Lubing 101: what matters, what doesn't, and where to stop

Lubing a switch is one of the few mods that survives every fashion cycle in this hobby. It changes the sound. It changes the feel. It does both more than keycaps do, more than plate material does, and more than most of the things builders argue about on Friday nights. It is also the mod most likely to swallow a weekend, ruin a tactile, and convince a first-time builder that they hate a switch they would have loved out of the bag. The honest version of this guide is short: most switches benefit from lube, many builders over-lube, and there is a minimum protocol that captures most of the audible win without the rabbit hole.

What lube actually does

Two things, and they are easy to keep straight. First, it damps the small high-frequency noises that come from plastic-on-plastic contact inside the switch — the scratch of the stem rails sliding against the housing, the tick of the spring rubbing against the bottom housing on its way down. Second, it dulls the spring's own ringing, the metallic ping that some builders find charming and most find distracting on a quiet build.

Neither of those changes is dramatic on a single switch. Both of them are dramatic across a board of seventy. A stock linear has a thin, slightly papery edge to the bottom-out and a faint hiss on the upstroke. The same switch with a careful coat of Krytox 205g0 on the rails has a deeper, rounder bottom-out and a quieter return — the difference between a typing sound that distracts and one that recedes. Theremin Goat reviews have been quietly cataloging this delta for years, switch by switch, and the pattern is consistent: smooth linears get smoother and quieter; rough linears get usable; tactiles get unpredictable.

That last clause is the one builders skip past, and it is why this guide exists.

The minimum that gets eighty percent of the benefit

Strip the protocol down to what actually moves the needle.

Use Krytox 205g0 for linears. It is the boring default for a reason — viscous enough to coat without sliding off, light enough not to mute the switch into mush. GPL 105 is the standard spring lube; a film on the springs kills ping with no real downside.

Open the switch. Pull the stem. Apply 205g0 with a small brush — the cheapest art-supply brush works fine — to the four stem rails (the long flat surfaces that ride against the housing), in a thin, even coat. Lay a film along the rails of the lower housing where the stem rides. Drop the spring into a baggie with two or three drops of GPL 105, shake for ten seconds, fish out the springs. Reassemble.

That is the protocol. Stem rails, lower housing rails, spring. Twenty seconds per switch once a builder has the rhythm, plus another fifteen for the open-and-reassemble. A board of seventy takes an evening and a podcast.

Where most builders go wrong

Three failure modes, in descending order of how often they show up in builds people end up disliking.

Lubing a tactile bump into mush. A tactile switch's bump comes from a specific shape on the stem riding past a specific shape on the leaf. Lube on the bump-bearing surface — the leg of the stem that contacts the leaf — softens that interaction. A pronounced bump becomes a vague swell. The switch the builder bought for its tactility now feels like a slightly-grainy linear. The fix is to skip the bump face entirely on tactiles, or to use a much lighter lube (Tribosys 3203 is the conservative pick) and a much thinner coat. The honest rule: if the switch was bought for its bump, lube only the spring and the housing rails, and leave the stem alone for the first build. Add stem lube on the second board if it still feels scratchy.

Touching the leaves. The two metal leaves inside the bottom housing are the electrical contact. Lube on the leaves causes inconsistent actuation, dead keys, and the kind of intermittent failure that only shows up under deadline. Brushes drift; that is why watching where the brush goes matters. If a builder is unsure whether they touched a leaf, they touched a leaf — wipe it with a dry cotton swab and move on. There is no sound benefit from lubing leaves, and the cost is a switch that fails in three months.

Lubing brand-new switches that do not need it. This one stings because the time is the cost. Modern factory lubing has gotten meaningfully better since the Cherry MX2A redesign and the rise of the HMX and Gateron Pro lines. The Cherry MX2A Red ships with a factory lube job that a careful manual application barely improves on for a daily-driver build. The HMX Cloud Linear lineup is sold pre-lubed and the coverage is tidy; tearing seventy of them down to add a second coat is, in most cases, a way to make a clean switch worse. The Gateron Oil King sits in a different bucket — factory lube is competent but uneven, and a careful manual pass is a real upgrade — but the principle holds: check what the switch ships with before deciding to lube it.

When not to lube

There is a specific reader for whom the right answer is: do nothing. The first-time builder who has never owned a custom should put the switches in stock, type on the board for a month, and only then decide whether the build needs a sound change. Lubing a switch the builder has never typed on stock is solving for a problem the builder has not yet diagnosed. It also makes the lube job harder to evaluate — there is no before to compare the after to.

Skip the lube on factory-lubed switches in their first board, on any tactile until the second build, and on silent switches across the board (the dampening pads do most of what lube would do, and lube interacts unpredictably with the pad material). Skip it on a board that will be rebuilt within a few weeks — pulling lubed switches out of hot-swap sockets is a way to spread lube to places it should not be.

The time math also matters. A careful lube job on seventy switches is three to four hours for a first-timer. If the builder is on the fence about whether they like the switch at all, that time is better spent typing on the stock build and figuring out the answer. Lubing commits a builder to the switch in a way that buying it does not — re-selling lubed switches is harder, and the time investment biases the builder toward defending the choice.

The practical take

Build the first board stock. If the switches have factory lube, trust it. Type for a month. If, at the end of that month, the build sounds papery, scratchy, or pings audibly in a quiet room, lube the second board with the minimum protocol — rails, lower housing, spring — and stop there. Skip stem lube on tactiles entirely on the first attempt. Do not lube leaves, do not over-coat, and do not lube a brand-new factory-lubed switch on the assumption that the factory job is rough. It is, increasingly, not.

The rabbit hole is real and it is interesting and it has produced some genuinely sublime builds. It is also optional. The eighty-percent version is enough for almost everyone, and the people who eventually want the last twenty percent will know it because the build told them — not because a forum did.