Analog Devices AN639 Application Notes
AN-639
ACTIVE POWER
REACTIVE POWER
APPARENT POWER
APPLICATION NOTE
One Technology Way • P.O. Box 9106 • Norwood, MA 02062-9106 • Tel: 781/329-4700 • Fax: 781/326-8703 • www.analog.com
Frequently Asked Questions (FAQs)
Analog Devices Energy (ADE) Products
By Rachel Kaplan
GENERAL
How do I get samples of a preliminary product and evaluation
board?
Request samples of prereleased products (products with
preliminary data sheets) through your local distributor or
sales representative. For our Sales and Distributors listing, go to www.analog.com/salesdir/continent.asp. Please
be sure to tell them that the product is prereleased. The
preliminary data sheet, if available, should include the
evaluation board part number in the Ordering Guide
section.
METERING
Why are electronic meters (solid-state meters) better than
electromechanical meters or analog electronic meters?
Electronic meters have high accuracy over a wide current dynamic range, are able to handle higher currents,
have low power consumption, are reliable and robust
(stable over time and temperature), and don’t have gears
that wear out or magnets that saturate with dc current.
They do not require precision mechanics or have large
tolerance variations over temperature. Electronic meters
more easily enable new functionalities such as automatic
meter reading (AMR), multitariff billing, tamper proofing, prepayment meters, load shedding, power outage
detection, and power factor detection. Electronic meters
offer fl exibility of design, and can easily be reconfi gured
and updated (e.g., software update). They have easy and
stable calibration without hardware adjustment and are
simpler to manufacture, transport, and install. Electronic
meters offer utilities a wider supply base of manufacturers—and the competitive environment helps keep the
cost of this solution down.
What is the life span of a solid-state meter?
The ADE ICs have been tested using an accelerated
life test. The results proved the ADE performance to
be accurate and reliable for 60 years. The life span of
the meter can be affected by the meter’s design and
component selection. The reference design described
in the AN-559 and AN-563 Application Notes provide a
proven meter solution, which is a good starting point
for designing a solid-state energy meter.
What is the difference between active, reactive and apparent energy?
Active energy is measured in kilowatt-hours, while reactive and apparent energy are VAR hours and VA hours,
respectively. Figure 1 shows the relationship between
active, reactive, and apparent energy. The relationship
described in the fi gure holds true for pure sinusoids at
the fundamental frequency. In the presence of harmonics, this relationship is not valid. See also the FAQ:
is power factor?
Figure 1. Power Triangle
The relationships are as follows:
Active Power = VI cos
Reactive Power = VI sin
Apparent Power = VI
Power Factor = cos
What
TABLE OF CONTENTS
CURRENT SENSORS . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
ALL ADE PRODUCTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
ADE775x: SPI
ADE775x: ANALOG CALIBRATION PRODUCTS . . . . . . 7
SINGLE-PHASE PRODUCTS . . . . . . . . . . . . . . . . . . . . . . 7
ADE7751/ADE7755 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
ADE7757 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
ADE7753 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
ADE7756 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
ADE7759 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
POLYPHASE PRODUCTS . . . . . . . . . . . . . . . . . . . . . . . . . 9
ADE7752 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .10
ADE7754 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .10
®
INTERFACE PRODUCTS . . . . . . . . . . . . 5
REV. 0
AN-639
What is power factor?
The quantity cos() is known as power factor, where
is the angle between active and apparent power vectors
(and no harmonics are present). Power factor is, therefore,
the ratio between real and apparent power. See Figure 1.
Some utilities will charge a penalty for low power factor.
Common causes of low power factor are induction motors
and transformers. Reactance is introduced onto the line
when current is displaced or shifted out of phase with the
voltage by an angle .
CURRENT SENSORS
What current sensor should I use with ADE ICs?
Analog Devices does not currently partner with or
recommend any current sensor manufacturer. Meter
manufacturers and other customers must perform their
own evaluation and selection of current sensors. If the full
dynamic range of ADE77xx performance is desired, then
care should be taken to use current sensors that have the
desired accuracy over this range.
What are the benefi ts and drawbacks of the different current
sensor technologies?
