Play a fast repeated note on a weighted keyboard — a sixteenth-note bass pulse, a stuttering lead line, anything that hits the same key over and over — and somewhere around ten notes a second the keyboard stops cooperating. Notes drop out. The ones that survive come back at wildly different velocities. On a synth-action keyboard the same passage is unremarkable. The difference is not build quality, and it is not the sound engine. It is that a hammer action was designed to reproduce the behaviour of a mechanism whose repetition ceiling was set in the nineteenth century.
What a hammer action is copying
A weighted key is not a spring with extra resistance. It is a lever carrying a pivoted mass — a small hammer — which the key throws upward and which then falls back under gravity. The resistance you feel is that mass being accelerated, and the reason it feels like a piano is that a piano works the same way.
The numbers a piano technician regulates to are specific. Downweight, the force needed to depress a key slowly, sits around 48 grams at middle C, with one well-known maker specifying a taper from 50 grams in the bass to 47 grams in the treble. Upweight, the force the key can lift on the way back, is expected to be at least 20 grams and consistent across the keyboard. Key travel is about 10 millimetres, and in a grand that 10 mm of dip drives roughly 47 millimetres of hammer travel through the lever system.
A synth action asks for something like 30 grams, returns on a spring rather than on gravity, and travels less far. Every one of those differences is deliberate, and every one of them shows up in fast passages.
The repetition ceiling, and where it came from
A grand piano can repeat a note quickly because of a mechanism patented by Sébastien Érard in 1821: the double escapement, in which a repetition lever catches the hammer on its rebound so the next stroke can begin before the key has fully returned. With that mechanism, a well-regulated grand permits genuine repeated notes at 12 to 14 times per second.
An upright has no repetition lever. Its hammer must fall back to rest and its key must return most of the way before the next stroke can be made, and its practical ceiling is around 7 repetitions per second — half the grand's figure, for a mechanical reason with a date on it.
Now put a musical demand next to those numbers. Sixteenth notes at 150 beats per minute are 10 notes per second. Thirty-second notes at 120 are 16 per second. A trill between two keys splits the work across two fingers and two hammers, so it clears the ceiling comfortably. A repeated single note does not — it is one key, one hammer, one return path, and the passage runs straight into the mechanism's limit.
Weighted digital actions live between the grand and the upright, and the ones without a triple-sensor arrangement live closer to the upright: without a sensor that registers a new stroke before the key has fully returned, the second note in a fast repetition simply does not report. The instrument is not dropping your notes. Your notes never happened as far as the sensors were concerned.
The second mechanism: velocity is measured, not received
The other half of the story is how a key tells the sound engine how hard it was struck. There is no force sensor under the key. There are two contacts, or two optical thresholds, at fixed points in the travel — commonly around 50 per cent and 75 per cent of the way down. The keyboard scans its keys frequently enough to resolve the interval between them, which is why scanning intervals in the order of 400 microseconds appear in the engineering literature. Velocity is the reciprocal of the time between those two crossings.
Two consequences follow, and both of them are felt rather than read.
First, the note-on message cannot be sent until the second threshold is crossed. Play loudly and the key crosses both in a few milliseconds. Play very softly and the key is still moving between them for a much longer interval — forum measurements of extremely slow playing report the sound arriving tens of milliseconds after the finger began to move. A hammer action makes that interval longer than a synth action does, because the mass has to be accelerated before the key reaches the thresholds at all.
Second, the mapping is a curve someone chose. Very quiet dynamics compress into a narrow band of MIDI values — the softest playing lands in the bottom sixteenth of the 1-to-127 range — so two pianissimo notes that felt different under your fingers can leave the keyboard as the same number.
For piano repertoire, none of this is a defect. An acoustic piano behaves the same way: it is quieter and later when struck gently, because the hammer took longer to arrive. Copying that behaviour is the entire point of a weighted action. For a synth line played with even dynamics, the same behaviour is a mechanism getting in the way of a part that has no use for it.
Weight is not one number
Two actions can specify the same downweight and feel nothing alike, because downweight is a static measurement and playing is not static. What separates them is inertia: how much mass has to be accelerated and stopped, and how quickly the key returns once released. A heavy, slow-returning action can measure a perfectly conventional downweight and still fight a passage of repeated notes, because the return is what sets the repetition ceiling, not the force needed to push the key down in the first place.
This is why the marketing vocabulary is so unhelpful. "Fully weighted", "hammer action" and "graded hammer" describe the presence of a mass, not the behaviour of the return. Two things predict fast-passage behaviour better than any of those phrases: whether the action has a third sensor for repetition, and how the key feels when you release it from half depth and let it come back on its own.
Choosing by repertoire rather than by principle
The useful question is not which action is better. It is what fraction of your playing is piano.
If most of your time is spent on piano and electric piano sounds, buy the weighted action and accept the repetition ceiling — you will meet it rarely, and everything else about the mechanism is working for you. Look for a triple-sensor action if you play repertoire with fast repeated notes; that is exactly the case those sensors exist to fix.
If most of your time is spent on synthesiser leads, clavinet and organ parts, drum programming from the keys, or anything involving glissandi, buy the synth action. Glissandi across a hammer keybed are slow and hard on the fingers, drum parts want the fastest possible repetition, and the mass is doing nothing for a sound that has no hammers in it.
If your playing is genuinely split, the honest answer is two keyboards rather than one compromise. A 49-key synth-action controller next to an 88-key weighted instrument costs less than the top of either range and gives both jobs the right mechanism. A semi-weighted action is the single-instrument compromise, and it is a real one, but it is a compromise in the literal sense: lighter than a hammer action and heavier than a spring, better than neither at what each does best.
The purchase to think twice about
Do not buy 88 weighted keys because they are the serious option. If your parts are synth lines, that mass is a tax you pay on every note and a ceiling you meet on every fast repetition, and the weight is not incidental — a full hammer keybed adds real bulk to anything you carry. The reverse warning holds too: a weighted controller bought for piano playing without a repetition sensor will drop notes in exactly the passages that made you want a weighted controller.
The check that takes a minute
In a shop, play a repeated note on one key as fast as you can and count what comes out. Then release a key from half depth and watch how it returns — a fast, clean return means a mechanism that can be re-triggered soon; a slow, floating return means the ceiling is low. Then play the same passage very quietly and listen for whether soft notes arrive when you expect them or a fraction late. Those three tests take a minute and tell you more than any specification on the box, because none of the three is printed on it.
How this was put together
Five independent sources sit under the figures above: piano-regulation references giving a downweight of about 48 grams at middle C, a maker's specification tapering 50 grams in the bass to 47 in the treble, and an upweight of at least 20 grams; the history and mechanics of Érard's 1821 double escapement, with repetition figures of 12 to 14 per second for a grand and around 7 for an upright; action geometry giving roughly 10 mm of key dip against about 47 mm of hammer travel, and the shallower travel and roughly 30-gram resistance of a spring-return synth action; patent and engineering literature on velocity sensing, describing two thresholds at approximately 50 and 75 per cent of key travel and scanning intervals of about 400 microseconds; and player reports of the delay between finger and sound when playing at the softest dynamics.
The derived figures are ours: the comparison of repetition ceilings against real tempi — 10 notes per second for sixteenths at 150 beats per minute, 16 for thirty-seconds at 120 — the reading of why a trill clears a ceiling that a repeated single note does not, and the argument that return behaviour rather than downweight is what sets the ceiling in a digital action.
