Van Winkle's Power Meter is a portable absorption dynamometer that measures the shaft power of a running pulley by clamping a friction belt around it and reading the tension difference on two spring scales. Mill engineers used it in the late 1800s and early 1900s to audit individual machines on a line shaft — looms, lathes, planers — without shutting the shop down. It works by converting mechanical power into heat in the friction belt, then computing power from belt-tension difference, pulley radius, and RPM. A skilled operator could measure a 5 HP machine to within 3% in a few minutes.
Van Winkle Power Meter Interactive Calculator
Vary belt scale readings, pulley radius, and RPM to see absorbed torque and shaft power from the strap-brake meter.
Equation Used
The Van Winkle meter is a strap-brake dynamometer. The difference between the tight-side and slack-side spring-scale readings gives belt friction force. Multiplying by pulley radius gives brake torque, and multiplying torque by angular speed gives absorbed shaft power.
- Belt forces act tangent to the pulley.
- Pulley radius is measured at the belt running surface.
- Wrap is at least 180 deg so the belt reading is stable.
- Power loss is absorbed as heat in the friction belt.
How the Van Winkle's Power Meter Actually Works
The Van Winkle is a strap-brake absorption dynamometer. You wrap a leather or canvas friction belt around the pulley you want to measure, hook each end of the belt to a spring scale, and tighten a hand screw until the belt drags on the rotating pulley. The pulley keeps spinning under load, but now it has to push the belt against friction. The tight side of the belt reads a higher force than the slack side, and the difference between those two readings — multiplied by pulley radius and angular speed — is the shaft power being absorbed.
The geometry matters. The belt has to wrap at least 180° around the pulley or the friction coefficient gets unstable and your readings jump. The spring scales must hang vertically and pull tangent to the pulley, otherwise the moment arm shortens and you under-read. If the belt heats up faster than it can shed heat — which happens above roughly 1 HP per square inch of belt-pulley contact — the leather glazes, the friction coefficient drops, and the meter slowly reads low as the test drags on. That's the classic failure mode: a test that started reading 8 HP drifts down to 6 HP over two minutes because the belt is cooking. Operators learned to take the reading in the first 30 seconds and let the belt cool between runs.
The other thing that bites you is RPM measurement. Power scales linearly with RPM, so a 2% tachometer error becomes a 2% power error directly. Period mechanics used a hand tachometer pressed against the shaft end — a Hasler or a Jaquet — and good practice was to take three readings and average them. Get sloppy with the tach and your whole brake horsepower number is meaningless even if the spring scales are perfect.
Key Components
- Friction belt (strap): A leather or canvas band, typically 50 to 100 mm wide, that wraps around the test pulley with at least 180° of contact. The friction coefficient must stay between 0.25 and 0.35 for stable readings — glazed or oily leather drops below 0.20 and gives drifting numbers.
- Tension spring scales (pair): Two calibrated spring balances reading the tight-side and slack-side belt forces. Common ranges were 0–50 lb and 0–200 lb depending on pulley size. Calibration tolerance must be ±1% of full scale or the power calculation inherits that error doubled.
- Hand-screw tensioner: Adjusts the static tension on the belt to load the pulley. The operator turns it until the spring scales read into the working range without stalling the machine. Backlash in the screw thread should be under 0.25 mm or you can't hold a steady load.
- Tachometer: Measures pulley RPM at the shaft end. A hand-held centrifugal tach reading 0–3000 RPM with ±1% accuracy is the period-correct tool. RPM error feeds straight into the power result, so this is not the place to cut corners.
- Pulley diameter caliper or tape: You measure the pulley diameter once at the start of the test. A 1 mm error on a 300 mm pulley is 0.3% — negligible. But if you measure the rim instead of the belt-running surface, you can be off by 5 mm easily, which is 1.7% straight into the answer.
Who Uses the Van Winkle's Power Meter
Van Winkle's meter and its cousins (Prony brakes, Alden absorption dynamometers, Webber transmission dynamometers) were the only practical way to audit individual machine power before the electric submeter became cheap. Anywhere a line shaft drove multiple machines through belt drops, a portable absorption dynamometer was the standard tool for finding the energy hog, sizing a replacement motor, or settling a warranty dispute over a machine that wasn't delivering rated horsepower. They show up in surviving heritage workshops today for exactly the same reason — you can't put a clamp-on ammeter on a leather belt.
- Heritage textile mills: Auditing individual loom power draw at Quarry Bank Mill, where curators verify a Lancashire loom pulls within its 1880s rated 0.5 HP before public running days.
- Vintage machine shop restoration: Sizing a replacement electric motor when converting a 1910 Rivett 608 lathe from line-shaft drive to standalone — measure the actual cutting-load horsepower under a representative cut, then add 30% headroom.