Sensor Benefi ts Drawbacks
Low Resistance Very low cost, Poor high current
Shunt good linearity capability, dc offset,
parasitic inductance
Current High current Hysteresis/saturation
Transformer performance, due to dc, phase shift
low power
consumption
Hall Effect High current Hysteresis/saturation,
Sensor performance, higher cost,
wide dynamic temperature drift
range
Rogowski Coil Low cost, no Output is derivative
(Air-Core CT) saturation limit, of voltage signal—
low power requires an analog
consumption, (or digital) integrator.
immunity to EMI sensitivity.
dc offset, wide
dynamic range,
very low tem perature range
The ADE7753 and ADE7759 have a built-in digital integrator
for easy interface with a Rogowski coil. In all cases, the
integrator can be turned off to interface with a current
transformer (CT) or shunt.
What are the considerations for selecting the shunt?
The following are several main considerations for shunt
selection:
Power consumption requirement: According to IEC 61036,
the power consumption per channel cannot exceed
2 W. Larger shunts consume more power.
Thermal management consideration: For a large (high
resistance) shunt, there will be signifi cant temperature
rise if the current is large.
Shunt quality: The self-heating of the shunt can increase
its resistance. The output signal can vary because of this,
and affects the accuracy of the meter.
Ta mpering consideration: The resistance of the shunt
should be as close to a wire as possible to minimize
the effect of any attempt to divert the current using an
external wire.
The shunt should provide reasonable signal levels to the
IC over the current operation range.
What are the considerations for selecting a current transformer (CT)?
Care should be taken to ensure that the dynamic range
for current sensing with a given CT is large enough for
the application. Current transformers can saturate under
large dc or high current, and designers should choose
CTs rated for their needs. CTs can introduce phase shift
and should be chosen according to the designer’s ability
to compensate for this error.
How do I compensate for the phase shift of my current sensor (or phase mismatch between channels)?
For products such as ADE7751 and ADE7755 (singlephase) or ADE7752 (3-phase), the only way to
compensate for phase mismatch is by hardware. The
phase mismatch at line frequency can be corrected by
adjusting the corner frequency of the RC fi lter (used for
antialiasing on the input channels) to create a phase shift
to offset the phase error from the CT. Application Note
AN- 563 has some detailed information about how to
adjust the phase mismatch.
For products like the ADE7753, ADE7756, ADE7759
(single-phase), or ADE7754 (3-phase), you can use the
internal PHCAL register to adjust the phase lead/lag.
Adjusting the phase mismatch is a simple procedure of
writing to the register. Refer to the data sheet of the
respective product for details. If the compensation range
is beyond that of the PHCAL register, a combination of
both hardware and software phase adjustment can be
used. For example, you can use the hardware method to
roughly compensate the default phase mismatch and use
the PHCAL register as a fi ne adjustment in production.
How do I calculate the burden resistor to use with my current transformer?
The burden resistor depends on the maximum current
(I
), the input level to the ADC (y), and the number of
MAX
turns in the CT being used (CTRN). At maximum current,
the input signal at the current channel should be at half
input full scale* to allow headroom.
Full Scale
y
== =
2
176.8
mV
peak
rms
, 500 mV
=
*Full Scale is 660 mV
Refer to the product data sheet for specifi cation.
500
peak
mV
2
, or 1 V
353.55
peak
, depending on the product.
peak
–2–
2
mV
rms
REV. 0
AN-639
The following equations apply:
I
MAX
CTRN
Solve for RB. For example, if I
How does the Rogowski coil work?
The basic operating principle of a Rogowski coil is to
measure the primary current through mutual inductance.
When current passes through a conductor, a magnetic
fi eld forms around the conductor. The magnitude of the
magnetic fi eld is directly proportional to the current. The
changes in the magnetic fi eld induce an electromotive
force (EMF) within a wire loop. The EMF is a voltage
signal and is proportional to the changes in the magnetic
fi eld inside the loop. The output voltage of the loop is,
therefore, proportional to the time differentiation (di/dt)
of the current.
A Rogowski coil is typically made with an air core, so, in
theory, there is no hysteresis, saturation, or nonlinearity.
Because the Rogowski coil relies on measuring magnetic
fi eld, it makes this type of current sensor more susceptible
to external magnetic fi eld interference than the CT.
Details and equations can be found on the ADI website in
the technical article entitled “Current Sensing for Energy
Metering.”
