A recording watt-hour meter is an electricity meter that continuously integrates real power over time and records the cumulative energy consumed in kilowatt-hours. The aluminium disc is its key component — it spins at a rate proportional to the instantaneous power flowing through the meter, with each revolution representing a fixed quantity of energy set by the meter constant Kh. The mechanism exists to give utilities and industrial users a defensible billing record. A typical Class 0.5 induction disc meter holds ±0.5% accuracy across loads from 5% to 200% of rated current.
Recording Watt-hour Meter Interactive Calculator
Vary voltage, current, power factor, time, and Kh to see real power, energy, disc RPM, and register revolutions.
Equation Used
The Ferraris disc meter converts real power into disc speed. Real power is V times I times power factor, each revolution represents Kh watt-hours, and the register energy is the accumulated revolution count.
- Single-phase real power with PF = cos(phi).
- Meter constant Kh is in watt-hours per disc revolution.
- Disc speed is proportional to real power after brake torque balance.
- Friction, creep, and calibration error are neglected.
The Recording Watt-hour Meter in Action
The classic recording watt-hour meter is a Ferraris-type induction disc meter. Two electromagnets straddle a thin aluminium disc — a voltage coil wound for many turns of fine wire connected across the line, and a current coil of few heavy turns in series with the load. The voltage coil produces a flux that lags the supply voltage by close to 90° because of its high inductance, while the current coil produces a flux in phase with the load current. These two out-of-phase fluxes induce eddy currents in the disc, and the interaction between flux and eddy currents generates a torque on the disc proportional to V × I × cos(φ) — in other words, real power. The disc would accelerate forever, so a permanent magnet brake creates a retarding torque proportional to disc speed, and the disc settles at an RPM directly proportional to power. A worm gear on the spindle drives the register dial train, which records cumulative revolutions as kWh.
Why this design? Because it integrates automatically. You don't need electronics, sampling, or a clock — the disc revolution count IS the energy integral. Get the geometry right and the meter just works for 40 years. The trade is in the tolerances. The disc-to-magnet air gap must hold to within roughly ±0.05 mm, because brake torque scales with the square of flux density. The voltage coil's lag angle must hit 90° within about 0.5° at unity power factor, otherwise the meter under-reads or over-reads on inductive loads — that's what the lag-plate adjustment screw on the side of the meter is for. Bearing friction at the spindle has to be near zero because friction creates a creep error at light loads.
Failure modes are mechanical and predictable. If you notice the disc creeping with no load connected, the light-load adjustment is misaligned or stray voltage-coil flux is unbalanced — meters have anti-creep holes drilled in the disc specifically to stall this rotation. If the meter under-reads at high current, the current-coil core is saturating or there's a shorted turn. If it over-reads on inductive loads but reads correctly on resistive ones, the lag adjustment has drifted. The brake magnet weakens over decades of service, which makes the disc spin faster than it should — utilities find these on routine accuracy testing and replace the magnet or recalibrate.
Key Components
- Aluminium Disc (Rotor): Thin aluminium disc, typically 0.5 to 1.0 mm thick and 80 to 100 mm in diameter. It carries the eddy currents induced by the coil fluxes and provides the rotating element whose revolution count represents energy. Disc thickness tolerance must hold �±0.02 mm because thickness affects eddy current distribution and braking torque.
- Voltage (Pressure) Coil: Many-turn, high-impedance coil wound on a laminated E-core, connected across the line. It produces a flux that lags the applied voltage by very close to 90°, achieved by a separate lag winding or shading loop. The lag angle must be trimmed within ±0.5° of 90° at unity power factor.
- Current Coil: Few-turn heavy-gauge coil wound on a U-core, connected in series with the load. It produces a flux essentially in phase with the load current. The core must avoid saturation up to 200% rated current — typical service current is 100 A or 200 A in a domestic Class 200 form 2S meter.
- Permanent Magnet Brake: An Alnico or ceramic permanent magnet straddles the disc opposite the coils. As the disc rotates through the magnet's field, it sees a retarding eddy-current torque proportional to disc speed. This is what makes disc RPM proportional to power rather than power's integral. Magnet strength must be stable to within ±1% over the meter's service life.
