Analog Devices AN624 Application Notes
AN-624
APPLICATION NOTE
One Technology Way • P.O. Box 9106 • Norwood, MA 02062-9106 • Tel: 781/329-4700 • Fax: 781/326-8703 • www.analog.com
Calibration of a 3-Phase Energy Meter Board on the ADE7754
By Etienne Moulin
INTRODUCTION
A 3-phase energy meter can interface with different
services like 3-phase 4-wire, 3-phase 3-wire, and
3-phase 4-wire with 2½ elements. It can also provide
active, reactive, and apparent energy cal c ulat ions as we ll
as power quality measurements, voltage rms, and current rms.
This application note describes the different steps to
calibrate a 3 -phase 4-wire, 3 - element energy meter
based on the ADE7754. It describes the sof tware used
with the ADE7754 evaluation board to perform the calibration.
The ADE7754 is comprised of six ADCs, a reference circuit, and all the signal processing necessary for the
calculation of active energy, apparent energy, and rms
value of the analog inputs. Circuitry is provided to null
out various system errors including gain, phase, and
offset errors. All registers of the ADE7754 are avail able through a 4-wire serial interface (SPI
to the ADE7754 data sheet for a detailed description
and the operation of the SPI interface.
3-PHASE ENERGY SERVICES
This section presents the different 3-phase services
as well as the different meter architectures available in
the fi eld. Each of these solutions uses a different formula
to calculate energy. The terms and descriptions of the
services are taken from the US ANSI C12.1 standard.
3-Phase 4-Wire Wye Service
This service is comprised of three phases and one neutral conductor, as described in Figure 1. Each voltage
phase is referred to the neutral and has ±120° phase
difference with the other phases (see Figure 2). The
energy measurement is done with three current sensors
and two or three voltage sensors. A meter with two
voltage sensors is generally called a 2½ -element or
2-stator 4-wire wye meter. A meter with three voltage
sensors is generally called a 3-element or 3-stator
4-wire wye meter.
®
). Please refer
PHASE A
SOURCE
Figure 1. 3-Phase 4-Wire Wye Service
PHASE B
Figure 2. 3-Phase 4-Wire Wye Phasor Diagram
3-Stator 4-Wire Wye Meter
This 3-phase meter is comprised of three voltage sen sors and three current sensors. The common point of
the voltage sensors should be connected to the neutral
conductor. The energy measurements are done based
on the measurement of the six entities involved in the
system. This method is accurate for all conditions of load
(balanced and unbalanced), power factor, or voltage.
Active Power = V I V I V I
PHASE B
PHASE C
LOAD
PHASE A
–120ⴗ+120ⴗ
NEUTRAL
PHASE C
×+ ×+ ×
AA BB CCφφ φφ φφ
REV. 0
AN-624
2-Stator 4-Wire Wye Meter
This 3-phase meter is comprised of two voltage sensors and three current sensors. The common point of
the voltage sensors should be connected to the neutral
conductor. According to Blondel’s Theorem, if the voltages between each line and the neutral are balanced
within acceptable limits, the accuracy is generally considered satisfactory. The energy measurements are
done by combining the fi ve entities (two voltages and
three currents) of the system.
3-Phase 4-Wire Delta Service
This service is comprised of three phases and one
neutral conductor (see Figure 3). The neutral conductor
is formed by a tap to the midpoint of one of the phase
windings (see Figure 4). The energy measurement is
done with three current sensors and two or three voltage sensors. A meter with two voltage sensors is
generally called a 2½-element or 2-stator 4-wire delta
meter. A meter with three voltage sensors is generally
called a 3- element or 3 - stator 4 -wire delta meter.
PHASE A
SOURCE
PHASE C
2-Stator 4-Wire Delta Meter
This 3-phase meter is comprised of two voltage sensors
and three current sensors. The common point of the
voltage sensors should be connected to the neutral conductor. If the neutral is a true midtap (voltages used to
defi ne the neutral are equal within acceptable limits),
then only two voltage sensors need be used (2-stator).
The energy measurements are done by combining the
fi ve entities (two voltages and three currents) of the
system.
3-Phase 3-Wire Delta Service
This service is comprised of three phase conductors
(see Figure 5 and 6). The energy measurements are done
with three current sensors and two or three voltage
sensors. A meter with two voltage sensors is generally
called a 2½-element or 2-stator 4-wire delta meter. A
meter with three voltage sensors is generally called
a 3-element or 3-stator 4-wire delta meter.
