±30A Energy Monitor IC Provides Accurate Board Level Measurements

With today’s emphasis on energy efficiency, designers need a relatively simple and accurate way to monitor the energy characteristics that determine efficiency. One possible solution is Linear Technology’s LTC2947, an energy-monitor IC that provides accurate digital monitoring of dc supply rails in the 0- to 15-V range, for applications up to 30 A (Fig. 1). Specified over the commercial and industrial temperature ranges, the LTC2947 comes in a 32-lead, 4- × 6-mm QFN package.

Monitoring solutions that use external sense resistors provide the flexibility of sizing the sense resistor per the exact application requirements. But designers usually want to choose the smallest-value sense resistor to minimize power loss, along with having low tolerance and drift to maximize measurement accuracy. When dealing with high currents, however, it can be difficult to find a sense resistor that meets all of the critical design parameters, including high accuracy, small physical size, and low cost.

Powerelectronics Com Sites Powerelectronics com Files Uploads 2015 06 0916 Pet Ltc2947 F1

1. The LTC2947’s VP pin is the Positive Voltage Sense Input that should be connected to the positive terminal of the voltage to be measured. In a similar vein, VM is the Negative Voltage Sense Input that should be connected to the negative terminal of the voltage to be measured.

The LTC2947’s integrated 300-µΩ temperature-compensated sense resistor addresses these issues, thus enhancing the IC’s monitoring capability. Featuring a footprint measuring a mere 24 mm², the device provides better than 1.5% accurate energy readings over its entire operating temperature range, and across a wide 15-mA to 30-A current measurement range

High accuracy is maintained with a factory trim of both absolute value and tempco compensation of the integrated sense resistor, together with an ultra-low offset analog-to-digital converter (ADC). Like all resistors of this type, the integrated sense resistor may exhibit minor long-term resistance shifts. However, drift will be much less at lower temperatures and/or currents.

When measuring 30-A full-scale current, the integrated sense resistor’s voltage drop is about 9 mV. When measuring a 6-A rail, power dissipation is about 10 mW. Besides low power dissipation, the LTC2947 offers high dynamic range due to 15-mA maximum current offset (4.5 µV) over temperature.

Three no-latency DS ADCs and an internal or external precision time base (crystal or clock) provide accurate measurement. With accuracy denoted in parentheses, measurement parameters include:

• Current (1.0%)

• Voltage (0.5%)

• Power (1.2%)

• Charge (coulombs) (1.0%)

• Energy (1.2%)

In addition, it provides run time and chip temperature. All digital readings, including minimum and maximum values, are stored in registers accessible by a selectable I²C or SPI interface. To aid the user, a dedicated alert pin signals if measurements exceed configurable warning thresholds.

The ability to accurately measure charge and energy requires precise timing. Therefore, the LTC2947 uses either a trimmed internal oscillator or an external clock as the time base for determining the integration period. It can use either an external square-wave clock in a frequency range between 200 kHz and 25 MHz, or a 4-MHz crystal as an external clock input.

The LTC2947 measures each input with an ADC specifically tailored for that task. Current through the internal sense resistor is measured with a ∆Σ ADC that has a measurement range of ±30 A and 3-mA resolution. Common-mode voltage can range from 100 mV below ground up to 15.5 V, regardless of the supply voltage at AVCC (analog power supply).

Powerelectronics Com Sites Powerelectronics com Files Uploads 2015 06 0916 Pet Ltc2947 F2

2. These two plots compare current and voltage waveform changing phases over a 20-µs interval for both typical and LTC2947 power measurements. With the LTC2947, power is calculated as the average of multiplied samples. If the power had been calculated instead by multiplying average current with average power, the resulting error in this example would be 7.8%.

This ADC uses a first-order architecture and continuous offset calibration to ensure that all input samples are averaged with equal weight and none are missed—the current ADC is never “blind.” A 10-MHz sampling rate maintains averaging accuracy for all current waveforms, including harmonics up to 2.5 MHz. A new average value is reported every 100 ms.

