Springs & Spring Design: The Complete Engineering Guide — The 0.615 Wahl Correction That Sets Every Coil's Peak Stress, the 45→65% Allowable Jump That Presetting Buys, the θD/360N of Shrink That Sizes Every Torsion Mandrel, 5.26·D of Slenderness Before Buckling, and the ₹5,000-Crore Indian Industry Wound Into Every Die Set

This series has taken apart bearings, gears, fasteners and seals — the machine elements that hold machines together — and this guide takes apart the one…

·

Springs & Spring Design: The Complete Engineering Guide

Every guide in this series has taken apart a machine element and found the same architecture underneath: a fastener is a preloaded spring, a seal is a spring with fluid on one side, a bearing is a spring that rolls, a die set is a spring with a cutting edge bolted to it. This guide takes on the element they were all secretly made of — the spring itself: the machine component whose entire purpose is to deflect under load, give force back, store energy, and do it millions of times without ever losing its calibration.

It is also the element with the most quotable arithmetic in mechanical engineering. A spring's stiffness obeys one of the most compact equations in design — k = \dfrac{G d^4}{8 D^3 N_a} — in which the wire diameter sits at the fourth power, which is why spring design is the art of choosing a wire size you can't easily adjust for anything else. Its peak stress carries a curvature correction named after a 1920s researcher, and the difference between a spring that fails at 100,000 cycles and one that survives 100 million cycles is not the design equation but three manufacturing steps — presetting, shot peening and heat setting — that change what the material is allowed to carry by 20 percentage points of its own tensile strength.

And there is a version of this element at every scale, from the 12 ₹ spring holding a 3D-printer bed to the 45,000-newton die spring holding a progressive stamping die together, from the torsion spring that shrank onto its own mandrel and seized, to the disc spring stack preloading a pressure-vessel flange. The physics is identical. The numbers are not, and this guide is about the numbers.

The guide runs in thirteen sections:

1. The Physics of a Spring: Geometry, Rate, and the Torsion Machine

1.1 One equation runs the industry

A helical compression spring is a length of wire wound into a helix. Push on the ends and the helix deflects; the deflection converts a force into a twist of the wire about its own axis. That is the central insight, and it is worth stating precisely: a compression spring is not a bending beam — it is a torsion bar with an identity problem. The axial force F acting at the mean coil radius D/2 applies a torque T = F \cdot D/2 to every cross-section of the wire, and the wire responds by twisting. The coil's spring rate falls out of torsion-bar mechanics:

where G is the shear modulus, d the wire diameter, D the mean coil diameter (outside diameter minus one wire diameter), and N_a the number of active coils. Every design conversation about a spring is a conversation about this equation's fourth power and cube. Double the wire diameter and the spring gets 16 times stiffer. Open the coils — raise D — and stiffness falls with the cube. Add turns and stiffness falls linearly: ten active coils are half as stiff as five, with identical wire and diameter.

The deflection under load F is the reciprocal statement, and the load–deflection line runs until the coils touch:

with U the energy stored. That energy equation is why springs exist: a suspension spring absorbing a 5 kN bump through 80 mm of travel stores \tfrac{1}{2} \times 62{,}500 \times 0.08^2 \approx 200 J, roughly the kinetic energy of a 40 kg mass at 11 km/h — and it gives most of it back. Compression springs, extension springs and torsion springs are all this machine in different costumes; the first two convert axial force to wire torsion, the third applies bending directly to the wire.

1.2 The spring index, and why 4 < C < 12

Define the spring index C = D/d. It is the single most useful dimensionless number in spring design because it sets manufacturability, stress correction (§2) and buckling resistance (§4) simultaneously:

1.3 The end-form table: what the ends do to your free length

The ends of a compression spring are not cosmetic. Squared and ground ends remove roughly two coils' worth of deflection; plain ends remove none. The standard table from Shigley's Mechanical Engineering Design (Table 10-1) fixes the arithmetic, and every spring drawing should follow it exactly:

End form · End coils, N_e · Total coils, N_t · Free length, L_0 · Solid length, L_s · Pitch, p

Plain · 0 · N_a · pN_a + d · d(N_t+1) · (L_0 - d)/N_a

Plain and ground · 1 · N_a + 1 · p(N_a+1) · dN_t · L_0/(N_a+1)

Squared or closed · 2 · N_a + 2 · pN_a + 3d · d(N_t+1) · (L_0 - 3d)/N_a

Squared and ground · 2 · N_a + 2 · pN_a + 2d · dN_t · (L_0 - 2d)/N_a

Three engineering notes live in that table. First, squared-and-ground is the default for anything important — a ground end presents a bearing surface that distributes load evenly and eliminates the eccentricity that plain ends introduce; it also lets the spring be treated as fixed at both ends for buckling purposes (§4), which is worth more than the grinding costs. Second, the solid lengths carry a well-known caveat: for squared and ground ends the theoretical L_s = d N_t overstates the true stack, because grinding flattens the end coils to less than full wire cross-section (the classic check is to count wire diameters in an actual solid stack, and the correction is on the order of 0.75 wire diameter). Third, an extension spring has no solid length at all — it has a maximum extended length, and the ends are specified as loops or hooks with their own stress limits (§6).

Series and parallel rules complete the geometry: two identical springs stacked in series (end to end) halve the rate and add the travels; two in parallel (side by side) double the rate at the same travel. These compound: a die with four 40 N/mm springs in its stripper plate has a stripper force characteristic of one 160 N/mm spring — which is exactly how die makers tune ejector force without new pockets.

2. Stress and the Wahl Correction: The 0.615 That Bounds Every Coil

2.1 From torsion to the peak-stress formula

The wire in a loaded compression spring sees two shear effects at once: the torsional shear from the torque T = FD/2, and a direct shear force F spread over the wire cross-section. A naive superposition gives \tau = 8FD/(\pi d^3)\,(1 + 0.5/C) — but it misses the effect that actually matters: the wire is curved, and curvature pushes the peak shear stress to the inner face of the coil. The stress is not distributed across the wire section the way it would be in a straight torsion bar; the inside of the coil carries more. The correction is named after A. M. Wahl:

and the design stress equation becomes

The modern refinement is the Bergsträsser factor K_B = (4C+2)/(4C-3), which runs within about 1.5% of K_w and is the factor Shigley's current editions prefer. For sanity it is useful to decompose: the direct-shear-only factor is K_s = (2C+1)/(2C), and the curvature factor is the rest.

C · K_s (direct shear) · K_w (Wahl, total) · K_i (torsion-spring inner)

5 · 1.1000 · 1.3105 · 1.1750

6 · 1.0833 · 1.2525 · 1.1417

8 · 1.0625 · 1.1840 · 1.1027

10 · 1.0500 · 1.1448 · 1.0806

12 · 1.0417 · 1.1194 · 1.0663

At C = 8 (typical of the worked design in §5), the Wahl correction multiplies the naive stress by 1.184 — the difference between 587 MPa and 695 MPa at the same load. That 18% is not academic: it is the difference between a design that passes the 871 MPa static allowable of a 2.5 mm music-wire spring and one that doesn't. And the curvature penalty grows as coils get tighter: at C = 5 the factor is 1.31, nearly double the direct-shear term.

2.2 What this means in design practice

Three rules follow directly from the table:

  1. Tight coils are stress multipliers. Shrinking D at fixed d to fit a smaller pocket raises the stress three ways — the moment arm shrinks (good), but the curvature factor rises and the index drops into the forming-danger zone (both bad). The net effect is usually negative below C \approx 5.
  2. A spring's published "stress at L₂" is a Wahl-corrected number when it is honest. Vendor catalogs that quote stress as a percentage of tensile (as most do) are quoting K_w-corrected stress — check which factor is used before comparing a catalog part to your calculation. The industry convention in the classic references: "to calculate the stress before set removal, use the K_{W1} correction factor" — i.e. Wahl.
  3. The peak is on the inside of the coil, at the wire surface. Every fatigue crack (§12) starts there. This single fact is why spring manufacturing spends so much effort on wire surface quality, shot peening, and avoiding nicks during coiling — the highest-stress point in the part is its most damaged surface.

