Different uses of wattmeters
A practical guide to measuring the power of a three-phase installation or machine: how a wattmeter works, how many are needed in each case, how its four terminals are connected and how to interpret the readings. Each method includes a worked example.
In alternating current, the power drawn by a load breaks down into three quantities related to each other by the power triangle:
In a balanced three-phase system, with line voltage \(V_L\) and line current \(I_L\):
Measuring \(V\), \(I\) and \(\cos\varphi\) separately and applying these formulas is only valid if the system is balanced. A wattmeter, on the other hand, measures active power directly, and by combining several wattmeters any three-phase system can be measured, balanced or not. Choosing how many wattmeters and how to connect them is the purpose of this guide.
All the examples use a 400 V line-to-line supply (\(400/\sqrt{3} \approx 230.9\) V line to neutral) with positive phase sequence L1 → L2 → L3.
A three-phase motor at 400 V draws 20 A at \(\cos\varphi = 0.8\):
With the methods in this guide we will obtain these same values from the wattmeter readings.
A wattmeter has two measuring circuits and therefore four terminals:
In the classic electrodynamic wattmeter, the torque that moves the pointer is proportional to the instantaneous product of current and voltage, and the inertia of the system makes the pointer show its average value. Digital wattmeters and power analysers do the same numerically. In both cases the reading is:
where \(V\) is the voltage between the top and bottom terminals, \(I\) the current through the side terminals and \(\widehat{V,I}\) the angle between them. The key idea is that a wattmeter does not measure "the cos φ of the load", but the cosine of the angle between the particular voltage and current we connect to it. If the voltage coil receives a voltage different from that of the phase whose current it measures, the reading is no longer the power of that phase. That is why, in some methods, a wattmeter can show a negative value.
One terminal of each coil is marked with * (or ±) to indicate polarity. For the reading to be positive when energy flows towards the load, the current must enter through the marked side terminal, and the marked voltage terminal (the top one in the diagrams of this guide) must be connected to the same line as the current coil, except when measuring reactive power, as we will see.
Line Voltage coil connection ● Electrical junction
A wattmeter in a 230 V single-phase circuit, with a load drawing 10 A at \(\cos\varphi = 0.8\), will read:
If one of the two coils were reversed, the pointer would try to swing to the negative side: that coil would have to be reversed and the reading recorded with a minus sign.
If all three phases carry the same load and the neutral is available, the three phase powers are equal and one wattmeter is enough: current terminals in series with one phase, top voltage terminal (*) to that same phase and bottom terminal to the neutral. It then measures the phase voltage \(V_F\) and the current of that phase.
A balanced star load at 400 V with neutral draws 20 A per phase at \(\cos\varphi = 0.85\):
Check: \(\sqrt{3}\cdot 400 \cdot 20 \cdot 0.85 = 11\,778\ \text{W}\).
Many three-phase loads have no accessible neutral: delta-connected motors, star loads with an isolated star point or three-wire lines. If the load is balanced, a single wattmeter can still be used by creating an artificial neutral N′: the bottom voltage terminal is taken to a common point to which the other two phases are also connected, each through a resistor.
For N′ to be at the same potential as a real neutral, the two resistors must have the same impedance as the voltage coil (see the wattmeter manual). This forms a balanced star and the wattmeter receives exactly the phase voltage.
A delta-connected motor at 400 V draws 15 A line current at \(\cos\varphi = 0.82\). With the artificial neutral:
Check: \(\sqrt{3}\cdot 400 \cdot 15 \cdot 0.82 = 8\,522\ \text{W}\).
In practice, if resistors of the right value are not at hand, the two-wattmeter method (section 7) is more convenient, as it needs neither a neutral nor auxiliary resistors.
When each phase has a different load (the typical case of a home, a building or a panel with single-phase loads spread across the phases) and a neutral is available, one wattmeter is placed on each phase: current terminals in series with its phase, top terminal (*) to that phase and bottom terminal to the neutral. Each one measures the real power of its phase.