PHASE A
SOURCE
PHASE C
LOAD
PHASE B
LOAD
Figure 3. 3-Phase 4-Wire Delta Service
PHASE C
–90ⴗ+90ⴗ
PHASE A
NEUTRAL
PHASE B
Figure 4. 3-Phase 4-Wire Delta Phasor Diagram
3-Stator 4-Wire Delta Meter
This 3-phase meter is comprised of three voltage sen sors and three current sensors. The common point of
the voltage sensors should be connected to the neutral
conductor. The energy measurements are done based
on the measurements of the six entities involved in the
system. This method is accurate for all conditions of load
(balanced and unbalanced), power factor, or voltage.
PHASE B
Figure 5. 3-Phase 3-Wire Delta Service
PHASE A
PHASE B – PHASE A PHASE C – PHASE A
+120ⴗ
–120ⴗ
PHASE CPHASE B
Figure 6. 3-Phase 3-Wire Delta Phasor Diagram
2-Stator 3-Wire Delta Meter
This 3-phase meter is comprised of two voltage sensors
and two current sensors. The common point of the
voltage sensors should be connected to one phase
conductor. The current sensors are connected to the
other two phase conductors.
–2–
REV. 0
AN-624
CONFIGURATION OF ADE7754 FOR 3-PHASE ENERGY METERING
The ADE7754 provides several registers to confi gure
the part depending on the meter connections and the
desired results. Some registers are specifi c to the meter
connections, but others are more generic and their
values can be defi ned early in the design.
The OPMODE and GAIN registers defi ne the general
confi guration of the ADE7754.
OPMODE (Address 0x0A)
Usually, the OPMODE register can be set to 0x00. With
this value, the high-pass fi lters and low-pass fi lters are
enabled and the pulse output proportional to active
power, CF, is activated. The part can be reset to its
default confi guration by setting Bit 6 of this register
to Logic 1. The default value of the OPMODE register
after reset is 0x04. In this state, the CF pulse output is
disabled.
GAIN (Address 0x18)
This register defi nes the PGA gain setting of both current
and voltage channels, the mode of accumulation of the
active powers (arithmetic sum or sum of the absolute
values), and the application of a no -load threshold on
the individual active powers.
The MMODE and WAVMODE registers confi gure the
measurements processed by the ADE7754.
MMODE (Address 0x0B)
This register defi nes the phase input on which the period
measurement and the peak detection are made. It also
defi nes the phases used for counting the number of
zero crossings in the line accumulation mo des. Th is re g ister can be set at a default value at initialization and
changed during the meter operation.
WAVMODE (Address 0x0C)
This register defi nes the speed and the analog input
used for waveform sampling. The value of this register
can be defi ned at initialization and changed during the
meter operation to access the different ADC outputs.
This register also selects the accumulation of the reactive energy in the LAENERGY register if needed.
Interrupt Mask (Address 0x0F)
This register defi nes which event will drive the interrupt
request pin (IRQ) low. The detected events are:
• Active energy register half full
• Low voltage on any of the three voltage inputs (SAG)
• Missing zero-crossing on any of the three voltage
inputs (ZXTOUT)
• Rising zero-crossing edge on any of the three voltage inputs (ZX)
• End of accumulation of energy over the LINCYC line
cycles (LENERGY)
• High voltage on a selected voltage or current input
(PKV and PKI)
• Sample available in the waveform sampling register (WFSM)
• Apparent energy register half full
The selection of the events for interruption depends on
the functionality needed in the meter. It is recommended
to select the interrupts for half-full energy (active and
apparent) in order to avoid any information loss due to
the overfl ow of these registers. The other interrupts
should be selected depending on the need of the design.
The confi guration and operation of the SAG, ZXTOUT,
VPEAK, and IPEAK interrupts are detailed in the ADE7754
data sheet.
REV. 0
–3–
AN-624
ACTIVE POWER-ENERGY MEASUREMENTS
Theory of Operation
Electrical power is defi ned as the rate of energy fl ow
from source to load. It is given by the product of the
voltage and the current inputs. The resulting signal is
called the instantaneous power signal and is equal to
the rate of active energy fl ow at every instant of time.
The unit of power is the watt or joules /sec. Equation 3
gives an expression for the instantaneous power signal
in an ac system.
v(t) V t= 2 sin( )
i(t) I ( t)= 2 sin
ω
(1)
ω
(2)
where V = rms voltage, I = rms current.
pt v t it
() () ()
=×
pt VI–VIcos t
() ( )
= 2ω
(3)
The average power over an integral number n of line
cycles is given by Equation 4.
nT
1
P
==
nT
pt dt VI
()
∫
0
(4)
where T is the line cycle period.