3. Materials: The Size-Dependent Strength Law and the 45% Budget

3.1 S_{ut} = A/d^m: strength is not a material constant

Spring wire is cold-drawn, and cold drawing makes thin wire stronger than thick wire of the same chemistry: the drawing strain accumulates more in thin sections. The industry standardizes this with a two-constant power law fit, and Shigley's Table 10-4 is the canonical compilation:

with d in millimetres and A in MPa·mmᵐ. The consequence is dramatic at the ends of each wire's size range. For ASTM A228 music wire (A = 2{,}211, m = 0.145): a 0.5 mm wire reaches 2,445 MPa minimum tensile, a 1 mm wire 2,211 MPa, a 2 mm wire 2,000 MPa, and a 6.5 mm wire only 1,685 MPa — the same steel, 31% weaker, because it was drawn fat. Any spring calculator that quotes one tensile strength per material is over-rating thick wire and under-rating thin wire, and the error runs in the unsafe direction.

The full working table, assembled from the standard references (Shigley 10-4/10-5 with the classic Associated Spring moduli noted where sources diverge):

Wire · ASTM spec · Size range, mm · A, MPa·mmᵐ · m · E, GPa · G, GPa · Max service temp

Music wire (piano) · A228 · 0.10–6.5 · 2,211 · 0.145 · 203.4 · 79.3–81 · 120 °C

Hard-drawn · A227 · 0.7–12.7 · 1,783 · 0.190 · 196.5 · 79.3 · 120 °C

Oil-tempered · A229 · 0.5–12.7 · 1,855 · 0.187 · 196.5 · 77.2 · 150–180 °C

Chrome-vanadium · A231/A232 · 0.8–11.1 · 2,005 · 0.168 · 203.4 · 77.2 · 220 °C

Chrome-silicon · A401 · 1.6–9.5 · 1,974 · 0.108 · 203.4 · 77.2 · 250 °C

302 stainless · A313 · 0.3–10 · 1,867–2,911 (3 bands) · 0.146–0.478 · 193 · 69 · 260 °C

17-7 PH stainless · A313 (631) · 0.5–6.4 · ~2,000 · — · 196 · 75 · 315 °C

A286 (high temp) · A286 · 0.5–8 · — · — · 200 · 78 · ~425–510 °C

Inconel X-750 · — · 0.5–9.5 · — · — · 214 · 82 · ~550–590 °C

Phosphor bronze · B159 · 0.1–7.5 · 913–1,000 (3 bands) · 0–0.064 · 103.4 · 41.4 · 100 °C

Read across that table and three selection logs emerge. Music wire is the strongest wire made below 6.5 mm, with the smallest permitted surface defects of any commercial grade — it is the default for everything from 3D-printer bed springs to precision instrument parts. Chrome-silicon (A401) gives up little strength in the 2–6 mm band and buys service temperature to 250 °C plus the fatigue pedigree of automotive valve-spring wire — it is what serious suspension and engine springs are made of. Austenitic stainless (302/304) trades 13% of stiffness (G = 69 vs 79.3 GPa) and a fifth of the allowable stress for corrosion resistance, and it cannot be strengthened by heat treatment. Note also a genuine industry trap flagged by the wire calculator references: A231 and A232 are chrome-VANADIUM specifications; A401 is chrome-SILICON — at least one manufacturer's catalog has them crossed.

Moduli footnote: references differ by 1–3% on moduli (music wire G appears as 79.3 GPa in Associated Spring-era tables and 81.0 GPa in Shigley's; 302 stainless as 69.0 or 69.6). All worked examples in this guide use the classic set — music wire G = 79{,}300 MPa — and readers recomputing with 81 GPa should expect about 2% higher rates.

3.2 From tensile to design stress: the percentage table

Spring wire is tested for tensile strength, but the spring lives in shear and torsion — so the design process converts S_{ut} into an allowable torsional stress, as a percentage of tensile, depending on material and duty. The percentages are the single most useful numbers in spring practice:

Material family · Static, as-wound (set not removed) · Static, preset (set removed) · Fatigue design

Music wire (A228), hard-drawn (A227) · 45% · 65% · ≈25–40% (peened) †

Oil-tempered (A229), Cr-V (A231/32), Cr-Si (A401) · 50% · 65% · ≈30–42% (peened) †

Austenitic stainless (302/304/17-7PH), non-ferrous · 35% · 55% · ≈20–35% (peened) †

Torsion springs — the wire in bending · ~75% · — · —

Extension spring hooks (torsion in a tight bend) · ~40% · — · —

The table's structure explains three industries. A 45% budget at d = 2.5 mm music wire means an allowable of 0.45 × 1,936 = 871 MPa, and a spring designed to sit just under it is exactly what a stock catalog sells as "static service". A 65% budget after presetting means 1,258 MPa — a 44% increase in allowable stress from a manufacturing step that costs a few rupees, which is why every automotive valve spring (and every serious die spring) is preset, peened and operated on the 65% curve. And the bending budget for torsion springs is higher than the torsional one — about 75% versus 45% — because bending strength exceeds shear strength, so a torsion spring is allowed more of its material's tensile than a compression spring of the same wire. This asymmetry recurs in §7 with real consequences.

One caution the measured data adds: these percentages are Shigley/Joerres-class guidance, and vendor references publish bands rather than single points (extension-spring body allowables are quoted anywhere from 30% to 45%, with the lower end for fatigue duty and the upper end for static). The fatigue column carries its own warning: serious fatigue design works from rating tables at defined stress ratios (or criteria like Sines–Zimmerli, §12), not from a single percentage of tensile — so treat the † ranges as indicative design bands, not as allowables to cite on a drawing. Where references disagree, this guide prints the conservative reading in the table above and says so. The one thing no serious source disagrees on: do not use a single universal percentage of tensile for every spring. Material, diameter, service and manufacturing steps all move the number.

3.3 Set removal, heat setting, and the residual-stress economy

The 45% → 65% jump deserves its own paragraph because it is the strangest deal in manufacturing. When a spring is wound, the coiling operation leaves residual stresses in the wire; when the finished spring is deliberately compressed beyond its elastic range — scragged or preset — the most heavily loaded fibers yield in a controlled place, at the factory, and on release the spring carries a residual stress field that opposes the service load: the inner surface of the coil ends up with compressive residual stress exactly where the peak stress wants to be (§2). The material is not stronger; the spring is pre-loaded against itself.

The design window for presetting is narrow and specified: a spring suited to presetting is designed so that its stress at solid height lands between 1.1 and 1.3 times the torsional yield strength of the wire (below 1.1×, the set is hard to control and the free length wanders; above 1.3×, the spring distorts rather than sets cleanly). And the standard rule for how much the operation removes: "springs to be preset should be designed so that 10 to 30 percent of the initial free length is removed during the operation." A related process, heat setting (presetting done at elevated temperature), buys something different: improved stress-relaxation resistance at service temperature — the property that decides whether your spring still delivers its load after ten years at 80 °C, discussed with the numbers in §12.

That completes the material kit: a size-dependent tensile law, a percentage budget, and two manufacturing operations that move the budget. Everything else in this guide is geometry and consequences.

4. The Four Constraints That Kill Springs Before Stress Does

A spring can pass every stress check and still fail in service — set flat on its first assembly, buckle sideways at half load, or shake itself apart at a frequency nobody computed. Each of those has its own number.

4.1 Clash allowance: keep 15% between working load and solid

If the working load ever reaches the load that closes the coils, the spring either sets (loses free length) or — at high cycle counts — hammers its own coils. The design rule, stated in the standard texts as a "robust linearity" requirement and visible in every good design example:

where F_s is the load at solid height. The Institute of Spring Technology's own worked design sheet uses the same practice from the other direction — it defines the design deflection to solid as working deflection + 15%. Concretely: if your mechanism needs 180 N, the spring should reach solid at 207 N or more. The check costs nothing at design time and prevents the most common field failure of catalog-picked springs.

There is a second check hiding in the same corner: the stress at solid for a spring that will ever be pressed flat (assembly, installation press, or the clash allowance itself being consumed). If that stress exceeds the static allowable of §3, the spring takes a permanent set the first time it is compressed to solid — this is exactly how "the spring was fine on the bench and 3 mm shorter after installation" happens. The exception is a spring designed for presetting, whose solid stress deliberately lands between 1.1× and 1.3× the torsional yield strength (§3.3).

