A Curved Cone Pulley is a single-piece pulley whose outer surface follows a smooth concave or convex curve rather than discrete steps, allowing a flat belt to slide axially and run at any diameter along the curve to give a continuous range of drive ratios. A typical machine-shop curved cone covers a 4:1 to 6:1 speed range with the belt shifting across 150-300 mm of axial travel. The curve solves the problem of needing variable spindle speed without a gearbox or stepped pulley jumps, and you still see it on textile spinning frames, line-shaft drives, and the original Hardinge bench lathes.
Curved Cone Pulleys Interactive Calculator
Vary driver speed, effective pulley diameters, and slip to see speed ratio, output RPM, belt speed, and torque multiplication.
Equation Used
The speed ratio for a curved cone pulley pair is set by the effective rolling diameters where the belt sits: output speed equals input speed multiplied by D1/D2. A small slip correction can be applied to estimate real output RPM.
- Open flat-belt drive with effective rolling diameters at the current belt position.
- Mirrored curved cones maintain approximately constant belt length and tension.
- Slip is applied only as a simple speed-loss correction.
- Torque multiplier is ideal and neglects bearing and belt losses.
Operating Principle of the Curved Cone Pulleys
A Curved Cone Pulley works by changing the effective rolling diameter the flat belt sees. Slide the belt to the small end of the curve and the pulley turns through a small circumference per shaft revolution, which drops belt speed and raises ratio. Slide it to the fat end and the opposite happens. Pair two cones nose-to-tail on parallel shafts — one tapering up, one tapering down — and the belt length stays constant at every position, which is the whole reason this geometry exists. If the two curves are not mirrored correctly the belt goes slack at one end of travel and snaps tight at the other, and you will hear it slap the guard before you see the ratio drift.
The curve itself is not arbitrary. For a constant-length open belt drive between two parallel shafts at fixed centre distance C, the sum of pulley radii at any axial position has to satisfy the open belt length equation, and that constraint defines the curve as a hyperbolic-like profile, not a straight cone. Cut a straight taper instead of the true profile and the belt will track but tension will swing by 15-25% across the range, which kills belt life on anything wider than 50 mm. The crown — that gentle convex bulge across the belt width at any axial slice — is what keeps the belt centred laterally; without crown the belt walks off the small end within minutes.
Tolerances matter more than people expect. Surface roughness should sit around Ra 0.8-1.6 µm — too smooth and the flat belt slips under load, too rough and the leather or rubber facing wears out in a season. Axial alignment of the two cones must be within 0.5 mm over a 300 mm centre distance, otherwise the belt edge rides one flange and frays. Failure modes are predictable: belt edge fraying (alignment), tension pulsing once per revolution (out-of-round cone or bent shaft), and ratio drift mid-cut (belt fork mechanism backing off because its detent spring fatigued).
Key Components
- Driver Cone: The cone mounted on the input shaft, usually tapering from large diameter at the motor end to small diameter at the far end. Cast iron or steel, machined to the true hyperbolic profile with a 0.05-0.10 mm crown across the belt width at every slice.
- Driven Cone: Mirror-image cone on the output shaft, tapering in the opposite direction so the belt-length constraint stays satisfied across the full travel. Must run on the same centre distance as the driver — typically 400-900 mm on machine tool sized cones.
- Flat Belt: Leather, rubber-canvas, or modern polyamide flat belt sized so the belt width is 60-80% of the narrowest cone slice it will ever sit on. Tension set to give about 1.5-2.0% strain at rest, measured with a sonic tension gauge or simple deflection check.
- Belt Shifter Fork: A two-pronged guide that straddles the belt and slides axially on a leadscrew or rack, dragging the belt to a new diameter while the drive runs. The fork must release the belt at the exact moment it transfers — pinch the belt between the prongs and you tear the edge.
- Shifter Leadscrew or Cam: Mechanical link between the operator handwheel and the fork. A 5-10 mm pitch acme leadscrew is common; the detent or friction lock holds the fork against the belt's lateral creep force, which can reach 50-100 N on a wide drive.
- Crowned Surface Profile: The slight convex curvature across the belt width at every axial slice. Crown height of 0.5-1.0% of belt width is the standard rule. Without it the belt drifts to one end within seconds, regardless of how perfectly aligned the shafts are.
Real-World Applications of the Curved Cone Pulleys
You find Curved Cone Pulleys wherever a continuously variable belt drive is wanted without the cost or maintenance burden of a Reeves drive, a CVT, or a VFD-plus-induction motor. They were the dominant variable-speed solution from roughly 1880 to 1950 and they still live on in textile machinery, restoration-grade machine tools, and a surprising number of lab-scale lapping and polishing rigs. The reason they survived is simple — they are passive, repairable with hand tools, and they don't care about voltage spikes or electrical noise.