A second ADC sequentially measures both temperature and differential voltage between the VP and VM pins while making the current measurement. The temperature measurement is both reported to the host and used internally by the LTC2947 to compensate for the temperature drift of the internal current sense resistor, resulting in very stable current measurements. The voltage measurement has 2-mV resolution and temperature resolution is 0.204°C. The differential-voltage measurement range (VP-VM) ranges from –0.3 to 15.5 V, independent of supply voltage.

A sensor on the die makes the temperature measurement, which can vary significantly from ambient temperature if the current in the internal sense resistor is high. A high supply voltage at AVCC/DVCC (digital power supply) increases the internal power as well as the internal temperature.

Power Measurement

The LTC2947 measures power with a third ADC that multiplies voltage (VP-VM) times current at the full 5-MHz sampling frequency, prior to any averaging due to the analog-to-digital conversion. This maintains accuracy even if current and voltage change in phase during the 100-ms conversion time, which can happen if the power is drawn from a source with significant impedance, such as a battery.

Current and power measurements are integrated over time to calculate charge and energy flowing to or from the load. It also keeps track of total accumulated time used for the integration. The integration time base can be provided by the internal clock, an external clock attached to CLKI, or an external crystal connected to CLKI and CLKO. If an external clock is used, the LTC2947 presents time, charge, and energy as a mathematical relationship to the external clock period.

Powerelectronics Com Sites Powerelectronics com Files Uploads 2015 06 0916 Pet Ltc2947 F3

3. Total unadjusted error (TUE) is shown for power measurements performed on 3.3- to 12-V supply rails running up to 30 A.

The IC provides two sets of registers for charge, energy, and time. Each register set can be separately configured to accumulate either based on the sign of the measured current, or by the level of the GPIO pin, or by the Control register settings. This allows the first set of accumulation registers to be configured to always integrate, while the second set only integrates if current is positive (to account for battery-charging efficiency, for example). A minimum current threshold can also be set below the integration that is stopped.

Operating Modes

• Single-shot mode: This takes four measurements (current, voltage, power, and temperature), and updates the corresponding registers and the Minimum/Maximum and Threshold registers. No time measurements are made and the Charge and Energy registers are not updated. One single-shot measurement cycle takes 100 ms.

• Continuous-measurement mode: This mode repeatedly measures current, voltage, power, and temperature; recalculates energy, charge, and time; and updates the Minimum/Maximum Tracking and Threshold registers. Each measurement cycle takes about 100 ms.

• Shutdown mode: Supply current reduces to about 10 μA in shutdown mode. If the device is in the middle of a measurement cycle, in either single-shot or continuous mode, it completes the cycle before entering shutdown. Shutdown clears the voltage, current, and temperature results, but preserves the values of the accumulated quantities charge and energy and all threshold and tracking values.

Figure 2 is an example of current and voltage waveforms changing phases over a 20-µs interval. In typical power or energy-monitoring ICs, power is calculated as the average current multiplied by the average voltage. With the LTC2947, power is calculated as the average of multiplied samples. When only two samples are used, the LTC2947’s 0.218-W power calculation more closely resembles actual power, whereas the typical power or energy-monitoring IC’s 0.234-W power calculation translates to a 7.3% error. The LTC2947 avoids this error and maintains accuracy with up to 50-kHz signals.

Powerelectronics Com Sites Powerelectronics com Files Uploads 2015 06 0916 Pet Ltc2947 F4

4. Regardless of using an internal or external clock when measuring the energy of a 12-V, 30-A supply, the LTC2947’s TUE is typically in the range of 0.25%.

Figure 3 illustrates the LTC2947’s high accuracy—the total unadjusted error (TUE) is close to 0% when measuring the power of 3.3- to 12-V supply rails running up to 30 A. Figure 4 shows that, when measuring the energy of a 12-V, 30-A supply using an internal or external clock, the TUE does not typically exceed about 0.25%.

TUE encompasses all of the errors in an ADC, including gain, offset, and INL (integral nonlinearity), and is generally a good indicator of overall measurement performance. As a result, these particularly low TUE figures may ultimately eliminate the need for calibration in some designs.

Evaluation of the LTC2947 is simplified with the DC2334 demo board and accompanying Windows GUI. With these tools, users can do things like choose the clock source; view the value of each measured parameter, its min and max value, and alert status; and even view the registry contents in its raw form.