4.2 Buckling: αL₀/D < 2.63, or the spring goes sideways

Compression springs are columns, and columns buckle. The governing parameter for steel springs, in the form used by the standard machine-design references (Samónov/Wahl/Haringx lineage, as reproduced in the engineering-data literature), is an effective slenderness ratio:

where \alpha is an end-condition constant that mirrors column theory exactly:

End condition · \alpha

Both ends on flat parallel surfaces (fixed) — squared and ground · 0.5

One end fixed, other pivoted · 0.707

Both ends pivoted (hinged) · 1.0

One end clamped, other end free · 2.0

For the everyday case — squared-and-ground ends compressed between rigid plates, which counts as fixed-fixed — the criterion reduces to the memorable form:

In words: a squared-and-ground spring whose free length exceeds 5.26 mean diameters is in buckling territory unless it is guided. And guidance is exactly the fix: buckling is defeated by a guide rod inside the coils, a guide sleeve outside them, or by redesign — a shorter, fatter spring (raise D, cut L_0) to pull L_0/D down. The fuller Wahl/Haringx treatment adds nuance: the critical deflection depends on how far the spring is compressed (stability boundaries move with the deflection ratio \delta/L_0, and the criterion above is the absolute stability condition), but the design-safe move is to stay inside 5.26·D for grounded ends or provide guidance. The worked example in §5 sits at L_0/D = 4.5 (λ = 2.24) — inside the line, but a free length 15 mm longer would cross it.

4.3 Surge: the spring loses a race against its own coils

Every spring is a distributed mass–spring system, and like a Slinky it has natural frequencies at which wave motion travels up and down the coils. Drive it near one and the coils surge: stress spikes far above the static value, rate goes non-linear, and fatigue life collapses. The fundamental surge frequency is set by exactly the quantities that set the rate:

with \rho the wire density. The design rule from the spring-technology references: keep the first natural frequency at least 15–20× above the highest forcing frequency the spring will see. Worked: the §5 spring (d = 2.5 mm, D = 20 mm, Na = 12) has f_n \approx 179 \times 0.0025/(0.02^2 \times 12) \approx 93 Hz, a 19× margin against a 5 Hz actuator — comfortable. (Be aware that end-condition conventions in the literature shift this result by up to a factor of 2 — some derivations give ~186 Hz for the same spring — so check critical cases against your boundary conditions, not a lookup number.) Where the margin cannot be found — cam-driven valve springs at thousands of rpm are the classic case — the fixes are a stiffer design with fewer active coils, damping from nesting or friction, and in engines the familiar countermeasure: nested dual springs, whose two different natural frequencies can't both be excited by the same cam harmonic. Valve springs live or die on this calculation; a surge resonance is a fatigue-fracture delivery mechanism, and the failure photo in Shigley's chapter — a valve spring broken clean along the 45° line of maximum principal stress — is the canonical image of what running a spring at its natural frequency buys you.

4.4 Presetting, revisited as a constraint

Two sentences from the manufacturing side of §3.3 become design constraints: a spring whose stress at solid exceeds its as-wound allowable must be preset (or it sets in service), and a spring intended for presetting must be designed with solid stress in the 1.1–1.3×Ssy window. Presetting is not a repair for a wrong design; it is a design decision that participates in the stress budget from the first sketch.

5. Compression Springs, Designed End to End

Time to run the whole machine. The requirements, borrowed from a realistic actuator-retrofit problem: the mechanism needs 120 N at working position 1 (60 mm) and 180 N at working position 2 (45 mm), room for a spring of ~20 mm mean diameter, a 5 Hz cycling rate, and the environment is a normal machine-shop atmosphere, under 60 °C.

Step 1 — Rate from the two load points

Two points fix the line: k = \dfrac{F_2 - F_1}{L_1 - L_2} = \dfrac{180 - 120}{60 - 45} = 4.0 N/mm.

Step 2 — Wire and diameter from the stress budget

Try music wire A228, d = 2.5 mm, and take the traditionally-proven index C = 8, so D = 20.0 mm (inside the ~20 mm budget).

Safety factor 871/695 = 1.25 — acceptable for static service, and note the Wahl correction did the deciding: the naive 8FD/\pi d^3 stress is 587 MPa, 18% lower. (Had the requirement been 190 N, the as-wound design would fail and the choice would be explicit: bigger wire, or preset and run on the 0.65 curve.)

Step 3 — Active coils

Reflow for the integer coil count: k_{actual} = 79{,}300 \times 39.06 / (8 \times 8{,}000 \times 12) = \mathbf{4.03} N/mm. Every subsequent number uses 4.03, not 4.0 — this is what separates a paper design from a buildable one.

Step 4 — Free length and the load-point check

Step 5 — The four constraint checks

Check · Computation · Result

Clash margin · L_s = 14 \times 2.5 = 35 mm; deflection to solid = 89.6 − 35 = 54.6 mm; F_s = 4.03 \times 54.6 = 220 N · 220 = 1.22 × 180 → 22% margin ≥ 15% ✓

Set at assembly · τ at solid = 695 × 220/180 = 850 MPa vs 871 MPa allowable · No set — but only by 21 MPa (2.4%); a production house would preset for comfort

Buckling · λ = 0.5 × 89.6/20 = 2.24 vs limit 2.63 (L₀ < 105.2 mm) · Stable, unguided ✓

Surge · f_n \approx 179 \times 2.5/(400 \times 12) \approx 93 Hz vs 5 Hz forcing · 19× margin ✓

Step 6 — The buildable geometry

Pitch p = (L_0 - 2d)/N_a = (89.6 - 5)/12 = \mathbf{7.05} mm; wire length ≈ π·D·N_t ≈ π × 20 × 14 ≈ 880 mm of 2.5 mm music wire; end form squared and ground; unguided installation acceptable per Step 5; suggest a finish (zinc plate or powder coat) if the shop air is damp.

Step 7 — The drawing numbers, as the shop wants them

Music wire ASTM A228, d = 2.5 mm — squared & ground — L₀ = 89.6 mm — N_t = 14 — rate 4.03 N/mm — load 120 N at 59.8 mm / 180 N at 45.0 mm — stress at 180 N: 695 MPa — solid height 35 mm — wind right-hand — tolerance class EN 15800 class 2 (load ±10%).

Seven steps, one spring, every number traceable. The single most common mistake this walkthrough prevents: picking a stock catalog spring by diameter and free length and then discovering its rate puts the working loads somewhere else entirely. Stock springs are sold by rate, load and length — design in that order.

6. Extension Springs: Initial Tension, Hooks, and the 80% Rule

6.1 The coil that fights back

An extension spring is wound tight — the coils press against each other under a residual initial tension F_i built in at coiling, and the load law therefore has an intercept:

The initial tension is real force that must be overcome before the coils separate, and it is invaluable: it keeps a spring quiet, keeps a tensioner taut at zero load, and cannot be produced by any other means at the price. But it is also the hardest property to control — vendor references put its standard tolerance at ±10–15% (against ±5–10% for compression load), because it depends on coiling tension, wire surface friction and the stress-relief schedule. And the stress-relief note cuts further: the usual low-temperature anneal after coiling reduces initial tension — references put the correction at 15–25% for stainless and 20–35% for piano/hard-drawn wire — so a spring that needs its initial tension must either skip the anneal or be designed with the loss included.

How much initial tension is achievable? Two classic bounds, from different traditions, converge:

Both formulations say the same thing: initial tension is an index-dependent budget, and a tight coiled spring (low C) can hold many times more of it than a slender one.

6.2 Hooks: the stress concentration you didn't design

The ends of an extension spring are formed into loops, hooks or crossovers, and those bends are the part's weak link — catalogs publish hook-limited maximum loads and maximum extended lengths precisely because the body would happily take more. The static allowable in a hook (wire in torsion inside a tight bend) is roughly 40% of tensile — lower than the body's 45–50% — before the stress concentration of the bend itself is counted. Practical rules: prefer machine loops (full, closed) for fatigue; cross-over center hooks save length but concentrate stress; and never use an extension spring's hook as a structural connection if the load cycles — that is what the body's 10-million-cycle rating is for, and it does not extend to the hooks.

6.3 The 80% rule and the worked example

Extension springs have no solid height, so their absolution limit is a maximum extended length, and the classic design practice (from the stock catalogs) is to never use more than 80% of the deflective capability — keep 20% in reserve for overload and fatigue.

Worked example: a door-tensioner spring needs a pre-tension around 10–15 N and a working load of about 55 N at 30 mm extension.

7. Torsion Springs: Bending, Shrinkage, and the Mandrel That Seizes

7.1 The mechanics flip to bending

A helical torsion spring stores energy by winding up — the legs rotate, the coils tighten, and the wire bends about its own axis rather than twisting. Two consequences follow immediately: the relevant stiffness is E (not G), and the relevant strength is the bending allowable, which is the higher of the two (≈75% of tensile vs 45–50% in torsion). The rate comes out of pure bending theory:

where 3,667 = 64 × 180/π is the beam-theory constant of the derivation. Catalogue practice rounds this to a rate quoted per revolution, k' = Ed^4/(10.8\,D\,N) — the 10.8 (against the theoretical 10.19) carrying the strain allowance for curvature and coil-to-coil friction, making real springs about 6% softer than beam theory predicts. The stress carries the inner-surface correction K_i (table, §2.1): \sigma = K_i \cdot 32M/(\pi d^3).

