Evans Friction Cones are a continuously variable speed drive that uses two tapered cones mounted on parallel shafts with a tensioned leather belt looped between them. Unlike stepped cone pulleys, which only give discrete ratios, the Evans design lets you slide the belt along the cones to get any ratio between the limits — no gear change, no stopping the machine. The drive transmits torque purely through friction at the belt-to-cone contact, and was used heavily in 19th-century mill work and early machine tools where smooth, on-the-fly speed adjustment beat swapping pulleys by hand.
Evans Friction Cones Interactive Calculator
Vary belt position and cone end diameters to see the continuously variable speed ratio and effective belt diameters.
Equation Used
The Evans friction cone drive changes speed by moving the belt to different effective diameters on two opposing cones. The ideal speed ratio is the driver diameter at the belt divided by the driven diameter at the belt: n2/n1 = D1/D2.
- No belt slip between leather belt and cones.
- Opposed cones have linear taper over the belt travel.
- Driver cone grows from small to large diameter as belt position increases.
- Driven cone tapers in the opposite direction.
How the Evans Friction Cones Actually Works
Two cones sit nose-to-tail on parallel shafts — one cone tapers left to right, the other tapers right to left. A continuous leather belt wraps around both. When the belt sits at the midpoint, the effective driving and driven diameters are equal and the ratio is 1:1. Slide the belt toward the fat end of the driver and the thin end of the driven cone, and the speed ratio climbs. Slide it the other way and the ratio drops. The belt shifter — a forked guide on a leadscrew — moves the belt axially while the machine runs.
The whole thing only works because of friction. The belt grips the cone surface through normal force generated by belt tension, and the available torque is governed by the capstan equation and the friction coefficient between leather and cast iron — typically μ ≈ 0.25 to 0.35 dry, lower if oil contaminates the contact. Push past that limit and the belt slips, the cones glaze, and you lose drive. That's the dominant failure mode. The cone surfaces also have to be true to within a few thousandths of an inch over their full length — if the taper isn't straight, the belt walks under load and refuses to hold the position you set with the shifter.
Belt tension is the other knob you tune. Too slack and the drive slips at the slightest load spike. Too tight and the leather stretches, the bearings overload, and the cones wear conical grooves at the most-used belt position. A well-tuned Evans drive on a 19th-century lathe typically ran with belt tensions around 15 to 25 lbf per inch of belt width — enough to grip, not so much that the leather creeps overnight.
Key Components
- Driving Cone: The input cone, fixed to the prime-mover shaft. Cast iron or steel, ground straight along the taper to within roughly 0.05 mm over a 300 mm length. Surface finish matters — too smooth and the leather skids, too rough and it abrades the belt.
- Driven Cone: The output cone, mounted on the parallel output shaft and tapered in the opposite direction so that summed diameters at any axial position stay constant. This keeps the belt centre-distance and belt length the same regardless of where the belt sits.
- Leather Belt: The friction transmitter. Traditionally oak-tanned leather, 50 to 150 mm wide depending on power, run hair-side-out for grip. Modern rebuilds sometimes use chrome-tanned leather or rubberised fabric, with μ values in the 0.25 to 0.4 range.
- Belt Shifter (Belt Striker): A forked guide that straddles the belt and moves axially along a leadscrew or hand-cranked rack. Shifts the belt 1 to 2 mm per revolution of the handwheel — slow enough that the operator can hold a target speed within ±2%.
- Belt Tensioner / Idler: A spring-loaded jockey pulley or weighted carriage that maintains belt tension as the belt walks across the slightly varying effective wrap geometry. Keeps tension within a 10 to 15% band over the full shift range.
Industries That Rely on the Evans Friction Cones
Evans Friction Cones showed up wherever a 19th- or early-20th-century engineer needed continuous speed variation without gear changes. They were never a high-power solution — friction limits cap them around 5 to 10 hp in a typical industrial size — but for small machine tools, instruments, and laboratory drives they were ideal. You see them today mostly in restored mill exhibits, antique lathes, and a handful of niche test rigs where the smooth, gearless ratio change is genuinely useful.
- Historic Machine Tools: Speed-change drive on 19th-century engine lathes, including some Holtzapffel ornamental turning lathes where operators needed fine spindle-speed control mid-cut without swapping back-gear ratios.
- Textile Mills: Spindle-speed control on early Lancashire ring-spinning frames, where the cone drive let an overlooker trim spindle RPM to match yarn count without stopping the frame.
- Paper Manufacturing: Calender-roll and reel drives on Fourdrinier paper machines from the 1880s through the 1920s, allowing the reel speed to track the wire speed as the sheet thickness changed.