- Agricultural equipment testing: Spot-checking PTO horsepower delivered by a restored John Deere Model B at threshing-show demonstrations, where you can't get the tractor onto a chassis dyno but you can wrap a strap brake on the belt pulley.
- Marine engine surveys: Verifying auxiliary shaft power on small steam launches and historic fishing boats, where the friction-brake meter sits on the propeller shaft pulley between the engine and the stuffing box.
- Industrial archaeology: Documenting genuine working horsepower of preserved machinery for museum interpretation — the Crossley gas engine collection at Anson Engine Museum has been audited this way to compare measured shaft power against builder's plate ratings.
- Education and engineering teaching labs: Undergraduate mechanical-engineering courses use modern reproductions of the strap-brake principle as a hands-on dynamometer lab, because the physics is visible and the maths is one equation.
The Formula Behind the Van Winkle's Power Meter
The formula gives shaft power as a function of belt-tension difference, pulley radius, and rotational speed. At the low end of the typical operating range — say a 0.25 HP loom countershaft at 200 RPM — the tension difference is small (10–15 lb on a 4-inch pulley) and spring-scale resolution becomes the limiting factor. At the nominal sweet spot, 1–5 HP at 300–800 RPM on a 6–10 inch pulley, you get clean readings with comfortable scale deflection and the belt runs cool enough for a 60-second test. Push above 10 HP or 1500 RPM and the belt starts cooking, friction coefficient drifts, and you have to take the reading fast and let the rig cool between runs. The formula is exact; the practical accuracy lives or dies on whether you stay in that middle band.
Variables
| Symbol | Meaning | Unit (SI) | Unit (Imperial) |
|---|---|---|---|
| P | Shaft power absorbed by the brake | watts (W) | ft·lb/s (divide by 550 for HP) |
| F1 | Tight-side belt tension (high-reading scale) | newtons (N) | pounds-force (lbf) |
| F2 | Slack-side belt tension (low-reading scale) | newtons (N) | pounds-force (lbf) |
| r | Pulley radius at belt-running surface | metres (m) | feet (ft) |
| D | Pulley diameter at belt-running surface | metres (m) | feet (ft) |
| N | Pulley rotational speed | RPM | RPM |
| ω | Pulley angular velocity = 2π × N / 60 | rad/s | rad/s |
Worked Example: Van Winkle's Power Meter in a vintage printing-press restoration shop
A letterpress restoration shop in Portland is auditing the shaft power drawn by a restored 1923 Chandler & Price 10×15 platen press running at 2,500 impressions per hour off a 200 mm flat-belt pulley turning at 420 RPM. The owner wants to confirm the press pulls under 1.5 HP before he commits to a 2 HP single-phase motor for the conversion. He wraps a Van Winkle strap brake around the flywheel pulley, takes spring-scale readings under a representative print run, and reads F1 = 145 N tight side, F2 = 38 N slack side.
Given
- D = 0.200 m
- Nnom = 420 RPM
- F1 = 145 N
- F2 = 38 N
Solution
Step 1 — compute the net belt-tension difference, which is the actual tangential force the pulley fights against:
Step 2 — compute the pulley radius and angular velocity at nominal 420 RPM:
Step 3 — compute nominal shaft power:
That sits comfortably under the owner's 1.5 HP threshold. Now check the operating range. At the low end of a typical Chandler & Price duty cycle — idle stroke with no impression, roughly N = 250 RPM and ΔF dropping to about 45 N because there's no print load — power falls to:
That's bearing drag and flywheel windage only — what you'd feel as a slight pull on the treadle if you were running it by foot. At the high end, full-bed forme printing heavy card stock at 480 RPM with ΔF spiking to 180 N:
Still under 1.5 HP, but now you understand the duty range. The 2 HP motor gives roughly 65% headroom over peak — appropriate for a press that sees occasional heavy lockups.
Result
The press draws a nominal 470 W (0. 63 HP) under a representative print run. That number tells the owner the press is loafing on his bench — well within the rating of any 1 HP motor and trivial for a 2 HP. Across the full duty range, idle is 0.16 HP and peak heavy-stock printing is 1.21 HP, so the 2 HP motor choice carries about 65% headroom over worst-case demand which is the right margin for intermittent peak loads. If the owner measured noticeably higher than 0.63 HP nominal — say 0.9 HP under the same conditions — three failure modes are likely: (1) the belt tensioner has crept and ΔF is reading high because the belt is binding rather than slipping cleanly, (2) the spring scales are out of calibration which you can check in 30 seconds with a known dead weight, or (3) the pulley diameter was measured at the rim instead of at the belt-running surface, adding 5–8 mm of phantom radius that scales the answer up by 5%.