7.2 The shrink law: ΔID = θD/360N

Here is the fact that writes off more torsion-spring installs than any stress error: as a torsion spring winds up, its inside diameter shrinks — and it shrinks onto whatever is inside it. The relationship is linear, elegant, and universal:

i.e. the mean diameter loses D/N of diameter for every full revolution of leg deflection. Equivalently, the loaded inside diameter is ID_{loaded} \approx ID_{free} \cdot N/(N + \text{REVs}). The industry's illustration is worth memorizing: a 10-coil spring rotated a quarter turn sits at 10/10.25 ≈ 97% of its free diameter — and through a full turn, at ≈91%; a short 4-coil spring at one full revolution lands at 80% of free ID — a nominal 20 mm bore closing to 16 mm. On a 20 mm spring that is nearly 2 mm of lost clearance, "more than most designers leave in the first place." And the body also grows axially by about one wire diameter per full revolution of wind-up — the growth that makes a 3 mm wire spring 0.75 mm longer at quarter turn and eats pocket end-clearance.

Two rules close the gap: specify the inside diameter at maximum working deflection on the drawing (not the free value), and size the mandrel at ≈90% of the loaded inside diameter — supplier guidance that already carries the margin for free-ID tolerance, wire tolerance, winding direction and alignment error. Also non-negotiable: the spring must be loaded in the direction that winds it tighter — never unwinds it — both because the coiling residual stresses then work in your favor and because a spring deflected the wrong way opens its coils and misbehaves immediately.

7.3 Worked example, and the leg warning

A latch spring: d = 2.0 mm, D = 15.0 mm, N = 8 body coils, music wire, working through 90° of rotation.

8. Disc Springs: Cones That Beat Coils for Force Density

8.1 The geometry and why the curve bends

A disc spring (Belleville washer) is a conical washer that flattens under axial load. Where a coil spring's load–deflection line is straight and its force density is set by how much wire you can pack, a disc spring's curve is shaped by geometry — and that is its superpower. The standard formula (DIN 2093 / DIN EN 16983 lineage) ties force to deflection through the cone height h_0 and thickness t:

with D_e the outside diameter, \mu Poisson's ratio, and K_1 a constant of the diameter ratio alone, C = D_e/D_i:

The bracket is the source of the famous curve family. At h_0/t \approx 0.4 the characteristic is nearly a straight line. As the ratio grows the curve gains a knee, and near h_0/t \approx 1.3–1.5 the region around 75% deflection is nearly flat — load stays almost constant across a range of travel — which is exactly what wear compensation and thermal take-up need. That is not a defect; it is the most useful non-linearity in machine design.

8.2 The three series, the working window, and the numbers

DIN 2093 sorts discs into three series by the ratio that shapes the curve (with DIN EN 16983 as the current European standard carrying the same system):

Series · h_0/t (range) · D_e/t typical · Character · Typical duty

A · ≤ 0.4 · ≈ 18 · Stiff, near-linear · Bolt preload, clutches, clamps

B · 0.4 – 0.75 · ≈ 28 · Compromise · General machine elements

C · 0.75 – 1.3 · ≈ 40 · Soft, travel-heavy · Thermal expansion take-up, long travel

Three numbers govern use. The working window: stay between s = 0.15\,h_0 (typical preload) and s = 0.75\,h_0, where the measured curve hugs theory; beyond 0.75·h₀ the stress at the bore approaches the fatigue limit and discs crack. The catalogue convention: disc springs are rated by the force at 75% deflection, F(0.75\,h_0) — and for a series-A disc that works out to approximately 0.77 × the flattening load (the closed-form ratio [0.3/0.4]\times1.025 \approx 0.77 for h_0/t = 0.4). The standard sizes: core DIN sizes run 8 mm to 250 mm outside diameter, in three thickness groups (t < 1.25 mm; 1.25–6 mm; > 6–14 mm).

Concrete: the smallest standard sizes flatten at 210 N (8 mm) and 320 N (10 mm); the mid-range that machine designers touch most often: a 25 × 12.2 × 1.5 mm series-A disc (h₀ = 0.55 mm) flattens at 2,910 N, so at the 0.75·h₀ working point (0.41 mm travel) it delivers ≈ 2,240 N on a curve that is nearly flat — force tolerance at that point ±2% even in standard grade.

Stacking is where discs win: in parallel (all discs same orientation) load multiplies — 8 stacked 25 mm discs deliver 8 × 2,240 ≈ 17.9 kN at 0.41 mm; in series (alternating) travel multiplies — the same 8 discs give 2,240 N over 8 × 0.41 ≈ 3.3 mm. Mixed stacks (n parallel × m series) interpolate freely, and every stack adds friction hysteresis between disc edges — which damps vibration as a byproduct. That is why disc stacks appear as: bolt-preload stacks on pressure-vessel flanges (where a 20 kN preload must survive thermal cycling), spindle bearing preload columns, clutch and brake release stacks, and the suspension of choice wherever force density per millimetre of height beats linearity.

Materials run with the standard's default: 50CrV4 / 51CrV4 spring steel quenched and tempered to 42–48 HRC, with 30WCrV6-3 for service above 250 °C and 17-7PH stainless for corrosive duty; dynamic stacks are shot-peened after hardening, for the same reason coil springs are (§10).

9. Die Springs: ISO 10243, the Four Colours, and Their Deflection Budgets

9.1 The standard and its rectangular secret

Nowhere is spring engineering more standardized — and more colour-coded — than in press tooling. ISO 10243:2019 ("Tools for pressing — Compression springs with rectangular section — Housing dimensions and colour coding") governs the compression springs that strippers, ejectors, pilots and blank holders ride on. Two engineering ideas hide in that title. First, rectangular section: die springs are coiled from profiled wire with a rectangular cross-section, which packs more material — and therefore more force — into the same bore than round wire ever could. Second — and this is the elegant part — the standard does not just fix colours; it fixes housing dimensions (D_h bore, D_d rod) so that two springs of the same size but different colour are mechanically interchangeable in the same pocket. Changing a die's stripping force becomes a spring swap, not a die modification.

The material of record for ISO 10243 springs is 52SiCrNi5 (DIN 1.7117) chromium-nickel-silicon steel. The four ISO colours cover bore diameters of 10–63 mm and free lengths of 25–305 mm, with forces for the core four series running from about 100 N to 7,000 N; extended grade families (lilac/gold/black and equivalents) stretch the same housings to 45,000 N (45 kN) and below 100 N. JIS B 5012 exists as the Japanese twin of the same idea — but because colour schemes differ between standards and manufacturers' own extended ranges, the professional rule is: read the load class off the datasheet, not off the paint.

9.2 The colour classes and their real meaning

The colour is a deflection budget label, not a style choice. Each colour class is limited in how far its coils may travel as a percentage of free length — and the stiffer the class, the shorter the allowed travel, because a stiffer spring reaches higher stress at any given deflection:

Colour (ISO 10243) · Load class · Max deflection, % of L_0 · Long-life design target · Typical duty

Green · Light · 40% · ≈ 25–30% · Long-stroke pilots, strippers, lifter returns

Blue · Medium · 37.5% · ≈ 20–25% · General stripping, blank holders — the workhorse

Red · Heavy · 30% · ≈ 15–20% · Heavy blanking clamps, pad return

Yellow · Extra-heavy · 25% · ≈ 10–15% · Maximum force in minimum pocket, short stroke

Two framing notes. First, deflection percentages trade directly against life: vendor-rated fatigue at max deflection runs on the order of 1 million cycles for light classes and down toward 150,000–300,000 cycles for the extra-heavies — which is why a die hitting 60 strokes per minute (≈ 30 million cycles per year on three shifts) must be designed at the long-life column, not the max column. Second, the over-deflection penalty is brutal and vendor guidance is unanimous: exceed a class's rated deflection even by ~5% and service life collapses — vendor testing puts it at up to an order of magnitude of lost life, because the stress at the coil's inner surface is rising steeply as the spring approaches its class limit. The safety practice recommended by die-spring suppliers: select one class stiffer than the calculated minimum and then run it below its rated deflection, giving the recommended ~20% margin.

9.3 Picking one: the worked selection

Requirement: a stripper plate needs 1,400 N of total force at the working position, with 10 mm of working travel, pockets 50 mm deep (free length ≈ 50 mm), and a die running at moderate cycle counts.