- Laboratory Test Rigs: Variable-speed input on early fatigue-testing machines like the Wöhler-style rotating-beam testers, where smooth speed ramp-up revealed resonance points without the discontinuities of stepped pulleys.
- Printing Presses: Inking-roller drive trim on 1900-era flatbed cylinder presses, used to dial in ink film thickness independent of impression-cylinder speed.
- Museum Restorations: Demonstration drives on working steam-mill restorations, including several at the Quarry Bank Mill in Cheshire and the Slater Mill site in Pawtucket — both run cone-drive line shafts as part of their public exhibits.
The Formula Behind the Evans Friction Cones
The speed ratio of an Evans cone drive depends entirely on where the belt sits along the cones. At the midpoint you get 1:1. At the extreme ends the ratio is set by the diameter ratio between the fat end of one cone and the thin end of the other — typically 3:1 to 5:1 in either direction on a practical industrial unit. The sweet spot for steady running is the middle 60% of belt travel, because at the very ends the wrap angle on the thin section drops and slip risk climbs sharply. Push the belt to within 10% of either end and torque capacity falls off a cliff.
Variables
| Symbol | Meaning | Unit (SI) | Unit (Imperial) |
|---|---|---|---|
| i | Speed ratio (output speed divided by input speed) | dimensionless | dimensionless |
| Nin | Input shaft rotational speed | rev/min | RPM |
| Nout | Output shaft rotational speed | rev/min | RPM |
| Ddrive(x) | Effective diameter of the driving cone at belt axial position x | mm | in |
| Ddriven(x) | Effective diameter of the driven cone at belt axial position x | mm | in |
| x | Axial position of the belt along the cone pair, measured from the midpoint | mm | in |
Worked Example: Evans Friction Cones in a restored 1890s ornamental turning lathe
You are restoring the spindle drive on an 1890s Holtzapffel-style ornamental turning lathe and replacing the original cone-pulley system with a working Evans Friction Cone drive. The driving cone runs at a constant 600 RPM from a countershaft. Each cone is 400 mm long, with diameters tapering from 60 mm at the thin end to 180 mm at the fat end. You need to know the spindle speed range you can dial in, and where the practical sweet spot sits.
Given
- Nin = 600 RPM
- Lcone = 400 mm
- Dmin = 60 mm
- Dmax = 180 mm
Solution
Step 1 — at the nominal midpoint position, both effective diameters are equal at (60 + 180) / 2 = 120 mm, so the ratio is 1:1:
This is the sweet spot. Wrap angle is symmetric on both cones, belt tension distributes evenly, and torque capacity is at its maximum. For a typical ornamental turning operation at 600 RPM you get clean cuts and predictable feed rates.
Step 2 — at the low end of the practical operating range, slide the belt 80% toward the thin end of the driving cone (x = -160 mm). Driving diameter drops to roughly 72 mm, driven diameter rises to roughly 168 mm:
At 257 RPM the spindle runs slow enough for heavy roughing cuts on hardwood, but you are now within 20% of the thin end of the driver. Wrap angle on that cone is shrinking, and a heavy cut spike will slip the belt before it stalls the spindle. You will hear a brief chirp from the leather when load peaks.
Step 3 — at the high end, slide the belt 80% toward the fat end of the driving cone (x = +160 mm):
1400 RPM is fast enough for fine finishing passes and rose-engine work, but the belt is now perched near the fat end of the driver where the slightest taper error makes it walk. In practice, restorers keep the working range between 350 and 1100 RPM — the middle 60% of belt travel — and accept that the extreme ends are theoretical, not operational.
Result
Nominal output at the belt midpoint is 600 RPM, matching the input one-to-one and giving full torque capacity. The full theoretical range runs from 257 RPM at the low end to 1400 RPM at the high end, but the practical operating window is roughly 350 to 1100 RPM — the extreme ends lose wrap angle and slip under any real cutting load. If your measured spindle speed differs from the predicted value, look at three things: (1) belt creep — leather stretches over the first few hours of run-in and loses 2 to 5% effective ratio, fix by re-tensioning the idler; (2) belt walking — if the cones aren't ground straight to within 0.05 mm over their length, the belt drifts axially under load and the actual ratio shifts away from where the shifter says it should be; (3) glazed cone surface — once the cast-iron cone polishes to a mirror finish, μ drops below 0.2 and the belt slips at any heavy cut, restored by a light scuff with 120-grit emery and a wipe with rosin.