P is referred to as the active or real power. Note that the
active power is equal to the dc component of the instantaneous power signal p(t) in Equation 3, i.e., VI. This
is the relationship used to calculate active power in the
ADE7754 for each phase. Figure 7 shows the active
power signal processing implemented in the ADE7754
for each phase.
Due to individual sensor characteristics, the active
power calculation needs to be calibrated to correct for
gain, phase, and offset errors for each phase independently (phase balancing).
Active Energy Accumulation
Besides the pulse output, which is used for calibration
verifi cation (see the Active Power Pulse Output section),
a solid state energy meter requires some form of display. This display should show the amount of energy
consumed in kWh (kilowatt hours). One convenient and
simple way to interface the ADE7754 to a display or
energy register is to use a microcontroller (MCU) that
reads one of the active energy registers, e.g., AENERGY
and LAENERGY. A full description of the functions
of these registers can be found in the ADE7754 data
sheet. The total active energy is accumulated in the
ADE7754 by adding the average active powers from each
phase and accumu lating them into the active energy
register (see Figure 8). The ADE7754 can be confi gured
to execute the arithmetic sum of the three active powers,
W = W
of these powers, W = |W
+ WøB + WøC, or the sum of the absolute value
øA
| + | WøB| + |WøC|.
øA
When the sum of the absolute values is selected, the
active energy from each phase is always counted positive in the total active energy. It is particularly useful
in a 3 -phase 4-wire installation where the sign of the
active power should always be the same. If the meter is
misconnected to the power lines, i.e., CT is connected in
the wrong direction, the total active energy recorded
without this solution can be reduced by two-thirds. The
sum of the absolute values assures that the active energy
recorded represents the actual active energy delivered.
In this mode, the reverse power information available
in the CFNUM register is still detecting when negative
active power is present on any of the 3-phase inputs.
To t r a n s f o r m the active energy register reading into a
usable form, a Wh/LSB conversion coeffi cient can be set.
The calibration of this Wh/LSB constant is described
later in this document.
APGAIN[11:0]
12
I
CURRENT SIGNAL – i(t)
1
V
VOLTAGE SIGNAL – v(t)
⌽ PHCAL
HPF
MULTIPLIER
INSTANTANEOUS POWER SIGNAL – p(t)
LPF2
24
28
Figure 7. Active Power Signal Processing
–4–
APOS[11:0]
12
12
WG[11:0]
ACTIVE POWER
SIGNAL – P
REV. 0
AN-624
POWER PHASE A
POWER PHASE B
POWER PHASE C
T
TOTAL
ACTIVE
POWER
AENERGY[23:0]
23
53
WDIV
53
0
0
0
Figure 8. Active Energy Accumulation
Active Power Pulse Output (CF)
The ADE7754 provides a pulsed output, CF, whose fre quency is proportional to the active power. It provides
a simple, single -wire, optically isolated interface to
external calibration equipment.
The energy-to-frequency conversion is accomplished
by accumulating the total active power in a 54-bit wide
register. An output pulse is generated each time the
register value is greater than 2
30
LSBs (see Figure 9).
The output frequency at CF, with full-scale ac signals
on all six channel inputs and CFNUM = CFDEN = 0x000,
is approximately 96 kHz. This can be calculated as follows:
with all the gain registers set to 0x000 (APGAINs,
WGAINs), the average value of the instantaneous active
power on each phase is 0xD1B717 or 13743895d. For all
three phases, the average value is 41231685d. An output
frequency is generated on CF when the internal register
accumulates 2
30
. The accumulation rate is CLKIN/4.
In the ADE7754, the CF frequency can be adjusted by
changing the different gain registers. As the ADE7754
is comprised of three independent inputs (phases), this
calibration needs to be done for each input independently. CFNUM and CFDEN are meant for global coarse
gain compensation and AWG, BWG, and CWG for fi ne
gain adjustment per phase.
Active Power Measurement
A solid state energy meter requires the display of the
active power in addition to the active energy. For 3-phase
applications, the requirement is generally for the display
of the active power per phase.
The ADE7754 does not provide a direct measurement
of the active power as defi ned in Equation 4, but instead
uses two accumulators for the active energy. Each
accumulator can be confi gured independently to hold
the active energy from a specifi c phase or the active
energy sum of several phases. One of the accumula tors (e.g., AENERGY) can be constantly used for the
regular total active energy accumulation (e.g., billing)
and the other accumulator (e.g., LAENERGY) can be
used for average active power measurement by dividing
the value read by the accumulation time.
