Tatham's Dynamometer is a transmission-type dynamometer that measures the power passing along a belt drive by routing the belt over a differential pulley arrangement and balancing the speed-difference torque against a deadweight on a lever arm. It typically handles shaft powers from about 1 to 50 hp on belt-driven line shafts running 200 to 600 RPM. Mill engineers used it through the late 19th and early 20th centuries to verify the actual power delivered to a machine without absorbing it as heat, unlike a Prony brake. The named application: textile mill power audits in Lancashire and New England.
Operating Principle of the Tatham's Dynamometer
Tatham's Dynamometer sits between the prime mover and the driven machine, in the belt run itself. The belt enters a frame containing two loose pulleys of slightly different diameters mounted on a common spindle — a differential pulley pair. The driving side of the belt wraps one pulley, the driven side wraps the other, and because the pulley diameters differ by a small fixed amount the spindle wants to rotate under the imbalance of belt tensions. We restrain that rotation with a horizontal lever arm carrying a sliding weight. Slide the weight outward until the lever balances horizontal, read the position, and you have your torque.
The whole point of this design is that it is a transmission dynamometer, not an absorption dynamometer. The power passes through to the load. You are not turning useful work into heat the way a rope brake or Prony brake does, which means you can run the test for hours, on a working machine, with the actual production load attached. That makes it indispensable for measuring the power consumed by a specific loom, lathe, or pump on a live mill line shaft.
Get the geometry wrong and the readings drift. The two pulleys must be concentric to within about 0.05 mm of true running — any eccentricity adds a sinusoidal torque ripple that hammers the lever and blurs the weight position. Belt tension also matters: too slack and the belt slips on the smaller pulley, throwing the differential ratio off; too tight and bearing friction in the spindle starts eating into the torque you are trying to measure. Worn spindle bushings are the classic failure mode — once radial play exceeds roughly 0.1 mm the spindle wobbles, the differential ratio walks, and you'll see a 5-10% low reading at full load.
Key Components
- Differential Pulley Pair: Two pulleys of slightly different diameters, typically D₁ = 300 mm and D₂ = 310 mm, fixed to a common spindle. The diameter difference creates the torque arm that the lever measures. The pair must be machined as a single piece or keyed together with zero relative slip — even 0.5° of windup between them invalidates the calibration.
- Lever Arm with Sliding Weight: A horizontal beam typically 600 to 1200 mm long, carrying a calibrated sliding mass. The operator slides the weight until the lever balances level against a fiducial pointer. Lever flex above 0.5 mm at full deflection introduces a noticeable nonlinearity, so the beam is usually a stout steel section with a slenderness ratio under 30.
- Spindle and Bushings: Carries the differential pulley pair and transmits the unbalanced torque to the lever. Bronze bushings with about 0.02 to 0.05 mm running clearance are typical. Once clearance exceeds 0.1 mm the spindle precesses under load and torque readings drop 5-10% low.
- Belt: The same flat leather or cotton belt that normally drives the machine, rerouted through the dynamometer frame. Belt slip on the small pulley is the single most common source of error — a properly tensioned belt should show under 1% creep at rated load.
- Frame and Fiducial Pointer: Cast iron frame supports the spindle bearings and provides the reference mark against which the lever is balanced. The pointer must read to within 1 mm of the lever's neutral plane to keep angular error under 0.1°.
Industries That Rely on the Tatham's Dynamometer
Tatham's Dynamometer earned its keep in industrial settings where you needed the actual power consumed by a working machine, not a bench number. It still turns up in heritage workshops, education labs, and any place where a transmission dynamometer beats an absorption dynamometer on test duration. Where a rope brake dynamometer dumps every watt as heat into a water tank, Tatham's lets the load do useful work while you measure.
- Textile Mills: Power audits on Lancashire weaving sheds — measuring the actual draw of a Northrop or Lancashire loom on a line shaft running 250 RPM, used to size replacement electric motors during the 1920s-30s electrification programmes.
- Heritage Engineering: Crossness Pumping Station and similar Victorian beam-engine museums use transmission dynamometers of the Tatham pattern to demonstrate live shaft power measurement on running engines without absorbing the output.
- Educational Laboratories: Mechanical engineering programs at institutions like IIT Kharagpur and University of Manchester maintain Tatham-type units in their fluid-and-power labs for undergraduate dynamometer comparison experiments.
- Agricultural Power Testing: Early threshing-machine and chaff-cutter manufacturers used Tatham's Dynamometer to certify shaft power consumption of belt-driven implements running off portable steam engines at 350-450 RPM.