  1. Deflection budget: 10 mm on a 50 mm spring is 20% of L_0. Green and blue can both tolerate it (40%, 37.5%), red is marginal at long-life (15–20%), yellow is out (10–15%). The long-life target points to blue (medium).
  2. Rate required: if four pockets are usable, each spring must make 1,400/4 = 350 N at 10 mm → rate ≈ 35 N/mm for a 25 mm OD × 50 mm blue — squarely inside the typical 30–45 N/mm band for that size and class.
  3. Verify against datasheet: load at 10 mm, preload length, solid clearance, rated deflection — the four numbers every die spring datasheet prints (nobody should select from this table's representative numbers; they select from the manufacturer's table for the exact size).
  4. Housing check: pockets per the ISO 10243 housing table for the bore — and remember the interchangeability payoff: if a die revision later needs 1,600 N instead of 1,400 N, the same pockets accept red springs; only the stroke must be re-checked.

That sequence — budget → rate → datasheet → housing — is the entire die-spring discipline, and skipping step 1 is the leading cause of premature die-spring failure.

9.4 Two adjacent members of the family

Gas springs replace the wire with compressed nitrogen: force is pressure acting on rod area, F = p \cdot A_{rod} — a 12 mm rod charged at 30 bar (= 3 MPa) carries ≈ 340 N, and tooling gas springs scale from ~100 N to many tonnes with force rising modestly over the stroke as the gas compresses and heats. Their virtues are force density and a built-in damping behavior in the variants that meter oil through the piston; their Achilles heel is temperature and seal creep — force drifts with temperature (pressure is proportional to absolute temperature), and a cheaper gas spring slowly loses its charge, taking its holding force with it. They dominate press-tool counterbalance and lift-assist duties where a coil spring would need impossible height.

Wire forms are the family's non-spring relatives: CNC wire formers bend round or flat wire into clips, guards, racks, hooks and retainers by geometry rather than by elasticity. No heat treatment follows for mild-steel forms — the part's function is shape, not stored energy — but the same tooling (CNC formers, 2D/3D bending, in the same 0.5–16 mm wire range) makes them the natural shelf-neighbor to springs in any wire-working shop, and they can often be prototyped and run in the same order as the springs they clip onto.

10. Manufacturing: The Five Steps That Decide Whether the Catalogue Curve Is True

A spring drawing specifies a shape, but a spring's performance is decided by processing. The chain runs: wire → coil → stress-relieve → ends → preset → peen → coat → test, and every station moves a number that design equations assume.

1 — Wire. Spring wire arrives drawn to specification — the European system splits it into EN 10270-1 (patented cold-drawn, ascending tensile classes from SL to SH), EN 10270-2 (oil-hardened and tempered, the automotive-grade family) and EN 10270-3 (stainless). The class determines the S_{ut} your §3 table predicted, so a material substitution at the wire mill is a design change, not a procurement event.

2 — Coiling. CNC coilers win the volume tiers: cold coiling (wire up to roughly 10–12 mm, where wire can be formed in its cold-drawn state) holds load tolerances to ±3% with 100% load testing; above ~10 mm, hot coiling (10–80 mm range) forms the wire at temperature and trades tolerance (±5–10%) for feasibility — hot-wound springs decarburize slightly and scale, so hot coiling is a different, wider tolerance world by nature.

3 — Stress relief. Cold-wound springs carry forming stresses that quietly consume part of the elastic budget; a low-temperature anneal returns them. The published practical window clusters around 200–230 °C for 10–30 minutes for cold-drawn carbon wires, with the standing rule to stay well below the wire's tempering temperature so you relieve forming stress without softening the material. This step is also where a tension spring loses some initial tension (§6.1) — one of the invisible trade-offs of the schedule.

4 — Ends. Compression springs get their ends squared and ground — the standard says bearing surfaces must be square enough to seat flat, and while commercial practice holds a few degrees of squareness, grinding is what makes a spring count as fixed under buckling rules (§4.2) and eliminates the side-loads that eat guide bores. The quality ladder vendors print for precision parts runs: squareness ~1° / 2° / 3° for precision / commercial / general classes.

5 — Presetting, peening, coating. Presetting (scragging) presses the spring beyond its elastic range so it "learns" its final length and banks the residual stress field that lifts the allowable to 65% (§3.3) — with the design-side rule that 10–30% of the initial free length is removed by the operation, and the stress-at-solid window kept between 1.1× and 1.3× Ssy. Shot peening then hammers the wire surface — precisely the inner coil face where peak stress and fatigue cracks live — into compressive residual stress: the technique "improves fatigue life and minimizes the harmful effect of surface defects," in the words of the Associated Spring design manual, and it is why the industry states that dynamic performance data is only valid for springs with a shot-peened, hardened surface. Finally, coatings: powder coat for die springs and general corrosion color-coding; zinc plating, black oxide or phosphate for carbon steels — with a real materials-safety footnote for high-strength wire: acid pickling and electroplating can introduce hydrogen that embrittles very strong steel, so plated high-strength springs are specified with a post-plating bake, or plated before winding, or replaced by stainless entirely. Stainless springs, in turn, deserve passivation to restore the oxide layer after coiling scratches.

Testing closes the loop. Springs are load-tested at specified heights — and the sampling standard is part of the trade: production is inspected per DIN ISO 2859-1 (AQL sampling, test level II) for standard classes, while precision orders move to 100% load testing, which vendors price at an 8–12% premium over standard sampling — the same premium that buys the ±3% capability quoted earlier.

11. Tolerances: What You Buy Between the Classes

Springs are sold to tolerance standards, and the standards are graded — so the gap between two quotes for "the same spring" is usually the tolerance class hiding inside them. For cold-coiled compression springs, the current European reference is EN 15800 (successor to the legacy DIN 2095 for round-wire cold-coiled springs; DIN 2096 covers hot-coiled), built on three classes and tolerances that scale with geometry. Representative values for a typical small compression spring (wire 1–3 mm, index 5–10):

Parameter · Class 1 (precision) · Class 2 (standard) · Class 3 (commercial)

Load at specified length · ±5% · ±10% · ±15%

Free length · ±1–2% · ±2–4% · ±4–6%

Outside diameter · ±0.5–1% · ±1–2% · ±1.5–3%

Squareness · ≈1° · ≈2° · ≈3°

Relative cost · +15–25% · baseline · −10–20%

Read the table like a shop drawing. Load at working height is the number that matters — a spring is a force-delivery device, and free length, OD and squareness are all proxies. Free length is the loosest of the three by design: it is a by-product of coiling and end grinding, which is why a ±10% load tolerance coexists with a ±2–4% length tolerance, and why buying "to length" is a rookie move. Class 2 at ±10% on load is the economic default for machinery; class 1 exists for calibrated instruments and latched mechanisms where a 5% force shift is a functional change. And the capability ladder doesn't stop at the standard: CNC coiling plus 100% load testing can hold ±3% on load for wire above ~1 mm, at a premium of roughly 8–12% over standard sampling — the precision that fuel injectors and medical mechanisms pay for. Quoted rate tolerances, it should be said, vary wildly between sources and geometries (±2% to ±20% depending on class) — another reason the load-at-height number, not the rate, belongs on the drawing.

The other spring types have their own contract numbers: extension springs are specified on initial tension (±10–15% — the hardest property in the trade) and load at extension; torsion springs on torque at angle (±5–10%) and leg angular position (±2–3°); die springs per ISO 10243's per-size tables; disc springs on force at the 0.75·h₀ point, where precision-graded stacks reach ±2%. Test sampling, finally, is part of the tolerance: standard production runs inspected per DIN ISO 2859-1 AQL sampling (level II), precision orders at 100%. Lead times follow the class: general/class-3 springs out the door in ~5 working days, commercial class 2 in ~7, precision class 1 in ~12.

12. The Failure Catalogue: Nine Ways a Spring Dies

1 — Fatigue. The dominant failure. The crack starts at the wire surface on the coil's inner face, where §2's Wahl-corrected peak stress lives and where any drawing scratch, seam or pit acts as a stress raiser — which is why surface quality ("valve spring quality") is a line item in the wire spec, and why the fractured valve spring breaks cleanly along a 45° line: pure torsional loading, maximum principal stress, no warning. Fatigue design is done against dedicated allowable tables (stress-ratio-zero values published for common wires, and the Sines–Zimmerli criterion in the standard design texts), not against the static percentages — and the two manufacturing steps that buy the most fatigue life are the cheapest ones: presetting (residual stress in the right direction) and shot peening (compressive skin at the crack-initiation site). A spring that must survive 10⁷–10⁸ cycles earns both, plus a stress budget near the 27–40%-of-tensile band rather than the static 45–50%.