Where can I fi nd a Rogowski coil?
Currently, the ADE product development group is looking
for an appropriate open-market Rogowski coil manufacturer. We will notify interested customers of our fi ndings.
If you would like this notifi cation, send your contact information and request to energy.meter@analog.com. Meter
manufacturers working with their proprietary sensors and
ADE products are very happy with the performance of
ADI’s digital integrator and sensor interface.
How can I use one CT or Rogowski coil in a single-phase, 3-wire
confi guration (ANSI 2S)?
In the United States (and some other locations), residential
power is distributed in a single-phase, 3-wire confi guration. Two wires, namely L1 and L2, have voltage signals
that are 180º out of phase with each other and share a
common neutral wire. In theory, two energy measurement
ICs and two sensors are required. However, an approximation method (which is generally very close to the actual
situation) can be used such that only one measurement
IC and one current sensor is needed. The assumption in
this case is that the amplitude of the two phase wires is
the same (they are 180° out of phase). One can simply use
the voltage difference between L1 and L2 and multiply by
==2
y
x
MAX
R
B
x
is 113.1 A rms, RB = 4.5 .
the sum of the currents in L1 and “reverse” of L2. Here’s
the math:
Instantaneous Power on L1 = V1N Current L1
Instantaneous Power on L2 = V2N Current L2
Instantaneous Total Power = Power on L1 + Power on L2
= V1N Current L1 + V2N Current L2
Assuming V1N = V2N = (V1N – V2N)/2
Instantaneous Total Power
= V1N Current L1 + V2N Current L2
= V1N Current L1 – V1N Current L2
= ((V1N – V2N)/2) (Current L1 – Current L2)
The divide by 2 factor is compensated for in the calibration process.
In the AN-564 Application Note, the CT is used for summing the current properly. Use one CT with both L1 and
L2 passing through in opposite directions to generate
the sum of the two currents, or use two CTs to monitor
individual phase currents and sum them externally (by
connecting the two in parallel). Take care when using a
single CT for the summation; the CT needs to be able
to handle the total current in both phases. For example,
if each phase wire has a maximum of 100 A, the CT needs
to have a 200 A capability.
ALL ADE PRODUCTS
Which metering standards do ADE products meet?
ADE ICs’ performance meets the IEC 1036, IEC 61036,
ANSI, and other derived specifi cations. Please refer to
individual product specifi cations found on product data
sheets for details or confi rmation of compliance with
other metering specifi cations.
Can ADE ICs be used in both 50 Hz and 60 Hz environments?
Yes. ADE ICs’ performance over frequency (45 Hz to 70 Hz)
can be seen in the Typical Performance Characteristics
section of each data sheet.
Do I have to use the recommended CLKIN frequency?
ADI performs extensive testing using the recommended
CLKIN frequency. The specifi ed CLKIN frequency is the
only frequency for which the product specifi cations and
part performance are guaranteed. Changing the CLKIN
frequency from 3.5 MHz (ADE7751, ADE7753, ADE7755,
ADE7756, ADE7759) or 10 MHz (ADE7752, ADE7754) will
change the constants given in the data sheet equations, as
well as register resolutions, CF, F1, and F2 pulsewidths.
Can I use a 3.3 V digital supply with ADE775x?
No, the digital supply is not supposed to work at 3.3 V.
How do I interpret the sign of the reactive or active power in the ADE ICs?
Figure 2 demonstrates how to interpret the sign of the
energy registers.
REV. 0
–3–
AN-639
ACTIVE (–)
REACTIVE (–)
ACTIVE (–)
REACTIVE (+)
ACTIVE (+)
REACTIVE (–)
I
60ⴗ = ;PF=–0.5
–60ⴗ = ;PF=+0.5
I
ACTIVE (+)
REACTIVE (+)
CAPACITATIVE:
CURRENT LEADS
VOLTAGE
V
INDUCTIVE:
CURRENT LAGS
VOLTAGE
Figure 2. Sign of Reactive and Active Power in ADE ICs
Do ADE chips measure power or energy?