- Register Dial Train: A worm on the disc spindle drives a gear train of typically 5 dial pointers reading in kWh, tens, hundreds, thousands, and ten-thousands. The gear ratio sets the meter constant Kh — typical values are Kh = 7.2 watt-hours per disc revolution for a residential 240 V meter.
- Spindle Bearings: Top jewel bearing and a magnetic suspension bottom bearing in modern meters, replacing the older sapphire-and-pivot design. Magnetic suspension reduces starting friction so the meter accurately registers loads down to 0.5% of rated current — required for Class 0.5 accuracy ratings.
- Lag Plate / Light-Load Adjustment: A small adjustable copper or steel plate near the voltage coil, used to trim the 90° flux lag. Combined with a separate light-load screw and a creep-prevention disc hole, these set the meter's calibration at full-load unity PF, full-load 0.5 PF lagging, and 5% load.
Real-World Applications of the Recording Watt-hour Meter
Recording watt-hour meters are the foundation of electricity billing — every house, every shop, every factory feeder has one. They show up wherever someone needs a defensible, integrated record of energy used, not just an instantaneous power reading. Even with smart meters now dominant, induction disc meters are still in service in millions of installations and are still installed where simple, electronics-free reliability matters more than remote reading.
- Electric utility billing: Form 2S single-phase induction disc meter on residential service drops — General Electric I-70-S and Westinghouse D5S meters in service across North American utilities since the 1950s, with Kh = 7.2 Wh/rev and Class 200 amp rating.
- Commercial sub-metering: Tenant sub-metering in shopping centres where each unit is billed for its share of total energy — a Landis+Gyr MX series 3-phase 4-wire meter on a strip-mall feeder records each tenant's kWh independently of the master utility meter.
- Industrial process plants: Feeder metering on motor control centres at a large bakery's mixing line — one meter per MCC bus tracks energy per production shift, used for cost-per-loaf accounting.
- Renewable generation: Bidirectional ratchet-equipped recording meters on small grid-tie solar installations — measure import and export separately on dual registers for net-metering tariffs in jurisdictions that still use mechanical meters.
- Rail and transit substations: Recording watt-hour meters on traction substation feeders, measuring energy delivered to overhead catenary or third-rail sections for cost allocation between transit operator and grid supplier.
- Marine shore-power: Recording meters on cruise ship and cargo vessel shore-power connections at the Port of Long Beach — tracks kWh delivered to docked vessels for billing under cold-ironing tariffs.
The Formula Behind the Recording Watt-hour Meter
The core relationship in a recording watt-hour meter is between disc speed and real power, and the integral of disc revolutions over time gives energy. The meter constant Kh ties revolutions to energy. At the low end of the typical operating range — a 5 W standby load on a residential meter — the disc creeps at well under 1 revolution per minute, and any bearing friction or stray flux dominates the reading. At nominal full load — say 5 kW on a residential service — the disc spins at a clearly visible 11 to 12 RPM. At the high end, 200% of rated load, the disc approaches 25 RPM and you start seeing aerodynamic drag on the disc edges add a small non-linearity. The sweet spot for billing accuracy sits between roughly 10% and 100% of rated current.
Variables
| Symbol | Meaning | Unit (SI) | Unit (Imperial) |
|---|---|---|---|
| E | Energy recorded by the meter | kWh | kWh |
| Nrev | Total disc revolutions over the measurement period | revolutions | revolutions |
| Kh | Meter constant — watt-hours per disc revolution, stamped on the meter nameplate | Wh/rev | Wh/rev |
| P | Real power flowing through the meter | W | W (or hp × 745.7) |
| Nrpm | Disc rotational speed at power P | RPM | RPM |
Worked Example: Recording Watt-hour Meter in an industrial bakery MCC feeder
A regional bread bakery in Hamilton Ontario sub-meters its dough-mixing motor control centre with a 240 V single-phase Form 2S General Electric I-70-S induction disc meter, Kh = 7.2 Wh per revolution, Class 200. The plant engineer wants to know what disc RPM to expect during a normal 5 kW mixing run, what to expect during the 500 W idle period between batches, and what to expect during the 15 kW peak when both the spiral mixer and the divider are running together. He also wants to verify the meter is reading correctly by counting revolutions with a stopwatch.
Given
- Kh = 7.2 Wh/rev
- Pnominal = 5000 W
- Pidle = 500 W
- Ppeak = 15000 W
- tcount = 60 s
Solution
Step 1 — at the nominal 5 kW mixing load, calculate disc RPM from the meter constant. One revolution equals 7.2 watt-hours, which is 7.2 × 3600 = 25,920 watt-seconds (joules) of energy.