- Machine Tool Verification: Pre-WWII Pratt & Whitney and Lodge & Shipley lathe builders ran Tatham-pattern transmission dynamometers in production cells to verify the spindle power matched the nameplate before shipping.
- Marine Auxiliary Drives: Belt-driven bilge and ballast pumps on early-20th-century steam-powered freighters were tested with Tatham units during sea trials to catch undersized motors before commissioning.
The Formula Behind the Tatham's Dynamometer
The formula computes shaft power from the lever balance reading, the differential pulley diameters, and the shaft speed. The interesting bit is the operating range. At the low end of practical use — say a 1 hp lathe drive at 200 RPM — the differential torque is so small that lever arm friction and bearing drag swamp the signal, and accuracy falls to about ±5%. At the nominal mid-range design point, typically 5-15 hp at 300-450 RPM, you can hit ±1.5% repeatably. At the high end, above 30 hp, belt slip on the smaller pulley becomes the limiting factor and accuracy drifts back out toward ±3% unless you increase belt tension, which then increases bearing friction. The sweet spot lives in the middle.
Variables
| Symbol | Meaning | Unit (SI) | Unit (Imperial) |
|---|---|---|---|
| P | Shaft power transmitted through the belt | W | hp |
| W | Sliding deadweight on the lever | N | lbf |
| L | Lever arm length from spindle centre to weight | m | ft |
| N | Shaft rotational speed | RPM | RPM |
| D1 | Diameter of the larger differential pulley | m | ft |
| D2 | Diameter of the smaller differential pulley | m | ft |
Worked Example: Tatham's Dynamometer in a heritage flour mill restoration
A working-museum team restoring a 1908 roller flour mill in Norfolk needs to certify the shaft power drawn by the No. 3 break-roll stand on the main line shaft. The line shaft runs at 320 RPM, the differential pulleys on their restored Tatham unit measure D₁ = 0.305 m and D₂ = 0.295 m, and the lever arm balances at L = 0.750 m with a sliding weight of W = 180 N at the nominal load. They want to know the shaft power, what the reading would mean if the load drops at the start of a milling run, and what to expect at peak grain feed.
Given
- D1 = 0.305 m
- D2 = 0.295 m
- L = 0.750 m
- Wnom = 180 N
- N = 320 RPM
Solution
Step 1 — compute the torque on the lever at nominal load. Torque equals weight times lever length:
Step 2 — at nominal, this lever torque equates to the belt-tension differential acting on the mean pulley radius. Convert to shaft power using the standard rotational-power relation P = T × 2π × N / 60:
This is the design sweet spot for the unit — comfortably inside the ±1.5% accuracy band where lever friction is small relative to the signal and belt slip stays under 1%.
Step 3 — low-end case. At the start of a milling run before grain reaches the rolls, the load drops to roughly 30% of nominal, so the lever balances at about Wlow = 55 N:
At this end of the range bearing friction in the spindle eats roughly 50-100 W on its own, so trust the reading to about ±5%. The lever pointer also tends to drift visibly because there isn't enough net torque to seat the weight firmly against any one position.
Step 4 — high-end case. At peak grain feed the load can spike to about 160% of nominal, calling for Whigh ≈ 290 N to balance:
At this load belt tension is high enough that you'll start to see slip creep on the smaller pulley unless tension is bumped up — which then loads the spindle bushings harder. Above about 11 hp on this size unit you cross out of the linear measurement window.
Result
Nominal shaft power comes out to 4524 W, or about 6. 07 hp at the No. 3 break-roll stand. That number means the roll stand is loaded right where a 1908-era mill designer would have expected — well inside the line shaft's capacity, with headroom for grain-feed surges. Across the full operating range you see the unit reads 1.85 hp at the empty-rolls condition and climbs to 9.77 hp at peak feed, with the cleanest measurements happening between roughly 4 and 8 hp where lever signal-to-friction is best. If your measured power runs 5-10% below the calculated value, suspect belt creep on the smaller pulley first — verify by chalk-marking both pulleys and checking phase after 50 revolutions. If readings oscillate by more than ±2% with the lever level, the differential pulleys are likely no longer concentric within 0.05 mm and need indicating in. If the lever sits low on one side regardless of weight position, the spindle bushing has worn past 0.1 mm clearance and the shaft is precessing under load.
