Maxim Integrated Teridian 78M6618 Instruction Manual

78M6618
Hardware Design Guidelines
APPLICATION NOTE
AN_6618_027 October 2010
Rev. 1.2 © 2010 Teridian Semiconductor Corporation 1
A Maxim Integrated Products Brand
1 Introduction
This application note provides hardware and system design guidelines for those incorporating the
78M6618 System on Chip in their products. These guidelines will help hardware engineers to reduce
design cycle times.
The following topics are discussed:
•Non-isolated Configuration
oSafety Precautions
o3.3 VDC Supply (V3P3) and System Connection
oLine Voltage Resistor Divider Selection
oShunt Selection and Connections
•Isolated Configuration
oCurrent Transformers
oOther Connections
oVoltage Transformers
•Calibration Considerations
•Basic Configuration
oReset Circuitry
oV2P5 Voltage Reference Pin
oV1 Pin
oIn Circuit Emulator (ICE) Pins
oConnecting 5 V Devices
oDriving External Loads
oConnecting I2C EEPROMs
oConnecting 3-Wire EEPROMs
oUART0 (TX/RX)
oUART1 Interface
oPower Supply Topologies
•Timing Reference
oOscillator Connections and Components Selection
oPCB Layout Recommendations
oOther Considerations

78M6618 Hardware Design Guidelines AN_6618_027
2 Rev 1.2
2 Non-isolated Configuration
When using a resistive shunt current sensor, the measurement IC and its power domain are not isolated
from the AC mains. In this configuration, the 3.3 VDC supply rail (V3P3) for the 78M6618 must be directly
connected to AC-Neutral for precision energy measurement. Isolation components, if required, are added
in between the measurement IC and the rest of the system.
The V3P3 connection to AC-Neutral can be eliminated when using current transformers (CT) as the
current sensing elements. Refer to the section on Isolated Connections for designing with CTs.
2.1 Safety Precautions
With V3P3 directly connected to NEUTRAL, the 78M6618’s Ground signal is -3.3 V below earth ground.
Therefore, any external test equipment attached to the 78M6618 will be subject to this -3.3 V ground
reference disparity.
External test equipment must
be floated from earth ground to avoid equipment damage due to
this ground reference disparity.
An additional safety issue may arise due to mis-wiring of the AC outlets. If the LINE and NEUTRAL wire
connections at the AC plug are reversed between the 78M6618 and the external test equipment, the
78M6618 and external test equipment will see a 120/240 VAC voltage difference rather than -3.3 V. This
scenario can occur when the 78M6618 and external test equipment are powered from different wall
outlets or power strips, one of which is mis-wired. With proper earth grounding, the external equipment is
always referenced to earth ground via their enclosures.
Any systems communication interface (UART, SPI, I2C) between the 78M6618 and external
circuitry must be isolated to accommodate the -3.3 V disparity in their GND pins (or in the event
of a LINE reversal).
Refer to the 78M6618 Safety Precautions document.

AN_6618_027 78M6618 Hardware Design Guidelines
Rev. 1.2 3
2.2 3.3 VDC Supply (V3P3) and System Connection
The 78M6618 requires a single 3.3 VDC supply. The 3.3 VDC (V3P3) also represents the reference
potential for the 78M6618. The basic connections for a shunt-based system are represented in Figure 1.
Shunt x8
NEUTRAL
LINE
VA
VB
750
1M
1M
1000pF
IA-IH
1000pF
x8
Unused
Inputs LOAD
x8
750 x8
78M6618
1M 1M
EARTH
V3P3
750 1000pF
0.1uF
0.1uF
x8
0.1uF
Figure 1: Basic Connection Diagram on Shunt-Based Systems
The analog inputs to the 78M6618 are used as follows:
•The VA input is used to measure the line voltage.
•The IA-IH inputs are used to measure the load currents. Terminate unused current inputs to V3P3.
•The VB input is used to flag a Line/Neutral polarity reversal. If this feature is not desired, the VB input
can be tied directly to the VA input or directly to V3P3.
Notes:
•The values used for the anti-aliasing filters are 750 Ω and 0.1 µF. The filters have in this case a cutoff
frequency of about 2.1 KHz. Since the sample rate of the ADC converter may vary depending on the
different firmware configurations, a different value may be required. To tune the filter, it is
recommended to keep the 750 Ω resistor unchanged and modify the value of the capacitor.
Do not tie the ground of the 78M6618 directly to earth ground. See the Safety Precautions
section.