Step 2 — convert that to revolutions per minute the engineer can count with a stopwatch over a 60-second window:
So during a normal mixing run he should see the disc complete a hair under 12 full turns in one minute. That's a comfortable counting rate — slow enough to mark the black spot accurately, fast enough that timing error stays small.
Step 3 — at the low end, the 500 W idle load between batches:
That's about 52 seconds per revolution. Visible motion, but barely — and at this load the meter's bearing friction and any residual creep starts to matter. A Class 2 meter is only specified to ±2% accuracy below 10% of rated current, so don't trust idle-load energy figures to better than a few percent.
Step 4 — at the high end, the 15 kW combined peak when mixer and divider run together:
The disc is now spinning faster than once every two seconds. You can no longer count individual revolutions reliably by eye — at this point use the LED test pulse output if the meter has one, or count over a longer window. The disc is well within its mechanical limit (the brake magnet is sized for around 200% rated load, which on a Class 200 meter at 240 V is roughly 96 kW), but you're approaching the upper end of the linear-response range.
Result
At the nominal 5 kW mixing load the disc spins at 11. 57 RPM, or just under 12 revolutions per minute — easily countable with a stopwatch and a black calibration spot. Compare to 1.16 RPM at the 500 W idle load (one slow revolution every 52 seconds) and 34.7 RPM at the 15 kW peak (faster than two revolutions per second), and you can see the meter operates over a roughly 30:1 dynamic range, with best accuracy in the middle of that band. If the engineer counts only 10.5 revolutions in 60 seconds at a verified 5 kW load instead of the predicted 11.57, three failure modes are most likely: (1) the brake magnet has weakened with age, raising disc speed for a given power — wait, that would cause over-reading, so suspect the opposite for under-reading; (2) the current coil core is partially saturated due to a DC offset on the load (common with half-wave SCR drives on older bakery equipment) which reduces working flux; or (3) the worm-to-register gear has slipped a tooth and the dial registers correctly but the visual revolution count is off because the calibration spot is referenced to a misaligned dial.
Choosing the Recording Watt-hour Meter: Pros and Cons
The recording watt-hour meter competes today with solid-state smart meters and with current-transformer-plus-instrument arrangements. Each has a real place. Choose based on accuracy requirement, remote-read need, lifespan expectation, and what happens when the electronics fail.
| Property | Induction Disc Watt-hour Meter | Solid-State Smart Meter | CT + Power Quality Analyzer |
|---|---|---|---|
| Billing accuracy class | Class 0.5 to Class 2 (±0.5% to ±2%) | Class 0.2 to Class 0.5 (±0.2% to ±0.5%) | Class 0.1 possible with revenue-grade CT (±0.1%) |
| Service lifespan | 30 to 50 years typical | 10 to 20 years (electrolytic capacitor and EEPROM limited) | 5 to 15 years for the analyzer; CTs last 30+ years |
| Accuracy at light load (<5% rated) | Poor — creep-prone, ±2% at best | Excellent — flat ±0.5% down to 0.4% rated | Excellent — limited only by CT phase error |
| Remote read / interval data | No (or pulse output add-on only) | Yes — 15-min interval data, AMI radio standard | Yes — full waveform capture, harmonics, sag/swell |
| Installed cost (residential) | $30 to $80 USD | $100 to $200 USD | $500 to $3000 USD |
| Power factor and harmonic measurement | Real power only, accurate on sinusoidal loads | True RMS, handles harmonics, measures kVAR and PF | Full harmonic spectrum to 50th order, IEEE 519 compliance |
| Failure mode | Slow drift, mechanical wear — fails gracefully | Abrupt failure, blank display — total loss of reading | Software/firmware lockup, often recoverable |
| EMP / surge tolerance | Excellent — passive electromagnetic device | Poor without surge protection | Poor — sensitive electronics |
Frequently Asked Questions About Recording Watt-hour Meter
That's a lag-angle calibration issue. The voltage coil is supposed to produce a flux that lags the applied voltage by exactly 90°. When the load is purely resistive, current and voltage are in phase, and a small lag-angle error barely matters. When the load is inductive at, say, 0.7 PF, the meter's response curve becomes sensitive to that 90° accuracy — every 1° of lag-angle error translates to roughly 1% reading error at 0.7 PF lagging.