A panel supplied at 230 V line to neutral has these loads:
Here it would be wrong to multiply a single reading by three: depending on the phase chosen, you would get 6,900 W, 8,280 W or 2,070 W.
On a three-wire line with an unbalanced load (for example, three different loads connected in delta), three wattmeters can also be used. As there is no neutral, their bottom voltage terminals are joined at a common point N′.
The sum is correct whatever the potential of the common point. In a three-wire system the line currents add up to zero, so if N′ shifts by a voltage \(V_0\), the total error is \(-\mathrm{Re}\big[V_0\,(I_1 + I_2 + I_3)^*\big] = 0\). What is no longer true is that each individual reading is the power of one phase: the individual readings change depending on where N′ ends up, and only the sum is meaningful.
Three delta-connected loads at 400 V: \(Z_{12} = 40\ \Omega\) (resistive), \(Z_{23} = 20 + j15\ \Omega\) (resistive-inductive, \(|Z| = 25\ \Omega\)) and \(Z_{31} = 50\ \Omega\) (resistive). Their actual power is:
Total: 12,320 W. With three identical wattmeters, whose coils form a balanced artificial neutral:
If the coils were not identical and the common point shifted, the readings could be, for example, 2,791 W, 7,116 W and 2,414 W: individually different, but their sum is still 12,320 W.
This property is Blondel's theorem: in an \(n\)-wire system, total power is measured with \(n - 1\) wattmeters. If the common point is taken to one of the lines, the wattmeter on that line has both voltage terminals at the same potential, reads zero and can be removed. This gives rise to the two-wattmeter method.
This is the most widely used method in three-wire systems. The current terminals of \(W_1\) are in series with L1 and those of \(W_2\) with L2. The top terminals (*) are connected to their own line and both bottom terminals to L3, the line without a wattmeter. So \(W_1\) measures the voltage \(V_{13}\) and \(W_2\) the voltage \(V_{23}\).
With a balanced load, the phase shift between the line voltage applied to each wattmeter and its current is \(30^\circ - \varphi\) in one and \(30^\circ + \varphi\) in the other. For an inductive load and positive sequence:
With a balanced load, the two readings also give the reactive power:
In fact, in the formula above the sign of (\(W_1 - W_2\)) depends on both the phase sequence and the nature of the load.
| \(W_1 - W_2\) | POSITIVE SEQUENCE | NEGATIVE SEQUENCE |
|---|---|---|
| Inductive load | \(\gt 0\) | \(\lt 0\) |
| Capacitive load | \(\lt 0\) | \(\gt 0\) |
Motor at 400 V, 20 A, \(\cos\varphi = 0.8\) \((\varphi = 36.87^\circ)\):
These match the values calculated in the introduction.
Look at \(W_2 = V_L I_L \cos(30^\circ + \varphi)\). Depending on the power factor:
| \(\cos\varphi\) | \(\varphi\) | Readings |
|---|---|---|
| 1 | 0° | \(W_1 = W_2\) (each one half of \(P\)) |
| between 1 and 0.5 | 0° – 60° | Both positive, \(W_1 > W_2\) |
| 0.5 | 60° | \(W_2 = 0\): all the power is read by \(W_1\) |
| less than 0.5 | over 60° | \(W_2\) negative |
When an analogue wattmeter tries to read negative, the pointer hits the left stop. Swap its voltage terminals (or its current terminals, never both), take the reading and subtract it instead of adding it. Digital wattmeters and power analysers show the minus sign directly.
The same type of motor running at no load draws 8 A at \(\cos\varphi = 0.3\) \((\varphi = 72.5^\circ)\):
The second wattmeter reads backwards. After swapping its voltage terminals it reads 695 W, and:
Check: \(\sqrt{3}\cdot 400 \cdot 8 \cdot 0.3 = 1\,663\ \text{W}\). That is why a negative reading almost always appears in motor no-load tests.
The Aron method is valid for any three-wire load, balanced or not, in star without neutral or in delta. It follows directly from Blondel's theorem in section 6: the sum is still the total active power. The connection is the same (figure in section 7).