78M6618 Hardware Design Guidelines AN_6618_027
4 Rev 1.2
Effective 3.3 VDC bypassing incorporates the combination of three different capacitor values. A 1000 pF
in parallel with a 0.1 µF ceramic capacitor must be placed as close as possible to the 78M6618 V3P3A
pin. Place the 1000 pF capacitor closest to the V3P3A pin of the 78M6618. An additional 22 µF bulk
capacitor is placed in the vicinity of the V3P3SYS pin to provide decoupling for the external DIO circuitry.
Connect the VBAT pin to the V3P3SYS pin, directly to V3P3. These three capacitor values provide
decoupling over a wide frequency spectrum. Do not connect V3P3D to V3P3. V3P3D only requires a
0.1 µF capacitor to ground.
GNDA
V3P3SYS
VBAT
V3P3A
GNDD
1000pF 0.1μF 22μF
V3P3
Place Capacitors
Close to V3P3A
V3P3D 0.1μF
Figure 2: Power Supply Decoupling

AN_6618_027 78M6618 Hardware Design Guidelines
Rev. 1.2 5
2.3 Line Voltage Resistor Divider Selection
The input line voltage must be scaled to match the 78M6618’s ADC input signal range of 176.78 mVrms.
In the example of Figure 3, the line voltage is scaled as follows:
The use of two 1 MΩ resistors instead of a single 2 MΩ resistor is required to meet the maximum voltage
rating of the resistor package and to provided adequate breakdown and arcing clearance. 1206 series
surface mount resistors are recommended.
An important aspect to consider is the accuracy of the resistors in the voltage ladder. Another
consideration that can affect the overall measurement accuracy is the resistor temperature coefficient
(TCR in Ωppm/°C).
VA
VB
1M
1M
750
LINE
V3P3 1000pF
0.1μF
Figure 3: Input Voltage Divider and Filtering
Initial component tolerance can easily be compensated for during calibration. Depending on the system
accuracy requirements, the incremental cost of a higher precision resistor (i. e. 0.1% versus 5%) may
result in a lower production line calibration cost (shorter calibration time). If 0.1% tolerance resistors are
used, a calibration would only need to be done on one system and then the voltage calibration coefficient
obtained can be used for all others in production thus saving time through the production line. The
variation from board to board would be within the tolerance of the resistors used in the divider.
Additionally, the higher precision resistor will have a smaller temperature coefficient. This eliminates the
source of error that can arise from changes in resistance from self heating as the line voltage changes.
TCR of 50 ppm/°C or lower (preferably 25 ppm/°C) is recommended.

78M6618 Hardware Design Guidelines AN_6618_027
6 Rev 1.2
2.4 Shunt Selection and Connections
The 78M6618 sensor input range is 12.5 μV (8.84 μVrms) to +250 mV (176.78 mVrms). The value of the
shunt to be used is usually a tradeoff between a higher shunt value to utilize the full analog sensor input
range of the IC and the power loss in the shunt.
Use the maximum rated load power when calculating the value of the shunt resistor for best utilization of
the ADC’s input range (+/-250 mVpp). Also, use the lowest operating LINE voltage (for example 90 VAC
for 120 VAC rated systems) for this calculation. The maximum input current is then:
For example, if the maximum input power is 1.0 KW, the maximum input current is 11.12 A rms.
The resultant peak-to-peak current is calculated to be:
In the example above, the peak to peak value is 31.4 A.
A 15 mΩshunt value fully utilizes the ADC input range. This shunt value produces a dissipated power of
1.85 W at maximum load current. In order to ensure more ADC signal margin due to transients and to
lower the power dissipation in the shunt resistor, a lower value shunt of 8 mΩis recommended. In this
case, the shunt’s power dissipation reduces to 0.99 W at the maximum load current.
The next steps involved in the shunt resistor selection include considerations for power dissipation, initial
tolerance and the device’s temperature coefficient. In the case selected above, the power dissipated in
the shunt at maximum load current is 0.99 W. A 2 W rated device package is recommended for good
long-term reliability. The initial tolerance can be compensated during calibration. However, the
temperature coefficient plays a role in the overall accuracy and cannot be easily compensated. For
example, a temperature coefficient of 100 ppm/°C causes a resistance variation of 1% over the 100 °C
operating temperature environment.
IA-IH
V3P3
SHUNT
750
0.1μF1000pF
Figure 4: Shunt Connections and Filtering