Open the meter cover (utility seal permitting) and look for the lag plate adjustment — usually a small screw or movable copper shading band near the voltage coil pole. Calibration shops set this on a phantom-load bench with a power-factor source. If the meter is in revenue service, call the utility and request an accuracy test rather than touching it.
Probably within a few percent, but not better than that. An induction disc meter responds to fundamental-frequency real power. A 6-pulse VFD draws current with significant 5th and 7th harmonic content, and the meter's current coil sees that current but the meter's torque response to non-sinusoidal current is non-linear. Typical errors run 1% to 4% under-reading on heavily distorted loads.
If billing accuracy matters on a VFD-heavy feeder, switch to a solid-state meter rated for true RMS measurement on distorted waveforms — ANSI C12.20 Class 0.2 electronic meters handle harmonic content properly. Disc meters were specified for 60 Hz sinusoidal service and that's where they belong.
Yes, but you don't usually choose Kh directly — you choose the meter form, voltage class, and current class, and Kh falls out of the design. The practical decision is: pick a meter whose rated current sits between roughly 25% and 100% of your expected peak load. A 200 A meter on a 20 A average feeder will creep, drift, and read poorly at light load. A 30 A meter on a 100 A feeder will saturate the current coil core and read low.
Rule of thumb: expected continuous load should be between Kh × 50 and Kh × 500 watts for the meter to operate in its sweet spot. For Kh = 7.2 Wh/rev, that means 360 W to 3600 W of normal load gives best accuracy. Outside that band you sacrifice 1% to 2% on the reading.
If the disc is genuinely turning with no load, that's called creep, and it's a calibration fault. Every induction disc meter has two small holes drilled diametrically opposite in the disc — when the disc rotates, one of those holes eventually lines up with the voltage-coil pole face and the disc stalls. If yours doesn't stall, either the disc has been replaced with one missing the anti-creep holes, the light-load screw is set too aggressively (overcompensating bearing friction), or the voltage-coil flux is unbalanced and producing a residual driving torque even with no current.
Diagnostic check: disconnect the load-side conductors entirely (not just open breakers — the meter's own voltage coil is still energised through the line side). The disc should come to rest within one or two revolutions. If it keeps turning past that, the meter needs to be pulled and recalibrated.
For a critical industrial feeder where billing disputes are expensive — yes, keep it in series as a check meter. Disc meters fail gracefully (slow drift over years) while smart meters fail abruptly (blank display, lost interval data, firmware lockup). Having a mechanical record alongside an electronic one has settled more than one utility billing dispute, particularly during AMI rollout problems where smart-meter firmware bugs caused systematic over-reading.
For residential or small commercial service the cost and panel space don't justify it. The smart meter is accurate enough, and disputes go through the utility's own check-meter program.
Two parts of the meter can disagree: the disc itself (the integrating element) and the register dial train (the display). If revolution count is correct but dial reading is wrong, the gear train between the worm on the spindle and the dial pointers has either slipped, has a damaged gear tooth, or was assembled with the wrong gear ratio for the meter's stated Kh.
Check the gear ratio printed inside the meter cover against the nameplate Kh. Mismatched register assemblies sometimes get installed during refurbishment — a register from a 240 V meter fitted to a 120 V meter base will read exactly half of true energy. Pull the register and verify the ratio stamp matches the meter's nameplate constant.
Because the brake magnet weakens over time. Alnico and ceramic permanent magnets lose roughly 0.1% to 0.5% of their field strength per decade depending on temperature exposure and mechanical shock. A weakened brake magnet allows the disc to spin faster for the same power, which makes the meter over-read in the customer's favour — or under-read in the utility's favour, depending on direction of drift.
Periodic in-shop testing on a phantom load bench at 100% PF, 50% PF lagging, and 10% load catches the drift before it exceeds the regulatory tolerance (typically ±2% in most North American jurisdictions). Meters that drift outside spec get the brake magnet re-magnetised or replaced, then go back into service for another decade.
References & Further Reading
- Wikipedia contributors. Electricity meter. Wikipedia
Building or designing a mechanism like this?
Explore the precision-engineered motion control hardware used by mechanical engineers, makers, and product designers.