When to Use a Evans Friction Cones and When Not To
Evans Friction Cones compete against stepped cone pulleys (their 19th-century rival), modern V-belt CVTs (their 20th-century replacement), and electronic VFD-driven motors (their modern replacement). Each wins on different axes — here's where the Evans drive actually lands.
| Property | Evans Friction Cones | Stepped Cone Pulleys | Modern V-Belt CVT |
|---|---|---|---|
| Speed change while running | Yes — continuous, on-the-fly | No — must stop machine to swap belt | Yes — continuous |
| Ratio range (typical) | 3:1 to 5:1 | 3:1 to 6:1 in 3-5 discrete steps | 5:1 to 8:1 |
| Maximum power | 5-10 hp | 20-50 hp | 50-200 hp |
| Slip under shock load | High — limited by μ ≈ 0.3 | Low — full belt wrap on each step | Low — variable-pitch sheaves grip via wedge action |
| Speed-holding accuracy under varying load | ±3-5% (creeps with load) | Exact at each step | ±1-2% |
| Maintenance interval (belt) | 6-18 months — leather stretches and glazes | 2-5 years — belt only flexes, no axial shift | 3-7 years — rubber/fabric belts |
| Capital cost (relative) | Low (period part) / High (modern rebuild) | Low | Medium |
| Best application fit | Light-duty machine tools, demonstrations, instrument drives | Any workshop where stopping to shift is acceptable | Industrial drives, scooters, snowmobiles, agricultural equipment |
Frequently Asked Questions About Evans Friction Cones
That's a wrap-angle problem combined with cone runout. Under load the belt momentarily slips, and slip on a tapered surface always pushes the belt toward the larger diameter because that's where surface speed is highest. If your cones are ground perfectly straight, this self-correcting drift is small and the shifter holds it. If the cones have even 0.1 mm of taper error, the drift compounds and the belt climbs.
Check the cones with a straight edge along the taper. If you see light under the edge, the cone needs regrinding. A short-term fix is to add a stronger spring on the shifter fork to physically resist the walk, but you're masking the real issue.
If the machine will be displayed running and authenticity matters — keep the cone drive. It's part of the machine's character, and visitors at places like Quarry Bank Mill specifically come to see line-shaft and friction drives in motion. If the machine will be used for actual production work, a VFD on a 3-phase motor gives you ±0.5% speed accuracy, no slip under load, and no belt to retension every six months.
The middle ground some restorers take: install a hidden VFD-driven motor that turns the original countershaft at constant RPM, and leave the Evans cone drive as the user-facing speed control. You get period-correct operation at the spindle and modern reliability upstream.
That's creep, not slip, and it's expected. Friction belt drives always lose 1 to 5% to elastic creep — the belt stretches slightly as it enters the loaded side of the wrap and contracts as it leaves the unloaded side. The contracting portion travels less distance per cone revolution than the geometry predicts, so the driven cone runs slow.
You can't eliminate creep, only manage it. Higher belt tension reduces creep but accelerates wear. Most period drawings of Evans drives assume 3% creep loss baked into the design ratios, so a 4% measured loss is well within the normal band. If you measure 8% or more, the belt is glazed or under-tensioned.
Rule of thumb for oak-tanned leather running on cast iron at 600 RPM is roughly 1 hp per inch of belt width at moderate tension. So 2 hp wants a 50 to 65 mm belt as a minimum, and you'd typically size up to 75 mm to give margin for the slip-prone end positions of the cone travel.
Don't go wider than the thin end diameter of the cone divided by 1.5 — a 60 mm thin-end cone with a 75 mm belt will overhang the cone at the extreme position and the edge of the belt will fold and abrade. Match belt width to the smallest effective diameter you'll actually run at.
Two suspects, in this order. First, the leather belt heats up under continuous running and the natural oils migrate to the contact surface, dropping μ from around 0.3 to closer to 0.18. The fix is rosin or belt dressing on the inner face — wipe a thin film on a stopped belt, run for 30 seconds at no load, then resume.
Second, thermal expansion of the cast-iron cones can slacken belt tension by 1 to 2% as everything reaches operating temperature. If your tensioner is a fixed dead weight you'll be fine, but a spring-loaded idler at the wrong preload setting can lose grip authority once the system warms. Re-check tension hot, not cold.
The geometry is fully reversible — there's nothing in the cone-and-belt arrangement that prefers a direction. You can drive from either cone, and the speed ratio relationship is identical. What does change is wear pattern: the loaded side of the belt wrap carries the higher tension, so reversing direction shifts which side wears faster.
If you reverse direction routinely, expect the belt to develop a more even wear pattern and last roughly 20% longer than a unidirectional install. Just don't reverse under load — slack-side-to-tight-side transition under torque snaps belts.
References & Further Reading
- Wikipedia contributors. Continuously variable transmission. Wikipedia
Building or designing a mechanism like this?
Explore the precision-engineered motion control hardware used by mechanical engineers, makers, and product designers.