AN_6618_027 78M6618 Hardware Design Guidelines
Rev. 1.2 7
2.4.1 Single-ended Shunt Connections
Multi-shunt measurement systems typically require use of differential input circuits. These differential
input circuits add cost due to added components. The 78M6618 Evaluation Board presents a lower cost
multi-shunt measurement system alternative. Additionally, the proposed circuit design results in a very
compact printed board layout for a high accuracy energy measurement system.
The key for single ended measurements is a noise free reference point. Attaching multiple shunts to a
common reference point presents mechanical challenges due to physical size and placement. The linear
spacing required (side by side placement) for multiple shunts compromises the “noise free” reference
point. The resistance along a linear surface from one shunt to another is not insignificant relative to the
low ohms value of the current sensing shunt.
Measurement errors occur in adjacent shunts as the collective currents produce voltages in the
interconnecting “reference point” structure. The small resistance present in the interconnecting structure
between adjacent shunts creates this voltage offset when high load currents are present. The voltage
offset results in a measurement error for each shunt. Reducing the resistance of the interconnecting
structure is critical to achieving high accuracy in a single ended multi-shunt measurement system.
The 78M6618 Evaluation Board minimizes the interconnecting structure’s resistance and minimizes
adjacent shunt influences (error voltage due to load current) by employing a radial disk topology. The
78M6618 reference point utilizes a copper disk. The copper disk provides a much lower resistance
structure compared to 1 oz. copper plating. The thickness of the copper disk is equal to the printed circuit
board thickness for ease of manufacture. Uniform distance to the reference “center” point is achieved by
placing the shunts radially around the copper disk. Placing the shunts on both sides of the board
minimizes the diameter of the copper disk reducing the overall reference point resistance.
Figure 5: 78M6618 Evaluation Board Copper Disk with Radial Shunt Placement
The 78M6618 Evaluation Board is designed for a typical 15A household service. Ideally, a large current
load at one outlet should not affect the remaining outlets. Due to the finite resistance of the copper disk,
a measurement error is unavoidable. The copper disk’s resistance is calculated as follows:
Copper resistivity at 25C = 1.7 x 10-6 ohm-cm
Resistance = Resistivity*Length/Area*(1 + (Temp_Co*(Temp - 25))
Assuming a straight conduction path from the shunt to the copper disk center point:
Thickness = 62 mils, same as printed circuit board thickness
Width = 200 mils, width of shunt pad
Length = 100 mils, distance of shunt to center point
The linear resistance from the shunt to the copper disk center point = 5 µohm
Copper
Disk

78M6618 Hardware Design Guidelines AN_6618_027
8 Rev 1.2
This represents a 0.1% error when using a 6 mohm shunt.
Current does not flow in a single straight line between two points in a conductor. Due to internal
resistance in the conductor, the current spreads out through the medium. Figure 6 shows how current
from one outlet produces a distributive voltage across the copper disk.
Figure 6: Single Load Voltage Distribution in a Disk
Observe how the presence of a current at one location (assume position 0 degrees), creates a small
voltage at positions 45, 90, 135, 180, 225, 270 and 315 degrees (the other shunt locations).The shunts
most affected by a single large load (at position 0 degs) are the two adjacent shunts on the opposite side
of the board (at positions 45 and 315 degs). The above voltage distribution assumes no current flow in
the other seven outlets.
When current flows in two or more outlets, the current distribution through the copper disk depends on the
difference between the adjacent currents. A small outlet current flowing next to a large outlet current will
endure a non-linear measurement error due to the larger load’s current distribution implied in Figure 6.
The smaller current’s path to the center point is no longer a straight line but a longer curved path. The
longer curved path effectively reflects itself as additional resistance in the path from the shunt to the
center point.
The above condition exhibits itself when the relative current magnitudes approach 100:1. For example, a
10 A load at one outlet affects the accuracy of an adjacent 100 ma load. The 100 ma load will be
measured as 105 ma. Similarly, a 1 A small load with a 10 A large load will be measured as 1.002 ma.
2.4.2 Differential Shunt Connections
Alternatively, the system designer can preserve the use of differential measurements on the 78M6618 by
using a differential op-amp between the current sense input of the 78M6618 and the shunt resistor.
Contact Teridian support engineers for more information on using op-amps for differential input
measurement of shunts.
Outlet Current
Exit Point
Edge of Disk
Source Current
Inlet Point
Disk Center Point

