Stem Wobble, Housing Tolerance and Typing Noise

Two keyboards built from the same switch model can differ in a way that is easy to hear and hard to name. One sounds tight and deliberate, the other has a faint rattle that shows up on off-centre presses. The difference is stem wobble, and it comes from housing tolerance rather than from the switch design.

We measured wobble directly across samples and logged both the movement and the sound. The correlation was strong, and it explained a complaint that people usually attribute to the keycaps or the case. Understanding where wobble comes from turns a vague annoyance into a specific, checkable property.

What is stem wobble, exactly?

The stem is the part that moves inside the housing, and it needs clearance to move freely. That clearance is wobble: the stem can tilt slightly within the housing before the housing walls stop it. A small amount is necessary, and too much is audible.

It shows up when you press a key off-centre. A press near the edge of the keycap pushes the stem sideways, the stem tilts until it contacts the housing, and that contact is the rattle people hear. A centred press produces almost none of it.

Why does it vary between units?

Because it is a tolerance, and tolerances have distributions. Two housings from the same mould can differ by enough to change the feel, and the stem itself has its own tolerance. The combination decides how much clearance the assembled switch has.

That is why two switches of the same model can feel different, and why a board can have some keys that rattle and others that do not. It is a manufacturing property rather than a design one, and it is invisible in any specification.

How do you measure it?

With a dial indicator or a simple lever gauge, push the stem sideways at the top and read the movement. Repeat at several angles, because wobble is often directional rather than uniform. The numbers are small, typically fractions of a millimetre, and the differences between samples are the useful part.

A cruder method is to fit a keycap and rock it gently, comparing by feel across several switches. That is subjective but it reliably separates the worst samples from the best, which is usually all you need when sorting a batch.

Does wobble affect typing accuracy?

Very little, because actuation happens well before the stem reaches its tilt limit. The switch registers regardless of where in the clearance the stem happens to sit, so wobble does not produce missed presses. That is worth stating, because it is often blamed for them.

What it does affect is confidence. A key that feels loose communicates imprecision even when it registers perfectly, and that perception changes how a board feels over a long session even though the measurements say nothing is wrong.

Where does the noise actually come from?

From the stem striking the housing, and from the keycap rocking on the stem. Both produce short, high-frequency sounds that are distinct from the click of the leaf or the thock of bottoming out. Once you learn to hear them, they are easy to identify.

The cap contributes more than people expect, because a loose fit between cap and stem adds its own movement on top of the switch’s. A tight cap can mask a wobbly stem, which is why the same switch can sound fine under one cap set and bad under another.

Can films or shims help?

Switch films add a thin layer between the housing halves, which tightens the assembly and reduces the clearance. They work, measurably, and they are the standard remedy. The improvement is largest on switches with loose housings and minimal on tight ones.

The risk is over-tightening. Too much material binds the stem, which adds friction and makes the press feel sluggish. That is a worse outcome than the wobble was, and it is the most common way this modification goes wrong.

Does lubrication interact with wobble?

It reduces the noise but not the movement. Lubricating the contact point softens the impact so the rattle becomes a duller sound, which many people find acceptable. The stem still tilts just as far, so the underlying looseness remains.

Treating both together is the usual approach: films to reduce the clearance and lubricant to soften what remains. Doing only one leaves half the problem, which is why partial treatments often produce disappointing results.

How does this compare with keycap fit?

Keycap fit is the cheaper thing to check first, because a loose cap adds movement even on a tight switch. Trying the same switch under two cap sets separates the two contributions immediately, and that test costs nothing if you already own another set.

If the problem persists across caps, the switch housing is the cause and films are the right remedy. If it disappears with a tighter cap, you have solved it without opening anything, which is always the preferable outcome.

Is wobble worse on certain switch types?

It is worse on designs with a taller stem or a looser top housing, and there is genuine variation between families. Some designs control the stem higher up, which reduces visible tilt at the cap even when the internal clearance is similar.

That is worth knowing when choosing, because a design that controls the stem well will feel tight even with average tolerances. It is one of the few structural differences that shows up consistently across samples rather than varying unit to unit.

Does it change over time?

It increases slowly. Repeated presses wear the contact surfaces, and wear increases clearance. Over a long period a board that was tight can develop a little rattle, particularly on the most used keys.

The progression is slow enough that most people never notice it, and it is much slower than the wear on keycap surfaces or the settling of lubricant. It is a real effect and a minor one.

What should you do about it in practice?

Sort your switches before assembling if you are building a board, because the variation within a batch is large enough to be worth exploiting. Put the tightest samples on the keys you press off-centre most often, which are usually the ones at the edges of your frequent reach.

If you are buying a prebuilt board, test the keys you use most, off-centre, before deciding. A board that rattles on the keys you press constantly will annoy you regardless of how good it is everywhere else.

What is the natural next question?

Does the plate hold the switch as tightly as the housing holds the stem?

Often less tightly, and that is a separate source of movement. A switch that sits loosely in a plate can shift slightly under the hand, which produces a sound and a feel distinct from stem wobble. Telling them apart requires pressing in different directions.

Plate-cut design matters here. Some plates grip the switch clips tightly and others are more forgiving, and the same switch behaves differently in each. It is one more reason the board matters as much as the switch.

Can wobble be measured without a gauge?

Approximately, by fitting a cap and comparing movement across several switches by hand. It is subjective but it reliably sorts the worst from the best, which is usually all that is needed when assembling a board.

Doing it with the caps you will actually use is important, because cap fit changes how much stem movement becomes visible. A tight cap hides wobble that a loose one reveals.

Does wobble increase with temperature?

Slightly, because polymers expand and clearances change. The effect is small and it is far below what anyone can feel, but it is real and it is one reason measurements taken in different conditions differ.

For practical purposes it can be ignored. Wear and manufacturing tolerance are larger by a wide margin, and those are the things worth sorting for.

Is a tighter housing always better?

Not beyond a point, because a very tight housing adds friction and can make the press feel stiff. The goal is minimal clearance with free movement, and both matter.

That balance is why the best samples are not the tightest ones. They are the ones with consistent clearance and clean surfaces, and those two together produce the best result.

Does wobble correlate with price?

Loosely, because tighter tolerances cost money. Expensive switches are usually more consistent, and the difference shows up as less variation rather than as a better average.

That is the same pattern as elsewhere: price buys consistency more reliably than it buys peak performance. Buying for consistency is the better use of money.

Does wobble affect the sound of bottoming out?

Slightly, because a tilted stem strikes differently from a centred one. The effect is smaller than the rattle on the way down, and it is usually masked by the louder impact.

Where it is noticeable, films help for the same reason they always do. Reducing clearance reduces every symptom that comes from movement rather than from the material.

Once wobble is understood, the next tolerance to look at is the fit between the switch and the plate, because that determines whether the switch itself moves rather than just the stem. Measuring plate retention is the logical continuation and it explains another family of rattles.