Helical Compression Spring Rate

A helical spring is a torsion bar in disguise. Pull on the ends of a coil and the wire itself is not stretched or bent to any useful degree — it is TWISTED. Follow one turn of wire around: the axial load sits at a lever arm of half the mean coil diameter from the wire's own axis, so it applies a torque of $FD/2$ to a bar that happens to be curved into a helix. Everything else follows from the standard torsion result, and the fourth power on the wire diameter arrives with it, because the polar second moment of a round bar goes as $d^4$.

That fourth power is the fact worth taking away from this page. Ten per cent more wire diameter is $1.1^4 = 1.46$, so a forty-six per cent stiffer spring. Twenty per cent more wire is more than double. Nobody's hand can feel a two-thousandth of an inch, which is why springs are calculated rather than eyeballed, and why a spring shop that substitutes "the next gauge up" without recalculating has changed the part rather than approximated it.

The three inputs that most often go in wrong are all definitional, not numerical. D is the MEAN coil diameter — the outside diameter less one wire diameter — and using the outside diameter makes the spring come out too soft by the cube of the error. n counts ACTIVE coils only: on the common closed-and-ground end, the two end coils are seated flat against the plate and do not flex, so n is roughly the total turns minus two. And G is the SHEAR modulus, not Young's modulus: for steel that is about 79 GPa against 200 GPa, so mixing them up gives an answer two and a half times too stiff.

The ratio $C = D/d$ is called the spring index, and it quietly governs how manufacturable the spring is. Below about 4 the wire is hard to coil without cracking; above about 12 the spring tangles in the bin, buckles under load and holds its free length poorly. Between those the spring maker is comfortable, and within that band the cheapest way to hit a target rate is almost always to adjust the active coil count, which is a machine setting, rather than the wire, which is an inventory decision.

A rate is only half of what the machine needs; the other half is where the spring sits, and that is where most installation errors come from. Deflection is measured from FREE LENGTH, never from the installed length. A spring 60 mm free fitted into a 40 mm space is already 20 mm deflected and already pushing $20k$ before the machine touches it — that standing force is the PRELOAD, and installing a spring short is how a designer sets the force at the start of the stroke independently of the rate. Compress it a further 10 mm and the force is $k \times 30$ mm, not $k \times 10$. Extension springs carry a second offset on top of that: most are wound with the coils pressed together under an INITIAL TENSION, so they make a real force at zero extension and only then begin to follow the straight line, which makes the honest form $F = F_i + kx$. If you are measuring a rate rather than calculating one, take TWO loaded points and divide the change in force by the change in length — that cancels the initial tension, any error in the free length and any friction in the fixture, all three of which bias a single-point reading the same optimistic way.

Three limits sit outside the equation. Solid height — total coils times wire diameter — is the shortest the spring can ever be, and a spring driven solid stops being a spring and becomes a rigid post, usually to the surprise of whatever was driving it. Buckling: a compression spring longer than about four times its mean diameter folds sideways under load unless it is guided over a rod or inside a bore. And SET: a spring overstressed at any point in its life comes back with a shorter free length forever after and quietly makes less force at every installed position than the day it was fitted, which is usually the answer when a mechanism goes weak with age and the spring "still looks fine". The relation itself is the classical one worked through in A. M. Wahl's Mechanical Springs and reproduced in every open machine-design course.