AN_6618_027 78M6618 Hardware Design Guidelines
Rev. 1.2 9
3 Isolated Configuration
Alternatively, the 78M6618 can sense the load current using a current transformer (CT) for an isolated
configuration. In this configuration, the 3.3 VDC supply rail (V3P3) for the 78M6618 is not directly tied to
AC mains. The analog inputs to the 78M6618 are used as follows:
•The VA and VB inputs are used to measure the line voltage differentially
•The IA-IH inputs are used to measure the load currents.
•Terminate unused current inputs to V3P3 as shown in Figure 4.
3.1 Current Transformers
The selection of a current transformer with respect for the desired measurement accuracy includes
factors such as line frequency, measured current range and the CT’s turns ratio. Also, subjecting a
current transformer to load currents above the manufacturer’s rated current specification may saturate the
CT and cause winding failures due to excessive temperature rise. On the other hand, a current
transformer that is rated much higher than the target load current might be restrictively too large and
expensive for its purpose.
Figure 7: Current Transformer (CT) Basic Connections
Usually, current transformers have turns ratios ranging from 10:1 to 2500:1. The higher the turns ratio
(TurnRatio = Nsecondary/Nprimary), the higher the resolution of the current measurement. A too high
turns ratio increases distributed capacitance and leakage inductance. These characteristics may
decrease the CT’s accuracy and capability to operate at higher frequencies. However, if the number of
turns is too low, the output signal may distort or “droop” (for positively sloped unipolar input signals).
Such distortion may cause measurement inaccuracies.We recommend a minimum turns ratio of 1000:1.
The next step towards selecting a current transformer is the calculation of the burden resistor’s value
(RBurden). The 78M6618 signal input range is 176.78 mVrms (250 mVpk).Therefore, the CT’s
secondary output voltage (Vout) must operate within this range. Assuming the maximum load current is
20 Arms (28.284A pk), a 1000:1 ratio current transformer will produce a secondary current of 20 mA rms
(28.284 mA pk). Per Figure 7, the burden resistor’s value is calculated as follows:
Using the values in the above example, the value of the burden resistor is:
A standard value 8.2 Ωresistor is recommended.

78M6618 Hardware Design Guidelines AN_6618_027
10 Rev 1.2
Figure 8 shows a basic connection diagram of a CT-based system.
Figure 8: Basic Connection Diagram of a CT-Based System
The use of a CT allows for the 78M6618’s V3P3 to be isolated from the plant NEUTRAL wiring. This
topology eliminates the safety issues stated earlier regarding shunt-based current sensing. Figure
shows a 2 MΩ isolation from the LINE voltage via the voltage divider network. In this topology, the line
voltage is a pseudo-differential measurement of VB-VA.
The V3P3 reference point critical to multi-shunt measurement performance is not an issue with CTs. The
output currents generated by the CT’s secondary winding is small enough that the sheet resistance of the
1 oz. copper plating does not present measurement errors from adjacent CTs. Shield the CTs secondary
pins, burden resistor and filter components with top and bottom printed circuit board layer V3P3 plane
surfaces. Insert multiple V3P3 vias to interconnect the top and bottom V3P3 structures for a low
impedance shield. Refer to the 78M6618 Printed Circuit Board Layout Guidelines application note for
additional layout design recommendations.
See Using Current Transformers with the 78M661x for additional information.
3.2 Other Connections
The same power supply decoupling circuit from Section 2.2 can be used for an isolated configuration.
The 78M6618 supply and ground connections, however, can be shared with other (isolated) components
in the system when V3P3 is not tied to AC Neutral.
Line voltage sensor recommendations for a non-isolated sensor configuration (resistor divider) can also
be used in an isolated configuration due to the high impedance connections between high voltage AC and
the 78M6618. Alternatively, a voltage transformer (VT) can be used for fully isolated voltage sensing.
NEUTRAL
LINE
VA
VB
750
1M
1M
IA-IH
Unused
Current
Inputs
LOAD
x8
750
78M6618
1M
1M
V3P3
1000pF
Rburden
x8 750 x8
1000pF
x8
0.1μF
x8
0.1μF
1000pF