Note: In the ADE7754, the average active power measurement per phase has to be processed one phase at a
time by changing the phase selected in the LAENERGY
accumulation. The switching between phases should
follow the descriptions of Figure 10 by changing the
MMODE and WATMODE register values.
CF Hz
()
Average Total Active Power CLKIN
=
30
×24
POWER PHASE A
POWER PHASE B
POWER PHASE C
×
TOTAL
ACTIVE
POWER
T
Figure 9. Energy-to-Frequency Conversion
(5)
CFNUM[11:0]
11
23
53
0
DFC
0
11
CFDEN[11:0]
0
CF
0
REV. 0
–5–
AN-624
ACTIVE ENERGY GAIN CALIBRATION USING THE LAENERGY REGISTER
The ADE7754 accumulates the active power synchronously to the line cycles. This mode is especially useful
for calibration purposes as the ripple effect in the active
energy accumulation is reduced to zero (see the Line
Energy Accumulation section in the ADE7754 data sheet).
In this line accumulation mode, the ADE7754 accumulates the active power signal in the LAENERGY register
for an integral number of half line cycles. The number
of half line cycles, the phases selected to be accumulated, and the phases involved in the counting are
specifi ed in the LINCYC register, the LWATSEL bits of
the WATMODE register (address 0x0D), and in the
ZXSEL bits of the MMODE register (address 0x0B),
respectively. Figure 10 describes how to set up the line
accumulation mode in the ADE7754.
SET CONFIGURATION REGISTERS
ADDR. 0x0A – OPMODE
ADDR. 0x0B – MMODE
ADDR. 0x0D – WATMODE
ADDR. 0x18 – GAIN
SET NUMBER OF LINE CYCLES
TO 200
ADDR. 0x13 = 0d200
SET INTERRUPT MASK
FOR LINE ACCUMULATION
ADDR. 0x0F = 0x0400
The number of phases sele cted is the number of ones in
the ZXSEL bits of the MMODE register.
WDIV is a register used for scaling the active energy
accumulations; it does not affect the CF frequency. It is
introduced in Equation 6 to compensate its effect on
the LAENERGY value.
When calculating the expected CF frequency with a
LAE
NERGY reading and Equation 6, the actual ADE7754
register values used during the test should be used.
The CF frequency calibration has to be done for each
phase individually. The gain correction carried by the
CFNUM and CFDEN registers affects all three phases
similarly and should be used as a coarse gain compensation. The WG registers should then be used to fi ne
adjust the CF frequency to the expected frequency for
each individual phase.
Note: If the active power is accumulated in both the
active energy and line active energy registers during
the same amount of time, the line active energy register
value is four times the active energy register value.
Wh/LSB Constant Calibration
The active energy Wh/LSB constant and the CF frequency
can be calibrated at the same time using the line accumulation mode. Equations 6 and 7 detail the relationship
between the different parameters.
Under the steady load test condition, the watt power
consumption, W, is known. The Wh/LSB constant for the
AENERGY register is estimated using Equation 8 :
RESET INTERRUPT STATUS REGISTER
ADDR. 0x11
Figure 10. Line Accumulation Mode Setup
CF Frequency Gain Calibration
There is a direct relationship between CF frequency and
the line active energy register value (see Equation 6).
CF
(Hz) =
4 Accumulation Time(s)
CFNUM
×××+
CFDEN
×
WDIV 1
LAENERGY
WG
12
2
(6)
where Accumulation Time is the period of time during
which the active power has been accumulated in the
LAENERGY register:
Accumulation Time(s)
LINCYC
LineFrequency No. of Phases Selected
××
=
[15 0]2:
(7)
Wh
LSB
W Accumulation Time(s)
cst
×
=
LAENERGY
×3600
(8)
4
It should be noted that once the CF frequency has been
adjusted for each phase to the same value, the AENERGY
and LAENERGY registers will give the same value from
part to part and phase to phase under the same condi tions. Therefore, the Wh/LSB coeffi cient is a constant
that does not need to be calibrated. It can be estimated
by design and stored as is in the MCU.
Line Period Measurement
The calibration of the CF frequency and the Wh/LSB
constant with the line accumulation mode requires an
estimation of the line frequency. A poor estimation of
this quantity leads to errors in the calibration of the
system. Some calibration systems do not provide the
line frequency. The ADE7754 provides a measurement
of the line period in the period register (address 0x07).
The selection of the voltage input is done by Bits 0 and 1
–6–
REV. 0
Loading...