ADE energy measurement products have ADCs on the
analog input channels that convert the ac voltage and
current signals to digital bit streams. The voltage and
current bit streams are multiplied in the digital domain;
the product is instantaneous power. Internally, this power is accumulated over time. This is energy. Therefore,
ADE chips measure energy and not power. The instantaneous power, if required, can be derived in our serial
interface (SPI) parts (ADE7753, ADE7754, ADE7756, and
ADE7759) by using the waveform sample register to read
the bit streams.
Why do I need antialiasing fi lters on the input channels?
Antialiasing filters are required for the ADCs at the
input terminals of the ADE IC to prevent possible distortion due to the sampling in the ADC. The ADCs in the
ADE775x family have a high sampling rate (approximately 800 kHz). As the Nyquist theory tells us, image
frequencies near the sampling frequency can get folded
back around half the sampling frequency (450 kHz) and
end up in the band of interest (between 50 Hz and 60 Hz),
causing distortion. A simple low-pass fi lter can attenuate
the high frequencies (near 900 kHz) so they will not end up
in the band of interest for metering (less than 2 kHz).
How do I design the antialiasing fi lters?
A simple RC low-pass fi lter is suffi cient for antialiasing
fi lters in this application. The AN-559 Application Note,
the ADE7755 reference design documentation, explains
how to design simple antialiasing fi lters for ADE77xx
products.
Where did the factor of 3 in 3SRC come from in the formula
H(s) = 1/(S
2ⴛR2ⴛC2
+ 3SRC + 1) for two RC fi lters in series
(AN-559, Figure 12)?
V
R1
IN
C1
R2
C2
V
OUT
Figure 3. Two RC Filters in Series
For a fi lter network with two RC fi lters in series, the
following equation applies:
H(s) = 1/((1 + sR1C1)(1 + sR2C2) + sR1C2)
with R1 = R2 and C1 = C2
An estimation of two fi lters in series is H(s) = G(s)W(s)
where G(s) and W(s) represent the transfer functions
of the individual fi lters. This estimation would give the
result (with R1 = R2 and C1 = C2) with a 2sRC term in the
denominator.
This estimation neglects the “crossover” term sR1C2 that
is seen in the more exact equation. You can prove this
by deriving the transfer function from the circuit without
estimation.
AN-559 explains how to use the pole location to calculate
the resistor and capacitor values for phase matching and
cancellation of parasitic shunt inductance.
What is the effect of phase mismatch in the voltage and
current channels?
The percentage measurement error in active power
caused by any phase mismatch between the voltage and
current signal paths can be approximated by the following formula:
Error Mismatch(Radians) %≈××tan( )θ 100
In the expression, represents the phase angle between
the voltage and current. As one can see, a phase mismatch of 0.1° will result in about 0.3% error at a power
factor of 0.5. Therefore, special care needs to be taken
to ensure phases are precisely matched between the
internal signal paths for the voltage and current. A large
error can occur at a low power factor with even a small
phase mismatch.
How do I calibrate the ADE metering IC?
For products such as ADE7751 and ADE7755 (singlephase) or ADE7752 (3-phase), calibration is done by
hardware. These products require resistor divider networks on the voltage channel. See the related application
notes and product data sheets for details.
For products like ADE7753, ADE7756, ADE7759 (singlephase) or ADE7754 (3-phase), calibration is done using
the registers through the SPI interface. Refer to product
data sheets and application notes for details.
At what test current do I calibrate my meter?
Meters are typically calibrated at a specifi ed base current (IB).
This current is usually 10% of the maximum current (I
Are there any differences internally between the ADE775x DIP, SSOP, or SOIC packages?
The difference is only in the package. The part’s performance is not affected by its package.
There are both analog ground (AGND) and digital ground
(DGND) on the ADE7755 and ADE7751. Why are both pins
connected to the analog ground plane on the reference design?
The ADE7755 and ADE7751 do not produce signifi cant
digital noise. Therefore, the whole IC can be set on the
quiet analog ground plane to minimize noise pickup from
other sources. Furthermore, this arrangement enables a
larger ground plane on the PCB. The key point here is to
have the digital output pins (F1, F2, REVP, and CF) connect
to the digital ground plane.
–4–
).
MAX
REV. 0
AN-639
To which ground plane should the current sensing be connected?
The digital ground plane. This can not only reduce the
noise from the noisy supply line entering the analog
ground plane, but can also divert the energy away from
the IC in an ESD event.
