Stem Review

We’ve once again brought back our favorite engineer, Jason Krantz. Jason’s a mechanical engineer whose graduate work focused on the intersection of composite materials and finite element analysis. Jason has worked for several companies in the bicycle industry and never fails to amaze me with the depth of his cycling related knowledge. He adds real value to all of the tests we perform.

Disclaimer: A lot of typing and numbers have gone into this article and we apologize in advance for any typos, but would warn that the possibility of mistakes is present.

Some Initial Notes on Testing

Testing Method: Each stem was mounted in the fixture and the testing performed 3 times and then averaged. All stems tested were 11cm with the exception of a few which are not available in those lengths. Each stem was mounted with it’s recommend torque specs and preloaded with 20 pounds of force applied 10 inches from the stem (about half way between a road bar and mtb bar) Once preloaded the equipment was zeroed and another 100 pounds of force was added and a measurement recorded. The measurement was taken at the point of load, again 10 inches from the stem centerline.

Loads Tested: While the actual load doesn’t matter because the response of the structure is linear. [That is, a stem with a s stiffness of 150-foot pounds per degree will deflect one degree with a torque of 150 foot-pounds and two degrees with a torque of 300 foot-pounds.] Jason selected a load of 100 pounds because it’s a nice round number, and pretty realistic to what some riders may experience. For a sprinting rider, let’s say he is pushing on the left pedal with 300 lbf (in other words, he can squat 600 lbs). The pedal is about half as far from the frame centerline as the handlebar is, so it would take a reaction force of 150 lbs at the right-hand bar to counteract the pedaling force. This is the same as applying 75 lbf up on the one side of the bar and 75 lbf down on the other; the moment at the stem is the same.

Normalizing for Stem Length: In theory, it’s possible to measure a particular stem of one length and extrapolate its results to longer or shorter versions of the same stem model. In technical terms, this is “normalizing for stem length.” We can do this because torsional displacement is directly proportional to length for a tube of a given cross section. It’s true that a good engineer will use thinner walls on shorter stems (which see lower stresses) and thicker walls on longer stems (which see higher stresses). Still, normalizing to length within a single stem make and model is valid to a first approximation, at least with metal stems. A composite stems is a different beast, because layups (AKA laminate schedules) can vary significantly with length.

While torsional displacement is directly proportional to stem length, bending displacement is proportional to the cube of the length. Our first pass at a stem test setup measures combined bending and torsional displacement, so it would not be valid to extrapolate to longer or shorter stems, even within the same model, at least if you’re looking for absolute values.

If you’re interested in relative values–for example, if you want to know whether stem A is stiffer than stem B–then these results should be quite useful regardless of the actual length you intend to use.

For this test we used 11cm stems in every case except 2 (Pro Vibe Sprint Carbon and Syntace F119) which are both 10.5cm. The stems were similar enough in length that we did not normalize any stems in this test, but in the future we may test some stems that need to be normalized for a fair comparison.

Stiffness to Weight Ratio: This was calculated using =((1/avg. defl) / weight) * 1000

Bolt Material: One question we wanted to answer was if there was a noticeable difference between stems with titanium and steel bolts. We selected a few random stems and tested them both with titanium and steel bolts and found no meaningful difference in deflection in stems with one material over the other.

Mounting Direction: We also wanted to see if mounting a stem with a rise vs a drop made a difference in the deflection. While we were able to measure a difference, less than 0.1 mm at 100 pounds, it was not enough to say that there is a meaningful difference, and was likely within margin of error of the testing.

Notes about Stem Angles. One other item we wanted to test was the difference in like stems with different angles. We tested a couple of stems that are available in both 7 and 17 degree angles and found that 17 degree stems deflected an average of 7% less. The tested stems vary from 5 to 10 degrees which should result in deflection variations of up to about 3%, so keep that in mind when comparing two stems of different angles.

Carbon Fiber vs. Aluminum: – Carbon fiber stems don’t offer the weight savings that, say, carbon fiber rims do. There’s a reason for this: stems are regularly stressed in three ways: bending, torsion and shear. In other words, the stresses in a stem are nearly isotropic (uniform in all directions). Carbon fiber is highly anisotropic (stiffer/stronger in one direction than in the others). When you try to lay up anisotropic carbon to deal with isotropic stresses, you end up with what engineers refer to as “black aluminum,” or pseudoisotropic carbon–and a weight that’s very close to an aluminum stem.

There may be other advantages to a carbon stem; vibration damping is one that is often put to the forefront of the debate. It’s true that composite structures generally damp vibration better than metallic ones, and there may be real-world damping benefits to a carbon stem. One way to test stem damping would be to wire an accelerometer to the stem and then strike it gently with a hammer, reading the accelerometer data to see how quickly the vibrations dissipated. This ring-down test may be on the Fairwheel menu for a future article, but is outside the scope of this test.

Mixed materials. Hybrid stems, a carbon overlay on an alloy stem. We tested only a couple of these but neither did well, we’ll do more testing on others in the future to develop out thoughts on this medium further.

We tested two similar stems one in full alloy (7s) and one in hybrid (vibe carbon) to see if there was an advantage to the hybrid stems. Both are the same weight and same dimensions with the same puzzle lock clamp. The difference being that the Carbon has thinner alloy, reinforced with carbon. Deflection was noticeably different, the full alloy was significantly stiffer than alloy/carbon.

To speculate on why tests put carbon/aluminum stems on the flexible end of the spectrum, one might consider two competing design constraints. First is galvanic corrosion. When aluminum and carbon touch each other, they’ve made a weak battery. The electron-swapping that goes on will cause an aluminum/carbon part to fail in short order. To avoid this, manufacturers use a layer of insulating material. This layer is often a fiberglass/epoxy composite but could be plain epoxy. Neither material is as stiff as aluminum, so you’re taking a stiffness loss right away.

Secondly, the market won’t tolerate a carbon-wrapped aluminum stem that is both more expensive and heavier than its plain-aluminum counterpart. So manufacturers are forced to use thinner-walled aluminum forgings for their carbon-wrapped stems than they do for their straight-up aluminum stems. Effectively, they’re replacing relatively stiff aluminum with relatively compliant fiberglass and/or epoxy.

Making matters worse, some carbon-wrapped aluminum stems use a 0°/90° weave for aesthetic reasons. This weave’s stiffness reflects the dreaded pseudoisotropy referred to earlier, and is particularly compliant in shear–which, in a stem, determines torsional stiffness.

The net result is a carbon-look stem that has adequate strength but more compliance (and possibly better vibration damping) than a “pure” aluminum stem.

To be clear, the above is what engineers call a SWAG: a Scientific Wild-Ass Guess. It’s a reasonable first pass, but there may be other factors at play that we haven’t considered and which will be further explored in future tests.

Now on to the results

Thoughts on Each Stem

A Few Graphs

Weight

Stiffness

Stiffness to Weight

Closing Thoughts

That’s as far as we go with the first instalment of our stem testing. We do have future installments planned with more models being tested as well as other variations such as different lengths and materials. We’ll also be posting handlebar tests for both road and mtb as well as some other tests we’ve devised.