When to Use a Van Winkle's Power Meter and When Not To
The Van Winkle is one option among several absorption-dynamometer designs, and the choice depends on power level, accuracy needed, and what you can wrap around the test machine. Here's how it stacks up against the two most common alternatives — the Prony brake (lever-and-block design) and a modern in-line torque transducer.
| Property | Van Winkle Strap Brake | Prony Brake | In-line Torque Transducer |
|---|---|---|---|
| Typical accuracy | ±3% with calibrated scales | ±2% with good lever geometry | ±0.1% to ±0.5% |
| Practical power range | 0.25 to 25 HP | 0.5 to 500 HP | 0.01 to 5,000 HP |
| Test duration limit | 30–90 sec before belt glazes | Unlimited with water cooling | Unlimited |
| Setup time on a running machine | 5–10 minutes | 30–60 minutes | Hours (requires shaft cut) |
| Capital cost (2024 equivalent) | $200–$600 reproduction | $500–$2,500 | $3,000–$25,000 |
| Suits portable field audits | Excellent | Poor (heavy) | Poor (permanent install) |
| Failure mode under sustained load | Belt glazing, drifting friction coefficient | Block charring, lever slop | Electronic drift, calibration loss |
Frequently Asked Questions About Van Winkle's Power Meter
The belt is heating up and the leather is glazing. Friction coefficient on cool tannin-treated leather sits around 0.30, but as belt temperature climbs past 80°C the surface oils migrate, the leather hardens, and the coefficient drops toward 0.20. ΔF falls with it because the pulley is now slipping more freely against the belt for the same tensioner setting.
Rule of thumb: take your reading in the first 30 seconds. If you need a longer test, swap to a wider belt — going from 50 mm to 100 mm halves the heat flux per unit area and buys you maybe two minutes of stable reading. For sustained measurement above 5 HP, a Van Winkle is the wrong tool — go to a water-cooled Prony brake.
15 HP is right at the edge of the Van Winkle's comfortable working range. The deciding factor is test duration. If you only need a 30-second reading to verify nameplate horsepower, the strap brake works fine and sets up in 10 minutes. If you need to load-test the motor for several minutes — say to verify it can hold rated load without overheating — the Prony's water-cooled blocks are the only option that won't glaze.
A second factor is pulley size. Van Winkles work best on 4–12 inch pulleys. If your motor drives through a 24 inch flywheel, the strap belt becomes unwieldy and the Prony is mechanically simpler.
Two suspects, in order of likelihood. First, your spring scales were calibrated vertically but you're reading them at an angle — any deviation from vertical introduces a cosine error and scales above 30° off-axis under-read by enough to throw the answer 5–10% high once you compensate by over-tensioning. Hang them plumb and re-read.
Second, you're measuring no-load shaft power, not delivered output. Motor nameplates show output power; the brake measures input shaft torque, which on an unloaded motor includes magnetising losses, windage, and bearing friction that don't appear at the rated operating point. Load the motor to at least 50% rated to get a meaningful comparison.
You can, but the friction coefficient is different and less predictable. Polyester webbing on cast iron sits around 0.20–0.25 and varies more with temperature than leather does. Nylon webbing is worse — it stretches under tension, so your ΔF reading is partially absorbed by belt elongation rather than transmitted to the spring scale, and you'll under-read by 5–15%.
If you must use synthetic, calibrate the rig against a known load (a hanging weight on a separate test pulley driven at a known RPM) before trusting any readings. Leather remains the standard for a reason.
Belt-tension theory (the Eytelwein/capstan equation) says the ratio F1/F2 grows exponentially with wrap angle. Below 180° wrap, you can't develop enough ΔF without slipping the belt, and ΔF readings become unstable — they jump every time the belt creeps a little.
Practical minimum is 180°. Comfortable working range is 200–270°. Above 270° the belt starts to bind and the slack side reads near zero, which compresses your useful scale range. If your test pulley geometry won't let you wrap a full 180°, add an idler pulley to extend the wrap rather than living with the error.
Yes. A crowned pulley has a larger diameter at the centre than at the edges, typically by 0.5–1.5 mm depending on width. The belt rides on the crown, so the effective belt-running radius is the crown peak, not the edge. Measure with a caliper across the highest point of the crown.
If you measure at the edge instead, you under-state radius by maybe 1 mm on a 200 mm pulley — that's a 0.5% error in the power result, small but systematic. For better than 2% overall accuracy on a crowned pulley, this matters.
References & Further Reading
- Wikipedia contributors. Dynamometer. Wikipedia
Building or designing a mechanism like this?
Explore the precision-engineered motion control hardware used by mechanical engineers, makers, and product designers.