To which ground plane should the crystal oscillator be connected?
The digital ground plane.
What are the considerations for designing the ground plane on an energy meter PCB?
The analog ground plane and digital ground plane should
be physically separated from each other and should be
connected only at one point (star ground confi guration).
Preferably, the two ground planes should be connected
through a ferrite to minimize the noise from the digital
ground plane entering the analog ground plane.
What are the system design considerations for electrical fast transient (EFT) burst testing?
The following are some useful tips:
• Use ferrites at points where the meter is connected to
the line.
• Use a metal oxide varistor (MOV) and shunt capac itor between the line wires.
• Maximize the physical distance between the areas
with possible high voltage to avoid sparks.
What are the system design considerations for electro magnetic interference (EMI) testing?
The following are some useful tips:
• The ground plane should be made as large as
possible.
• Use a short signal path on the analog portion of the PCB.
• Eliminate ground loops.
• Use short and tight twisted-pair wires.
• Consider physical shielding.
ADE775x: SPI INTERFACE PRODUCTS
Can the ADE775x (ADE7753, ADE7754, ADE7756, ADE7759)
handle bidirectional energy fl ow?
Ye s, t he AD E7 75x (ADE7 753, ADE775 4, ADE7756, ADE7759)
can handle bidirectional energy fl ow. The energy registers
are signed.
Can ADE775x (ADE7753, ADE7754, ADE7756, ADE7759) be used for dc energy measurement?
Ye s, the ADE775x (ADE7753, ADE7754, ADE7756, ADE7759)
can be used for dc energy measurement when the HPF
in Channel 1 is turned off. Note, however, that there are
dc offsets from the ADC in both Channels 1 and 2, so
you need to perform a dc calibration to offset the error.
You can achieve this by writing to the CH1OS, CH2OS,
or APOS register to offset the error term CH1_OS1
CH2_OS2.
Which products can be used with the Rogowski coil (air-core CT, di/dt sensor)?
The ADE7753 and ADE7759 are the single-phase products that can be used with a di/dt sensor. In all cases,
the integrator that enables this direct interface can be
disabled (refer to product data sheet for register maps)
so that a current transformer (or shunt) can be used as
the current sensor.
Are there any special considerations for SPI timing to interface with the MCU?
During a multibyte data transfer, there must be at least
4 s between bytes (t
This includes writing to the communication register (the
command byte that initiates SPI communication), i.e., the
rising edge of SCLK should not occur until 4 s after the
falling edge of the write to the communication register.
For a fast MCU, the transfer time could be fast enough to
violate this timing specifi cation.
Which ADE products give reactive energy?
ADE7753.
How does ADE77xx VAR (reactive energy) calculation work?
ADE energy measurement products calculate VAR using a
single-pole low-pass fi lter with a constant 90° phase shift
over frequency and attenuation of 20 dB/decade. The cutoff frequency of the low-pass fi lter is much lower than the
fundamental frequency, so it provides a 90° phase shift
at any frequency higher than the fundamental frequency
and attenuates these frequencies by 20 dB/decade. This
solution is susceptible to variations of the line frequency.
However, a dynamic compensation of the gain attenuation with the line frequency can be achieved by evaluating
the line period of the signal. The ADE products also have
a period register that may be used for this compensation
purpose (ADE7753). For a full description and comparison
of VAR calculation methods, see the technical article
entitled, “Measuring Reactive Power in Energy Meters,”
on the ADI website.
How many line cycles do I have to accumulate over to get a
stable energy register reading from the ADE775x?
The reading will be stable in one half cycle, but the issue
is the accuracy. The accuracy of the reading will be 1/n of
LSB accumulated. This is similar to a quantization error
in an ADC.
What is the smallest number of LINCYC to get a meaningful
energy reading?
You can adjust the number of half line cycles and make
a tradeoff between low current accuracy and the time
it takes to read out each phase.
100 half cycles @ 60 Hz = 1.667 sec 32 LSB/sec = 53.333
= 53 (due to rounding, and causes the error). The error in
any measurement is going to be ± 0.5 LSB.
Therefore, the accuracy = 1 LSB of error/53 LSB = 1.875%.
and t10 in the product data sheet).
7
REV. 0
–5–
Loading...