2 — Set and relaxation. The spring comes back shorter, or holds less load at the same height, without any crack. Cold-drawn wire relaxes under sustained stress and temperature — the Associated Spring design manual notes that highly stressed music wire "exhibits appreciable stress relaxation after only 100 hours at temperatures as low as 100 °C," which is uncomfortably close to machinery-cabinet temperatures. The defenses, in order of cost: keep the service temperature under the wire's limit (music wire 120 °C; chrome-silicon 250 °C; see §3.1's table), heat-set the spring (preset at temperature — the reference texts note it "significantly improves stress-relaxation resistance at moderate temperatures" and is often cheaper than jumping to a premium alloy), and design with the loss accounted (the "creep-limited" design case). A spring that sets in service is telling you the design's stress × temperature × time budget was exceeded — the failure is arithmetic, not bad luck.

3 — Corrosion and hydrogen embrittlement. Rust thins the wire and pits the surface — and every pit is a fatigue starter, so a mildly rusty spring can fail at half the fatigue life of a clean one. Stainless grades trade strength for immunity (302 for general, 316 where chlorides threaten — austenitic stainless is susceptible to chloride stress-corrosion cracking, and passivation restores the oxide layer after forming). The sharper-edged hazard: hydrogen embrittlement of high-strength wire — acid pickling and electroplating can charge the steel with hydrogen, and very strong spring steel (above ~1,400–1,800 MPa classes) under sustained tensile stress can then crack without warning. Mitigations are standard: post-plating bake per the plating spec, mechanical rather than acid processes, or peening after plating.

4 — Buckling. §4.2's sideways collapse, and its tell is unmistakable: the spring bows into a banana, then rubs its coils against the guide or wall and wears through its coating within hours. The design answer is the slenderness ratio before the failure is the slenderness ratio: λ = αL₀/D under 2.63, or a guide.

5 — Surge. §4.3's resonance; the tell is a rate curve that goes soft or noisy at speed, and a fatigue fracture with no overload history. The answer is a frequency margin (15–20×), a stiffer spring with fewer active coils, damping, or the industry fix of nested springs.

6 — Wear and fretting. Coil-to-coil contact at solid, grinding against guides, abrasive dust in a shop environment: every one of these mints fine metal debris and polishes stress raisers off the wire. Guides should be hardened and polished; coatings carry a wear function as well as a corrosion one; and a failure-analysis finding "black fretting oxide at the break" usually means the spring was clashing, not overloaded.

7 — Over-extension. The tension-spring specialty: exceed the rated extension and the result is a straightened hook or a permanently longer spring, because the wire yielded — which is what the 80% rule and the hook-limited maximum loads (§6) exist to prevent.

8 — Seizure (torsion). §7.2's shrink law, weaponized: a spring that spun freely on its arbor at rest binds at full wind-up, or the body grows into the housing face. Both are drawing-table mistakes — the free-state ID on the print was never the loaded ID.

9 — Disc spring cracking and stack errors. Discs run past 0.75·h₀ crack at the bore; discs installed with one flipped orientation in a stack produce a wrong-curve spring; and stacks without hardened end washers imprint their edges. All three are caught by a stack drawing with orientation marks — a cheap control for an expensive failure.

The pattern across all nine: the shop can build exactly what the drawing says, but it cannot know the assumptions behind it — the 5 Hz, the 80 °C, the missing guide, the 5% over-deflection. Most spring failures are bought, not built — purchased at the design desk, delivered years later on the plant floor.

13. The Ledger: What Springs Cost, and the Industry Behind Them

13.1 The Indian spring industry

The Indian spring market generated USD 563.7 million in 2024 and is projected to reach USD 861 million by 2030 at a 7.4% CAGR — about 4.8% of the global market, and the fastest-growing country market in Asia-Pacific. (Company investor material sizes the same industry at "about ₹5,000 crore" — the same number through a different exchangerate, which is a healthy sign for both.) The mix is telling: helical springs are the largest single type segment at 46.7% of revenue, with leaf springs and disc springs dividing the rest — meaning the automotive aftermarket and OEM supply chains are the industry's spine.

That spine has a remarkable concentration. In leaf and parabolic springs, Jamna Auto Industries holds roughly 68–70% of the Indian OEM market — with an installed capacity around 240,000–300,000 t/year, it is a top-two global leaf-spring maker by capacity, in a table where the next entries are Rassini (Mexico, ~250,000 t), Hendrickson (Canada/Mexico, ~216,000 t) and NHK (Japan/Thailand, ~150,000 t); Toyo (~20%) and Soni (~5%) trail domestically and the rest of the market is fragmented. And the high end still flows in from outside: India's automotive suspension-spring import shipments grew at roughly 20% CAGR across 2020–2024, led by Germany, the USA, China, South Korea and Japan — a reminder that the hardest applications (high-cycle valve-spring-quality wire, pre-set, peened, certified) remain an import story at the top of the market even as volume production is thoroughly domestic.

13.2 What a spring costs, and why

For custom springs, the piece price is dominated by four levers, in this order of impact:

  1. Material and wire class — music wire is the cheap default while its strength suffices; chrome-silicon and valve-quality grades (EN 10270-2 territory) can multiply wire cost; stainless trades capability for corrosion immunity priced by the alloy.
  2. Tolerance class and testing — class 1 over class 2 adds 15–25%; class 3 subtracts 10–20%; ±3% capability with 100% load testing adds 8–12%; and specifying tight free length while leaving load loose is paying for the wrong precision.
  3. Processing steps — presetting, heat setting, shot peening, and coating each carry a line item; on a high-cycle part they are the cheapest reliability you will ever buy, and on a one-off prototype they are usually skipped.
  4. Setup amortization and lead time — a CNC coiler setup is coiling-machine hours, not material, which is why 100 pieces and 10,000 pieces of the same spring differ in unit price by several fold, and why custom lead times cluster at 5–12 working days across the classes (plus the possibility of buying a "stock equivalent" today for a fraction of the price and a wider tolerance).

The design-side ledger matters more than the price list: the three decisions that actually move total cost are the fatigue margin (a peened, preset spring in a common material outlasts an unpeened one in an exotic material, at lower cost), the tolerance budget (specify class 2 and design for the spread; class 1 is a last resort), and the constraint discipline of §4 (a buckling or surge failure retrofits at 10× the cost of designing the guide or the frequency margin in the first place). And the cheapest line of all remains the first one: a spring specified by rate, load at heights, material spec and test standard — not by "about 20 mm and stiff" — hits the shop as a manufacturable part, not a negotiation.

Frequently Asked Questions

What is spring rate and how is it calculated?

Spring rate is the force per unit of deflection — k = Gd⁴/(8D³Na) — and it is the first number of every design because two load points fix it: a spring that carries 120 N at one position and 180 N at another 15 mm closer has a rate of 4 N/mm. The fourth power on wire diameter makes wire size the coarse adjustment (a 10% wire change moves rate ~46%) and coil count the fine one (rate is inversely proportional to active coils). Real springs are then checked for the fitted rate — for example a design computing Na = 12.1 rounds to 12 active coils and the rate re-flows to 4.03 N/mm — and the working loads move accordingly. Deflection is linear until coils touch: F = kδ, energy = ½kδ².

Music wire, stainless, or chrome-silicon — which spring material should I choose?

Decide in this order: temperature, corrosion, then strength. Music wire (ASTM A228) is the strongest wire below ~6.5 mm (2,445 MPa at 0.5 mm falling to 1,685 MPa at 6.5 mm), the cheapest, and fine up to 120 °C — the default for small mechanical springs. Chrome-silicon (A401) is the fatigue and heat champion: service to 250 °C, the valve-spring pedigree, and marginally lower strength than music wire in most sizes. 302 stainless costs you ~13% stiffness (G = 69 vs 79.3 GPa), a fifth of the allowable stress (0.35 vs 0.45 of tensile), and buys corrosion immunity to 260 °C; upgrade to 316 only where chlorides are actually present. Above 250–300 °C, it's A286 or Inconel X-750 territory, with prices to match. And remember the allowable-stress rule underneath the choice: static service runs at 45–50% of tensile (as-wound) or 65% if the spring is preset; austenitic stainless and non-ferrous wires stay at 35% (55% preset).

My spring lost its free length / its load — what happened?

