Why Two Keyboards With the Same Switch Feel Nothing Alike

There are two ways to buy a keyboard. The first is to read the switch name printed on the page, decide that one linear switch is much like another, and pick whichever board is cheaper that week. The second is to put two boards that list the same switch on the same desk, type the same paragraph on both, and record what the switch actually does. The first method produces a confident order in about four minutes. The second takes an afternoon, and it very rarely agrees with the first.

We ran both methods on a pair of boards that advertise the same switch family, because the gap between them is where the useful information lives. Two keyboards can share a part number and still differ in where the key registers, how far it travels, how hard it hits the bottom, and what it sounds like when it gets there. None of that appears on a spec sheet, which is exactly why the spec sheet method feels so tidy and tells you so little.

What does a switch name actually guarantee?

A switch name tells you the family: roughly the stem shape, the intended feel, and the factory that made it. It does not tell you the housing tolerance on this batch, the spring that ended up inside, or the lubricant that the assembler used that month. Those three variables move the feel more than the name does, and none of them is printed anywhere. So the first question to ask is not which switch it is, but how tightly that unit matches the one you tried before.

When we pulled samples from two boards and pressed them on a rig, the actuation points sat several tenths of a millimetre apart, and the bottom-out forces differed by a noticeable margin. Both boards were honest about their switch name. Neither was wrong. The name was simply never meant to describe a tolerance band, and the moment you treat it as one, the surprise is guaranteed.

Where does the same switch start to diverge?

The first place divergence shows up is the housing. A switch is a two-part shell that holds a stem, a spring and a leaf, and the clearance between those parts decides how much the stem wobbles on the way down. A loose housing turns a clean press into a slightly gritty one, and that grit is what most people describe as a cheap feel. It has nothing to do with the switch family.

The second place is the spring. Two springs with the same rated weight can have different lengths, different coil counts, and different behaviour over the last millimetre of travel. Our traces showed one board ramping smoothly and the other stiffening sharply near the bottom, which is why the same fingers reported one board as soft and the other as harsh. Same name, different curve.

How much of the feel comes from the plate?

More than most buyers expect. A switch does not float; it is clipped into a plate that is either stiff metal or a more forgiving polymer, and that plate decides how much of the bottom-out force comes back into your fingertip. On a stiff aluminium plate the same switch reads as sharp and short. On a flex-cut polymer plate the identical switch reads as rounder and slightly longer, because some of the energy goes into the plate instead of your hand.

We swapped the same switches between two plates and logged the difference in how testers described the keypress. The words changed before the numbers did, which is a useful reminder that feel is a whole-system property. If you are comparing two boards, the plate is at least as important as the switch, and it is usually the part nobody mentions in a listing.

Why does the case change the sound more than the switch?

Sound is mostly the case. A switch click is a small, dry event; what you hear is that event amplified by a hollow enclosure, reflected off a desk, and coloured by whatever foam or packing sits underneath. Fill the case and the same switch becomes quiet and lower pitched. Leave it empty with a stiff plate and the same switch becomes bright and clangy, with a long tail after the press.

Our recordings made this unambiguous. The keyboard that testers called the louder one was not the one with the louder switch; it was the one with the larger empty volume under the plate. That is good news if you care about sound, because case treatment is cheap and reversible, while a switch swap is neither.

Does actuation point move with temperature and wear?

It does, though slowly enough that you will not notice day to day. Polymer parts relax under repeated load, springs settle, and lubricant migrates away from the contact surfaces. Over a long run, the key tends to register slightly earlier and bottom out slightly softer than it did new. The direction is consistent even when the size of the change is not.

We saw a measurable drift in actuation point after extended use, small enough to be irrelevant for typing and large enough to matter if you are tuning a board for gaming. Warm rooms shift the same numbers as well, which is why a board can feel different in summer than it did in winter without anything having broken.

What does a force curve tell you that a spec sheet cannot?

A force curve is a plot of resistance against travel, and it shows the shape of the press rather than two isolated numbers. From it you can see whether the switch ramps evenly, whether there is a tactile bump and where it sits, and whether the force spikes right at the bottom. That shape is what your fingers actually read, and it is invisible in a two-number spec.

Reading curves also settles arguments quickly. Two switches that feel equally heavy at rest can have completely different shapes, one flat until it suddenly firms up, the other rising the whole way. When testers disagreed about which board felt heavier, the curves explained why: they were describing different points on the same stroke.

How do you test two boards without fooling yourself?

Use the same desk, the same chair height, the same keycaps and the same paragraph, and switch back and forth within a single session rather than a week apart. Memory for feel is poor over days and surprisingly good over minutes. Keep the boards at the same height and angle, because wrist position changes perceived weight far more than the switch does.

It also helps to test blind where you can. Once you know which board cost more, your hands tend to confirm the price. Our cleanest results came from sessions where the tester could not see which unit was on the desk, and the preference flipped more than once under those conditions.

What should you check before you trust a switch name?

Check the plate material, the case construction, the mounting style, and whether the listing mentions foam or gaskets. Those four items tell you more about how the board will feel than the switch family does. Then check whether the seller states a tolerance or a batch, because a seller who tracks batches is usually a seller who cares about consistency.

Finally, treat the switch name as a starting hint rather than a promise. It narrows the field from hundreds of boards to a handful, and at that point the plate and the case decide the rest. Boards that feel nothing alike usually diverge in the parts nobody photographs.

What does a settled board feel like compared with a fresh one?

A fresh board has tight housings and un-migrated lubricant, so the first week produces a slightly firmer, slightly quieter press than the same board produces later. After sustained use the parts bed in, the press becomes a touch smoother, and the sound drops a little in pitch. The change is small in absolute terms and easy to notice in a side-by-side test.

That settling is worth waiting for before you judge a purchase. A board that feels slightly stiff out of the box often becomes the board you wanted a fortnight later, and a board that feels perfect immediately may already be at its best on day one. Either way, the comparison you make in the shop is not the comparison you will live with.

How do you keep a comparison honest over time?

Keep one reference board that never changes and use it as the anchor in every session. Press the same short paragraph on the reference first, then on the candidate, and write down what changed while the memory is fresh. Without an anchor, every board gets compared with a half-remembered one.

Log the conditions as well: room temperature, keycap set, desk surface, and how long you have been typing that day. Those variables move the result more than people expect, and a note taken under different conditions is not comparable with an earlier one.

The same question reappears one layer down, at the spring and the stem. Once the plate and case are accounted for, the next thing worth measuring is how a switch behaves over its first month, which is where the difference between a fresh board and a settled one becomes obvious.