- Machine Tools: Hardinge HLV-H toolroom lathe headstock used a stepped-and-curved cone arrangement to give 125-3000 RPM at the spindle without changing gears.
- Textile Machinery: Saco-Lowell ring spinning frames used curved cone drives to taper spindle speed during package build-up, matching yarn tension as the bobbin diameter grew.
- Line-Shaft Power Transmission: Pre-1940 New England textile mills like the Boott Mills in Lowell, Massachusetts ran curved cone pulleys off the main line shaft to drive individual looms at independently variable speeds.
- Lapping and Polishing: Logitech PM5 lapping machine variants used a small curved cone drive to vary plate RPM during a single lap cycle without stopping the machine.
- Drill Presses and Pillar Drills: Pre-war Buffalo Forge No. 18 column drills shipped with curved cone heads giving 60-1200 RPM through a single belt shift.
- Paper and Pulp: Beloit Jones beater drives used cone pulley speed control on the refining roll to ramp fibre treatment rate during a batch.
The Formula Behind the Curved Cone Pulleys
The core calculation is the speed ratio at any axial position along the curve, which sets the output RPM you will actually see at the driven shaft. At the small-diameter end of the cone the belt sees the lowest output speed and highest torque multiplication — useful for heavy roughing cuts but also where belt slip risk peaks because contact arc shrinks. At the large-diameter end you get top spindle speed but the contact arc on the driven cone tightens, so transmissible power drops. The sweet spot for steady-state machining sits roughly mid-travel where both arcs of contact are within 10% of 180°.
Variables
| Symbol | Meaning | Unit (SI) | Unit (Imperial) |
|---|---|---|---|
| N2 | Output (driven) shaft speed at axial position x | RPM | RPM |
| N1 | Input (driver) shaft speed, set by the motor | RPM | RPM |
| D1(x) | Driver cone diameter at axial belt position x | mm | in |
| D2(x) | Driven cone diameter at axial belt position x (constrained so D1+D2 satisfies constant belt length) | mm | in |
| x | Axial position of the belt along the cone, measured from the driver large-end face | mm | in |
Worked Example: Curved Cone Pulleys in a restored Hardinge bench lathe headstock
Picking the belt position on a restored Hardinge cone-head bench lathe in a private machine shop in Sheffield. The motor runs the driver cone at a fixed 1450 RPM. Driver cone tapers from 180 mm at one end to 60 mm at the other over 240 mm of axial travel; driven cone mirrors it (60 mm to 180 mm) with centre distance 520 mm. You want to pick the belt position for a finishing pass on a 25 mm diameter mild steel shaft, target surface speed roughly 30 m/min.
Given
- N1 = 1450 RPM
- D1 range = 60 to 180 mm
- D2 range = 180 to 60 mm
- Cone axial travel = 240 mm
- Target surface speed = 30 m/min
- Workpiece diameter = 25 mm
Solution
Step 1 — convert target surface speed to required spindle RPM. Surface speed v = π × Dwork × Nspindle, so:
Step 2 — at the nominal mid-travel position (x = 120 mm), both cones sit at 120 mm diameter, so the ratio is 1:1:
That is far too fast for the finishing pass — at 1450 RPM the 25 mm shaft sees 114 m/min, which will glaze the HSS tool nose inside one cut. So the belt has to shift toward the driven-cone large end.
Step 3 — at the low-speed end of travel, belt sits on driver D1 = 60 mm and driven D2 = 180 mm:
That gives a surface speed of 38 m/min on the 25 mm shaft — close to target and the practical low-speed limit of this cone. At the opposite high-speed end (D1 = 180, D2 = 60):
Useful for small-diameter brass or aluminium work but well above what the plain bronze headstock bearings on a Hardinge bench lathe will tolerate continuously — you would smell the oil burning off inside 10 minutes. Step 4 — solve for the exact axial position that gives 382 RPM. Required ratio is D1/D2 = 382/1450 = 0.263, and along a linear taper approximation that puts the belt about 15-20 mm in from the driver small end.
Result
The belt should sit roughly 15-20 mm from the driver small end, giving about 480 RPM at the spindle and 38 m/min on the 25 mm workpiece — close enough to target for a finishing cut. At the low-end position the spindle creeps at 483 RPM and you can hear individual tool-tip contacts; at mid-travel it spins at 1450 RPM which is fine for small drilling but glazes carbide on steel; at the high-end 4350 RPM the headstock oil starts smoking within minutes on plain bronze bearings, so treat that range as occasional-use only. If you measure spindle speed with a tach and read 380 RPM instead of 480, the most likely causes are: belt slip from a glazed driver surface (clean with belt dressing, not sandpaper), a stretched flat belt giving 0.5-1% slip per pass under load, or the shifter fork creeping back toward mid-travel because the detent spring has lost preload. Check fork position against the headstock witness mark before blaming the calculation.