It set: the material yielded (or crept) and the spring is now shorter, or it relaxes under sustained stress at temperature and delivers less load at the same height. Cold-drawn music wire relaxes measurably after only 100 hours at ~100 °C, so a warm environment plus a hot-built stress budget explains most field cases. Fixes in order: verify the service temperature against the material limit (music wire 120 °C; oil-tempered 150–180 °C; chrome-silicon 250 °C); ensure the stress at solid is below the static allowable so assembly can't set the spring; and for the next order, specify heat setting (presetting at temperature), which the design references credit with significantly improved stress-relaxation resistance at moderate temperatures — often cheaper than moving to a more expensive alloy.

How far can I compress a die spring?

By its colour class's deflection budget, and no further: ISO 10243 die springs allow 40% of free length (green/light), 37.5% (blue/medium), 30% (red/heavy) and 25% (yellow/extra-heavy) at maximum, with long-life figures roughly two-thirds of those (≈25–30%, 20–25%, 15–20%, 10–15%). The stiffness classes trade force for travel: a stiffer spring develops more force in the same pocket but may only use a shorter stroke, and vendor data is blunt about the penalty for over-deflection — around a 5% overshoot can cut fatigue life by up to an order of magnitude to the point that supplier guidance is to pick one class stiffer than the calculated minimum and keep ~20% in reserve. Practical rule: compute the required force and stroke, run the deflection budget first, then pick the class — and read the load class off the datasheet, not the paint.

Why does my torsion spring seize or bind on its shaft?

Because a torsion spring shrinks as it winds up, and the free inside diameter on the drawing is not the diameter it will have at working deflection. The loaded diameter follows ID_loaded ≈ ID_free × N/(N + revolutions) — a 10-coil spring rotates a quarter turn at ~97% of free ID, a full turn at ~91%, and a short 4-coil spring ends its first full turn at only 80% of the nominal bore. A 20 mm bore can close by nearly 2 mm, more clearance than most designs carry. Specify the ID at maximum working deflection on the drawing, size the mandrel at ≈90% of that loaded ID, remember the body also grows axially by about one wire diameter per full revolution, and always wind in the direction that tightens the coils.

What is the 15% "clash allowance" and why does it matter?

It is the margin between the maximum working load and the load that closes the coils to solid: F_solid ≥ 1.15 × F_max, i.e. keep the working load at or below ~87% of the solid load. Without it, transient overloads (assembly, end-stops, thermal growth) press the spring to solid — once is enough to set a plain spring permanently shorter, and repeated solid contact is a fatigue accelerator. The check is part of the standard design procedure, appears as the "maximum working load +15%" in design-sheet worked examples, and costs nothing before the spring is ordered. It also protects the second condition of the same paragraph: that the stress at solid stays under the static allowable so that even an accidental full compression doesn't cost the spring its calibration.

Can I 3D print springs?

Yes, with honest limits. FDM springs in PLA/PETG/TPU print fine and work as light-duty, low-cycle, low-precision devices — demonstrators, compliant mechanisms, light clips — because the geometry math in this guide still applies: rate k = Gd⁴/(8D³Na) with a polymer shear modulus ~60–80× smaller than steel's (≈1.3 GPa for PLA versus 79.3 for music wire) means a printed spring of the same size is dramatically softer, and polymer creep means it will keep losing force under sustained load even at room temperature. What 3D printing does not deliver is a calibrated force element with fatigue ratings, temperature limits and ±10% load tolerance — for any mechanism where the number on the drawing must be the number in the field, the spring is still a piece of steel formed on a coiler, preset and peened. Use printing for the prototype, buy the spring for the product — and consider printed springs as replaceable compliant elements where "give" matters more than "grip."

The Discipline in One Page

A spring is one equation with four consequences. The equation is k = Gd^4/(8D^3N_a) — rate set by wire (fourth power), diameter (cube) and active coils — and the peak stress carries the Wahl correction K_w = (4C-1)/(4C-4) + 0.615/C, which at the workhorse index of 8 adds 18% over the naive coil stress (§1–2). The material is not a constant but a curve — S_{ut} = A/d^m, music wire 2,445 MPa at 0.5 mm and 1,685 MPa at 6.5 mm — and the static allowable starts at 45% of it, rising to 65% the moment the spring is preset (§3). Four constraint numbers bound the design independently of stress: the clash margin (F_solid ≥ 1.15 F_max), buckling (λ = αL₀/D < 2.63; L₀ < 5.26·D for grounded ends), surge (f_n ≈ 179·d/(D²N_a) Hz, kept 15–20× above the forcing frequency), and the presetting window (stress at solid 1.1–1.3× Ssy, 10–30% of free length removed by design) (§4). The same arithmetic dresses up as every family: extension springs add initial tension (10–25% of load, bounded by 0.4–0.8·Sut/C, ±15% tolerance) and hook limits (§6); torsion springs bend instead (M = Ed^4θ/3667DN, bending allowable ≈75% of tensile) and their bores shrink by ΔID = θD/360N — size the mandrel to 90% of the loaded ID (§7); disc springs trade linearity for force density in the A/B/C series, working between 0.15·h₀ and 0.75·h₀, with a 25 mm disc flattening at 2,910 N and stacks multiplying force in parallel or travel in series (§8); die springs encode their deflection budgets in four colours — green 40%, blue 37.5%, red 30%, yellow 25% — in interchangeable ISO 10243 housings, with one-class-up selection and never a 5% over-deflection (§9). Manufacturing decides whether the catalogue curve is true: coiling, 200–230 °C stress relief, squared and ground ends, presetting, shot peening, and load testing at sampling standards — or 100% at a premium for ±3% capability (§10–11). Failures are bought at the design desk and delivered years later: fatigue at the inner coil face, set and relaxation at temperature, corrosion and hydrogen embrittlement, buckling, surge, fretting, over-extension, torsion seizure, disc cracking (§12). And the economy around it all: an Indian spring market of ₹5,000 crore — USD 563.7 million growing 7.4% to USD 861 million by 2030 — 46.7% of it helical, its leaf-spring half concentrated 70% into one company, with the highest grades still imported and the design decisions (fatigue margin first, tolerance class second, material third) costing less than the failures they prevent (§13). The springs that last are not the strongest; they are the ones whose engineer computed the 15%, the 2.63, the 90%, and the 0.615 — before anyone ordered wire.

Previous guides in this series: Industrial Wastewater Treatment & Effluent Treatment Plants · O-Rings, Gaskets & Mechanical Seals · Machine Safety & Guarding · PCB Fabrication & Bare Board Manufacturing · Industrial Ventilation, Fans & Dust Collection · Lean Manufacturing & Production Systems · Belt Conveyors & Bulk Material Handling · Cranes, Hoists & Rigging · Cleanroom Engineering & Contamination Control · Industrial Refrigeration & Cold Chain Systems · Industrial Electrical Power Distribution · Metrology & Dimensional Inspection · Industrial Furnaces, Kilns & Refractories · Structural Steel Design & Fabrication · Pressure Vessels & Storage Tanks · Process Piping & Pipe Fabrication · Corrosion Engineering · Industrial Steam Boilers & Steam Systems · Industrial Valves · Industrial Pumps · Rolling-Element Bearings · Gears & Gearboxes · Fasteners, Threads & Bolted Joints.