Choosing the Tatham's Dynamometer: Pros and Cons
Tatham's Dynamometer competes with two well-known alternatives in the shaft-power measurement space: the Prony brake (an absorption type) and modern strain-gauge torque transducers (a transmission type using electronics rather than mechanics). Each wins on different engineering dimensions.
| Property | Tatham's Dynamometer | Prony Brake | Strain-Gauge Torque Transducer |
|---|---|---|---|
| Measurement type | Transmission (load runs) | Absorption (load dissipated as heat) | Transmission (load runs) |
| Typical accuracy | ±1.5 to ±3% | ±2 to ±5% | ±0.1 to ±0.5% |
| Power range | 1 to 50 hp | 0.5 to 200 hp | Fractional hp to 5000+ hp |
| Speed range (RPM) | 150 to 800 | 50 to 600 | 0 to 30000+ |
| Test duration | Unlimited (live machine) | Minutes to ~1 hour (heat limited) | Unlimited |
| Initial cost (relative) | Moderate | Low | High |
| Lifespan | 50+ years with bushing service | Decades, brake-band wears | 10-20 years, electronics dependent |
| Best application fit | Line-shaft and belt-drive audits | Engine bench testing, bursty loads | Production test cells, datalogging |
| Setup complexity | Moderate — must break belt run | Low — clamp onto shaft | High — install in driveline + DAQ |
Frequently Asked Questions About Tatham's Dynamometer
Once belt tension and pulley runout are ruled out, the next suspect is lever-arm flex under the deadweight. A 1200 mm lever loaded near its end with 200 N can deflect 1-2 mm if the section modulus is undersized, which tilts the weight slightly forward of true horizontal and biases the reading low.
Check by clamping a dial indicator under the loaded end of the lever and noting deflection at full load. If it exceeds 0.5 mm, either reinforce the beam or recalibrate with the lever loaded statically against a known weight to capture the flex offset.
The diameter difference sets the sensitivity. A small difference — say 5 mm on 300 mm pulleys — gives high sensitivity (large lever weight movement per unit power) but amplifies any concentricity error. A large difference, like 20 mm, is forgiving on machining tolerances but compresses the lever scale and hurts low-load resolution.
For shaft powers in the 5-15 hp range at 300-450 RPM, a difference of 8-12 mm on roughly 300 mm pulleys is the practical sweet spot. Below 1 hp consider 5-6 mm; above 30 hp open up to 15-20 mm to keep belt-slip torque manageable.
You can, but the math changes and so does the error budget. V-belts wedge into the pulley grooves, so the effective contact diameter shifts with tension and wear. The differential pulley assumption — that both belt sides ride on a clean machined diameter — breaks down. Expect accuracy to fall from ±1.5% to roughly ±4-5% even with careful setup.
If the original drive is V-belt and you must measure it, the cleaner approach is to fit matched flat-belt pulleys for the duration of the test, or step up to a strain-gauge torque transducer in the shaft itself.
Steady-state lever oscillation almost always traces back to belt-tension surge from a non-circular pulley or an out-of-round belt joint. Each rotation, the belt joint passes through the wrap arc and momentarily spikes the tension, which the differential pulley dutifully transmits to the lever as a sinusoidal torque pulse.
Diagnose by counting the oscillation frequency. If it matches shaft RPM, suspect pulley runout; if it matches belt-circuit frequency (much slower), the belt joint is the culprit. A properly cemented or laced belt joint should produce under ±3 mm of lever movement at typical load.
This is a classic tell for an asymmetry in the spindle bearings or lever pivot. Forward and reverse should produce identical magnitudes — if they don't, friction torque in the spindle is adding to the reading in one direction and subtracting in the other.
Quantify the asymmetry: half the difference between forward and reverse readings equals the spindle friction torque. If it exceeds 2% of the working torque, repack or replace the bushings. Bronze bushings older than about 20 years of mill service are almost always past their useful clearance band.
For motor sizing, ±2-3% is plenty. Motor frame sizes step in roughly 25% increments, so a Tatham reading inside ±3% will always pick the right frame. The bigger question is whether the measurement captures peak demand, not average — line shafts feeding looms or rolling mills have load factors as low as 0.4 with short peaks 2-3× the average.
Run the dynamometer through a full production cycle and watch for the highest sustained reading over 10-20 seconds. Size the motor for that peak with a service factor of 1.15, not for the time-averaged power. This is exactly how Lancashire mill engineers spec'd motors during the 1925-1935 conversion programmes.
Nine times out of ten this is a unit-conversion error in the formula — specifically forgetting that N is in RPM and needs the /60 to convert to rev/s before multiplying by 2π. The other common mistake is using pulley radius instead of diameter, or vice versa, halfway through the calculation.
Sanity-check by computing the belt speed: v = π × Dmean × N / 60, which for a 0.30 m pulley at 320 RPM should land near 5 m/s. If your belt speed is 0.08 m/s or 300 m/s, you've found the missing factor of 60.
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.