This equality always holds in three-wire systems.
However, the formulas for \(Q\) and \(\tan\varphi\) in the previous section only apply to balanced loads: with an unbalanced load, the system does not even have a single \(\cos\varphi\).
With the same unbalanced load as in section 6 (40 Ω, 20 + j15 Ω and 50 Ω in delta, which actually draws 12,320 W and 3,840 var), the two wattmeters read:
The active power is correct and the reactive power is not. Note also that \(W_2 > W_1\), which is impossible with a balanced inductive load: a good warning sign that the load is unbalanced.
The method is not valid in four-wire systems with current in the neutral: part of the energy returns through the neutral and the two wattmeters do not see it. In that case three wattmeters must be used (section 5).
A wattmeter measures \(V I \cos\big(\widehat{V,I}\big)\). If a voltage shifted 90° from the phase voltage is applied across its voltage terminals, it measures \(V I \sin\varphi\), which is proportional to the reactive power. In a three-phase system, the line voltage between the other two phases is exactly in quadrature with the phase voltage: you just need to change where the top and bottom terminals go.
Current terminals in series with L1; top voltage terminal (*) to L2 and bottom terminal to L3:
Motor at 400 V, 20 A, \(\cos\varphi = 0.8\) \((\sin\varphi = 0.6)\):
The result is the same as with the Aron method.
As seen in section 7, the Aron method gives \(P\) and \(Q\) at the same time with \(Q = \sqrt{3}\,(W_1 - W_2)\). It is the most practical approach if the active powers are already being measured.
If the load is unbalanced, one wattmeter per phase is used, each with the voltage between the other two phases, in this order: \(W_1\) with \(I_1\) and \(V_{23}\) (* on L2), \(W_2\) with \(I_2\) and \(V_{31}\) (* on L3), and \(W_3\) with \(I_3\) and \(V_{12}\) (* on L1). Keeping this order and the polarities is essential for the sum to be correct.
This arrangement is valid with or without neutral and with any load imbalance, as long as the supply voltages are symmetrical, as is normally the case in distribution networks.
With the loads from section 5 (230 V line to neutral: 10 A at \(\cos\varphi = 1\), 15 A at \(\cos\varphi = 0.8\) and 5 A at \(\cos\varphi = 0.6\)), the actual reactive powers per phase are 0, 2,070 and 920 VAr, 2,990 VAr in total. The quadrature wattmeters read \(\sqrt{3}\) times the reactive power of their phase:
| System | Wattmeters and voltage terminals | Active power | Reactive power |
|---|---|---|---|
| Balanced with neutral | 1: phase (*) – neutral | \(P = 3W\) | 1 in quadrature: \(Q = \sqrt{3}\,W\) |
| Balanced without neutral | 1: phase (*) – artificial neutral, or 2 Aron | \(P = 3W\) or \(P = W_1 + W_2\) | \(Q = \sqrt{3}\,(W_1 - W_2)\) |
| Unbalanced with neutral | 3: each phase (*) – neutral | \(P = W_1 + W_2 + W_3\) | 3 in quadrature: \(Q = \sum W / \sqrt{3}\) |
| Unbalanced without neutral | 2 Aron (bottom terminals to the 3rd line) or 3 with a common point | \(P = W_1 + W_2\) | 3 in quadrature: \(Q = \sum W / \sqrt{3}\) |
Measuring power with wattmeters is still the basis for understanding modern power analysers, which internally apply these same connections. Knowing which voltage reaches the terminals of each wattmeter lets you detect wiring errors, interpret negative readings and choose the right method depending on whether there is a neutral and whether the load is balanced.
If you find a lot of reactive power, you can size the capacitor bank you need with our power factor correction calculator. To convert loads from delta to star and vice versa, use the delta–star calculator.
The examples use rounded values and a 400 V supply with positive sequence. Actual readings may vary slightly due to instrument accuracy and the consumption of its own coils.