[1] Shigley's Mechanical Engineering Design, Chapter 10 "Mechanical Springs" (8th edition chapter text and lecture notes; Budynas–Nisbett) — governing equations and design practice used throughout: rate k = Gd⁴/8D³Na; stress τ = Kw·8FD/πd³ with Kw = (4C−1)/(4C−4) + 0.615/C; Table 10-1 end-treatment table (end coils, Nt, free length, solid length, pitch for plain/plain&ground/squared/squared&ground); preferred index band 4 ≤ C ≤ 12 and active-coil range 3 ≤ Na ≤ 15; "robust linearity" clash requirement ε ≥ 0.15 (Fs ≥ 1.15 Fmax); buckling criterion L0 < 2.63D/α (α = 0.5 squared and ground → L0 < 5.26D); presetting design window (10–30% of initial free length removed; stress at solid between 1.1× and 1.3× torsional yield); Ssy = 0.45 Sut for A228 as-wound; fatigue via Sines–Zimmerli criterion; the valve-spring fatigue-fracture photograph (45° maximum-principal-stress line). [2] MechSimulator "Spring Design Calculator" (Shigley-verified calculation engine, mechsimulator.com) — size-dependent strength table Sut = A/dᵐ for A228 (2,211 / 0.145), A227 (1,783 / 0.190), A229 (1,855 / 0.187), A231/32 (2,005 / 0.168), A401 (1,974 / 0.108), A313 302 in three bands (1,867–2,911), B159 phosphor bronze; moduli table (E/G per material, 193–203.4 / 69–81 GPa); allowable stress 0.45 Sut ferrous, 0.35 stainless/non-ferrous, rising to 0.65/0.55 with set removed; buckling λ limit ≈2.6; surge frequency kept ≈20× forcing rate; worked example dataset cross-checked against Shigley 10-4/10-5. [3] CalcEngines "Spring Wire Strength & Allowable Stress Calculator" (calcengines.com) — allowable-stress reference set: 45% music/hard-drawn in torsion, 50% oil-tempered and alloy, 35% austenitic stainless and non-ferrous, ≈75% in bending (torsion springs), ≈40% extension hooks; wire-size strength gradients verified against Suhm Spring Works minimum-tensile tables; moduli values (music wire G = 79.3 / E = 206.8 GPa; 302 stainless 69 / 193 GPa; phosphor bronze 41.4 / 103.4 GPa); A231/A232 chrome-vanadium vs A401 chrome-silicon naming caution. [4] Associated Spring (Barnes Group) "Spring Design" guide (2025 edition, amft.com) — static allowable torsional stress values 35–50% of tensile without set removal; KW1 correction convention for pre-set stress; stress-relaxation note ("music wire exhibits appreciable stress relaxation after only 100 hours at 100 °C"); heat setting improving relaxation resistance; shot peening "improves fatigue life and minimizes the harmful effect of surface defects"; Tables 5-2/5-3 static and fatigue allowables; load-at-test-height creep criteria. [5] Roy Mech "Helical Spring Natural Frequency / Surge" (roymech.org) — surge frequency derivation fn = (d/4πNaD²)√(G/2ρ) (fundamental) and the 15–20× margin rule. [6] IJERA Vol.9 "Natural Frequency, Ride Frequency and their Influence in Suspension System Design" — independent listing of the same surge formula set and the 15–20× guidance. [7] University of Fairfield ME311 Machine Design Notes 10 — surge equation presentation and spring fatigue context. [8] EngineersEdge "Helical Spring Buckle Stability Calculator" (after Samónov; citing Wahl and Haringx) — effective slenderness λeff = αL0/D with α = 0.5 fixed-fixed, 0.707 fixed-hinged, 1.0 hinged-hinged, 2.0 clamped-free; absolute stability condition reduced for steels to L0 < 2.63D/α, i.e. L0 < 5.26D for squared-and-ground ends. [9] Institute of Spring Technology (IST) Design Data Sheet No. 9 — design-chart method; initial-tension guide (≈25% of max operating load at index 3, ≈10% at index 12); solid-length formulas per end form; "maximum working load + 15%" design practice; BS 1726-2 tolerance referencing; worked design examples. [10] ScienceDirect "Spring Index" topic compilation (after Juvinall & Marshek) — extension initial tension typically 10–25% of maximum design force; recommended initial shear stress between 0.4·Sut/C and 0.8·Sut/C; solid-stress table limiting long-term set to 2%. [11] MISUMI spring technical pages — initial-stress correction values (stainless −15%; low-temperature annealed piano/hard wire −20 to −35%, stainless −15 to −25%); coil-end and active-turn conventions. [12] Lee Spring extension spring catalogue — load law F = S·x + F0; "not extended beyond 80% of deflective capability"; stainless ratings ×0.833 of music wire. [13] Acxess Spring extension design guide — working-load formulas and hook-limited maximum load/travel guidance. [14] Hagens "Torsion springs shrink onto the shaft" — loaded ID = IDfree × N/(N + REVs); 10-coil examples (97% at quarter turn, 91% at full revolution); 4-coil spring at 80% of free ID; body growth ≈1 wire diameter per revolution; drawing checklist; mandrel ≈90% of loaded ID. [15] CalculatorWebTools torsion spring reference (after Century Spring) — rate per revolution R360 = Ed⁴/10.8DN; 90%-of-loaded-ID mandrel guideline; loaded-diameter approximation. [16] BQUQ "How to Design a Torsion Spring" — rate equation k = Ed⁴/3667·D·Na and diameter-shrink equation ΔID = θD/360N (worked: 12 mm, 5 coils, 90° → 0.6 mm); index range 4–16 guidance. [17] SPIROL "Disc Spring Design Guide" — series A/B/C h0/t ≈ 0.4/0.75/1.3 and De/t ≈ 18/28/40; force/deflection predictability to 75% of h0; safe flattening to h0/t ≈ 1.5 with the flat plateau for wear compensation; preload at 0.15·h0. [18] CBallast "DIN 2093 Disc Spring Catalogue" — series definitions and h0/t ranges; core sizes 8–250 mm; load convention F(0.75h0); materials 50CrV4/51CrV4 (42–48 HRC), 30WCrV6-3 above 250 °C, X7CrNiAl17-7 for corrosion; working range 0.15–0.75 h0 and the bore-fatigue limit beyond. [19] Engineering Hardware "DIN 2093 disc springs" table — standard size/load pairs incl. 8 mm → 210 N, 10 mm → 320 N, 25 × 12.2 × 1.5 → 2,910 N flattening load. [20] ISO 10243 die-spring references: Surisa (ISO 10243:2019 scope; housing interchangeability; 52SiCrNi5 material; Dh 10–63 mm, L0 25–305 mm; forces ~100 N–7 kN core, to 45 kN extended; eight colour series), Axiom Molds (deflection-per-class table 40/37.5/30/25%; long-life vs max guidance; "select one class above minimum" 20% margin; ~10× fatigue-life penalty for 5% over-deflection), Dingli (load-class semantics), JIS B 5012 cross-reference. [21] BQUQ "What Is a Die Spring?" — representative 25 mm OD × 50 mm rate bands (15–25 / 30–45 / 50–70 / 80–110 N/mm by class), fatigue-life tiers at max deflection, 220 °C service context. [22] EN 15800 / DIN 2095 tolerance references (BQUQ spring-tolerance tables; Yinfeng spring-machinery tolerance guide) — class load tolerances ±5/±10/±15%, free-length and OD ranges, squareness 1–3°, cost deltas (class 1 +15–25%, class 3 −10–20%), lead times 5/7/12 working days, ±3% achievable capability. [23] Gutekunst + Co. KG standard spring data sheets — presetting performed in production for springs with setting tendency; tolerances per DIN EN 15800; spring testing per DIN ISO 2859-1 test level II; dynamic data valid only for shot-peened hardened surfaces. [24] Market and industry data: Grand View Research "India Spring Market Size & Outlook 2025–2030" (USD 563.7 M in 2024 → USD 861.0 M in 2030, 7.4% CAGR; helical springs 46.73% of 2024 revenue; India 4.8% of global market; leaf/helical/disc segmentation); Jamna Auto Industries investor presentations (Indian spring market ≈ ₹5,000 crore; 68–70% OEM share; capacity ~240,000–300,000 t/yr and global leaf-capacity comparisons: Rassini ≈250 kt, Hendrickson ≈216 kt, NHK ≈150 kt; Toyo ≈20%, Soni ≈5% domestic shares); 6Wresearch India automotive suspension spring market (imports +20.4% CAGR 2020–2024; origins Germany/USA/China/South Korea/Japan). [25] Stress-relief practice and research: ENG-Tips thread on music-wire stress relief (200–230 °C, 10–30 min guidance) and ScienceDirect "Stress relief effect on fatigue and relaxation of compression springs" (heat-treatment time/temperature classes including 190 °C studies; relaxation as function of stress, temperature, material). [26] MachineCalcs "Music Wire Spring Calculator" and material references — classic moduli set (G = 79.3 GPa / E ≈ 207 GPa for A228) and 250 °F / 121 °C service limit; used as the moduli/comparison basis for the worked examples.

Note on sources: the governing texts for every formula, allowable and limit in this guide are the standard machine-design reference (Shigley/Budynas–Nisbett, Chapter 10), the spring-technology design sheets (IST), and the manufacturer standards literature (Associated Spring, DIN 2093/DIN EN 16983, ISO 10243, EN 15800/DIN 2095, EN 10270 wire classes, DIN ISO 2859-1 sampling). Where vendor and academic references disagree — moduli by 1–3%, allowable stress bands, surge-formula end conditions, die-spring deflection percentages by a few points between sources — this guide prints the working value and the conservative alternative beside it, because a spring is a contract between a specific wire, a specific fixture and a specific duty cycle, and no general table has ever seen any of the three. Design each spring against your own load points, your own temperature, your own cycle count and your own space — then buy it from someone who can hold ±10% of the number you calculated.

FabFlow connects factories with the manufacturing muscle that keeps them running — including spring and wire-form fabrication, press-tool components, and precision machining. Post a job at fabflow.app and get quotes from verified fabricators.

More FabFlow blog posts