When to Use a Curved Cone Pulleys and When Not To
Curved cone pulleys compete against three modern alternatives whenever a designer wants variable spindle or drive speed. The honest tradeoff is between mechanical simplicity and ratio precision — cones win on robustness and lose on accuracy. Here is how they stack up on the dimensions that actually matter when you are choosing a drive.
| Property | Curved Cone Pulley | Reeves Variable Speed Drive | VFD + Induction Motor |
|---|---|---|---|
| Speed ratio range | 4:1 to 6:1 typical | 6:1 to 10:1 | 20:1+ with vector control |
| Ratio setting accuracy | ±5% (operator-set, no scale) | ±1-2% with indicator | ±0.1% with encoder feedback |
| Cost (machine-shop scale) | Low — castings + flat belt | Medium — sliding sheaves | Medium — drive + motor |
| Maintenance interval | Belt replace every 3-5 years, otherwise zero | Sheave regrease every 500 hr | Effectively zero mechanical |
| Lifespan | 50+ years on the cones themselves | 15-25 years before sheave wear | 10-15 years on drive electronics |
| Power capacity | Up to 15 kW with wide flat belt | Up to 75 kW | Limited only by motor frame |
| Speed change while loaded | Possible but rough on belt | Smooth and continuous | Smooth and continuous |
| Application fit | Restoration, textile, low-noise lab | Conveyors, mixers, mid-power process | Modern machine tools, pumps, fans |
Frequently Asked Questions About Curved Cone Pulleys
Almost always a missing or insufficient crown across the belt width. A flat belt naturally climbs to the highest-diameter point on the pulley face — that is the self-centring effect crowned pulleys rely on. If the cone surface is dead-flat across the belt width at every axial slice, there is nothing pulling the belt back to centre and any tiny shaft tilt or belt seam asymmetry sends it sideways.
Check with a straightedge laid across the belt path at the small end specifically. You should feel a 0.3-0.8 mm convex rise from edge to centre on a 50 mm wide belt. If it is flat, the cone was either machined wrong or worn flat — either way the fix is to re-machine the crown back in.
You can, and on narrow belts (under 30 mm) on short cones it is hard to tell the difference. On anything wider or longer, the belt tension swings noticeably across the travel because the sum D1+D2 no longer holds constant for a fixed centre distance — a straight taper makes that sum vary by a few millimetres across the range.
The practical consequence is belt life drops by roughly half and you get tension pulsing at the ends of travel that you can feel through the handwheel. For a quick demo cone or a hobby drill press, fine. For a production lathe, cut the true profile.
The decision comes down to whether you ever change speed mid-cut. If you set spindle speed once at the start of a job and leave it there, a stepped cone with 4-6 fixed ratios is simpler, cheaper, and the belt sits in a defined groove so it never drifts. If you want to ramp speed during a facing cut to hold constant surface speed as the diameter shrinks, the curved cone is the only mechanical option short of a Reeves drive.
Rule of thumb — production shops doing repeat work pick stepped, toolroom and restoration shops pick curved.
This is almost always belt creep, not slip. Under load the tight side of the flat belt stretches more than the slack side, and that tiny strain difference makes the belt advance slightly less per driver revolution than the geometry predicts. Creep of 1-2% is normal on a leather flat belt at rated load; 5-10% means the belt is undertensioned or near end of life.
Quick check — measure belt deflection at mid-span with a 10 N push. It should sag about 1.5% of the centre distance. If it sags 3% or more, the belt has stretched and either needs re-tensioning at the takeup or replacing.
For leather flat belts, around 25-30 m/s belt speed at the largest cone diameter. Rubber-canvas belts go to 35 m/s, modern polyamide flat belts to 50 m/s. Above the limit the belt centrifugal tension reduces the effective grip force on the pulley and you lose transmissible torque before you see any obvious failure.
What fails first is grip, not the belt itself. You will notice it as a sudden inability to take a heavy cut at the high-speed end of travel even though the spindle freewheels fine. The belt is still intact — it just cannot push power through the contact arc anymore. Drop one cone diameter step and grip returns.
A consistent offset across the full range — not just at one position — points to a wrong assumption about effective rolling diameter rather than a mechanism fault. Flat belts roll at their neutral axis, which sits roughly half the belt thickness above the pulley surface. If your belt is 6 mm thick and you used the bare cone diameter in the calculation, you underestimated effective diameter by 6 mm at each cone.
Recompute using Deffective = Dcone + tbelt on both pulleys. The 8% error usually disappears. If it does not, check the tach itself against a strobe — cheap optical tachs read low on shiny shafts.
References & Further Reading
- Wikipedia contributors. Pulley. Wikipedia
Building or designing a mechanism like this?
Explore the precision-engineered motion control hardware used by mechanical engineers, makers, and product designers.
