Preload

[Daniel Beardsmore]

Currently, the [wiki]force[/wiki] page has "preload" down as: "The force required to begin depressing a key. Also reported to be the force on the spring at rest, i.e. from the keycap."

The force curve for an ideal linear switch fits y = mx + c. c is typically a figure between 30 and 40 cN.

These figures are presumably given with the keycap removed. As I understand it, the spring in a switch is partially compressed when the switch is assembled, so presumably that is what c represents. Whether "preload" represents this, or the keycap weight, I don't know, but I would imagine it would be the former.

Since the spring is under a constant minimum tension, would this not weaken it after a while, just as how it's harmful to leave a keyboard with the keys depressed for long periods?

[Muirium]

Interesting question. My intuition is that springs are contained slightly compressed inside switches at rest. If you were to remove the spring and let it assume its natural shape, it will lie a little longer. This force is what you must counterbalance when pressing the cap to make it move at all. Any less force, and the spring is still pushing against the switch's insides.

Hooke's Law (which is analogous to a straight line formula as you described) is for springs that aren't damaged. Any reasonable switch won't exceed its spring's design limits. So I'd wager that no damage happens to springs contained in static, relaxed switches. Even fully depressed switches may not damage their springs, for all I know, at least from the rigid perspective of linear over compression as read by the force curve.

Fatigue is what wears things down. Motion, friction, and use. Every spring falls off Hooke's Law eventually with use. But I don't think this is what happens in keyboards stored badly under a pile of heavy junk. Just going by my intuition though.

[Halvar]

Currently, the [wiki]force[/wiki] page has "preload" down as: "The force required to begin depressing a key. Also reported to be the force on the spring at rest, i.e. from the keycap."

Like Muirium wrote, both is the same, and c is the same, too. That how Hooke's law works. A force c on the spring is needed to keep it at the length that is has in non-operating state. This force is exerted on the spring by the switch case that keeps the spring at length (and partly also by the weight of the cap and the stem, but that's a small part -- compare the preload force of typical switches (20-50 gf) to the weight of a key cap (1-2 gf)).

The force that the finger has to exert on the spring to start it moving down is exactly the same force c.

I agree that springs should weaken over time by this, but as far as I know, this effect is very small. Springs weaken much more by usage, i.e. when the are compressed often, than by a constant prestress.

[Daniel Beardsmore]

I found the idea that preload referred to the keycap, suspect; I don't know what they weigh (you reckon 1–2 g), but I know it's far less than the y intercept on force graphs! You can rest a keycap on another key and it won't depress at all.

What I do assume is that the force graph should start at (0, 0), not (0, 40), and I presume that the partial compression of the spring is what accounts for that, possibly to prevent the keycaps from being too bouncy. I don't know how the weight of the slider comes into it, if the spring is already compressed slightly (i.e. the spring isn't tensioned by free weight, but by compression within the space it occupies).

So, if a keyboard is stored with the springs compressed, that won't really harm it? Is this just another widespread myth?

Basically I'm just trying to settle the definition of preload.

[Halvar]

I would take the part "i.e. from the keycap" out. This part is wrong, because the keycap weight plays no role at all for the value of the preload force if the spring is precompressed, which it is in all switches I know.

If the keycap is lighter or heavier only means that the case needs to exert more or less force, but the sum of both is always the same, namely the force that is given by the spring's properties and the length of the spring when the switch is in resting state.

[Daniel Beardsmore]

Looking at the graphs on the [wiki]force[/wiki] page, there's something else weird.

Take the graph for ML. ML requires ~37 cN of force to start pressing the switch. When releasing the switch, the force against your finger is only 20 cN at the point that the slider position returns to 0 mm. If you let the slider return to 0.1 mm, and then start pressing the switch again, would you feel ~23 cN or ~40 cN?

What happens between 0 mm and 0 mm to cause the initial force to jump from 20 cN to ~37 cN?

[scottc]

ML is a really strange switch. All of the tactility is on the upstroke, which just feels completely wrong.

[Daniel Beardsmore]

That was just an example. The MX Black graph is the same.