the best possible grip on the nut with your nut driver (e.g., put the bolt head on the inside and the nut on the outside of a joint). And . . . Make Them Tight!
  • Use multiple bolts on a joint wherever you can. And . . . Make Them Tight!
  • It’s hard to get the nut driver onto a nut in the inside corners of C-channels… better to put bolts in the mid-line holes of the 3 rows so your nut driver can get a solid grip on the nut.
  • Do everything you can in your structure to brace joints against moving (e.g., add triangulating supports). And . . . Make Them Tight!
  • Don’t feed bolts through your parts in such a way that making them tight bends the metal; they won’t stay tight. (E.g., if you need to bolt the small side of a C-channel to something, don’t put a long bolt all the way through both sides of the C-channel; put a short bolt in from the inside of the short side.)
  • Nyloc nuts are useful for a few cases: when you *don’t* want a tight connection between two parts, such as those that rotate against each other around a loose bolt, or for putting together plastic parts. (All these rules of tightness won’t work on plastic parts like bearing flats, because the part will crack before it’s tight enough.) Other than that, the only value of nyloc is to keep the nuts from falling off *after* they loosen.
  • There’s no need to buy keps nuts rather than plain nuts, but if you have lots of keps nuts, don’t throw them away, just . . . (all together now) . . . Make Them Tight!