# CN0100

Manufactured by:

### Circuit Function & Benefits

The −48 V power rail is widely used in wireless base stations and telecommunication equipment. Used in network central offices, it can vary between −48 V and −60 V. Measuring the current at that voltage typically requires components that operate on dual supplies, such as ±15 V. Typically, only the front-end conditioning amplifiers that interface directly with the −48 V rail use dual supplies. The remainder of the system operates on single supplies. However, eliminating the negative supply reduces complexity and cost. Using the AD629 and the AD8603 in this circuit allows designers to measure current at −48 V to −60 V while operating only on positive supplies.

Compared with low-side current sensing, the high-side current sensing rejects ground noise and can detect short circuits during operation.

Figure 1: Circuit Used to Measure −48 V Current (Simplified Schematic)

### Circuit Description

This circuit uses the AD629 difference amplifier to condition voltages beyond its supplies. The minimum and maximum allowable input common-mode voltage is determined by the following equations:

VCOM_MAX = 20 × (+VS – 1.2) – 19 × VREF
VCOM_MIN = 20 × (−VS + 1.2) – 19 × VREF

When VREF = +5 V, +VS = 12 V, and −VS = 0 V, the AD629 common-mode input range is −71 V to +121 V. This is high enough to cover the entire expected range of the −48 V rail. The AD629 difference amplifier senses the differential voltage, IS × RS, which is generated by the current flowing through the shunt resistor. Since the AD629 has a fixed gain of 1, its output voltage is equal to IS × RS +VREF.

The shunt resistor is 100 mΩ with 0.1% tolerance and a maximum power rating of 1 W. When selecting the shunt resistor, both current measurement accuracy and self-heating effects should be considered.

The AD8603 is configured as a subtractor so that it can reject the 5 V common-mode voltage and amplify the signal of interest, IS × RS. The signal is amplified by a factor of 20 to span the 2.5 V full-scale input range of the AD7453 ADC. A full-scale 2.5 V signal to the ADC corresponds to a current of 1.25 A on the −48 V supply. The AD8603 was chosen because it has low input bias current, low offset drift, and rail-to-rail input and output features. The rail-to-rail output allows the AD8603 to share the same supply as the ADC. It should be noted that the output of the AD8603 can only decrease to approximately 50 mV above ground due to its output stage. This corresponds to an input current IS of about 25 mA. Therefore, currents less than about 25 mA cannot be measured. However, accuracy for very low currents is not usually required.

The ratios of the four resistors that form the subtractor must be matched to obtain maximum common-mode rejection (CMR). In this stage, the subtractor will have to reject the 5 V common-mode signal from the AD629.

The AD7453 12-bit ADC is used because of its pseudo-differential input that can simplify the interface between the AD8603 and the ADC. In addition, its small package and low cost make it useful in cost sensitive or size limited cases.

The AD780 was chosen as the voltage reference for the AD7453 12-bit ADC because of its accuracy and ease of use.

The circuit was tested by measuring the digitized output voltage as a function of current for supply rails of −48 V and −60 V. The results are shown in Figure 2. It shows close correlation to the expected values and good linearity under different common-mode voltages.

Figure 2: Digitized Output Voltage vs. Current for Common-Mode Voltages of −48 V and −60 V

The error due to the CMR of the AD629 is the largest error. The total offset error is amplified by the signal gain of the AD8603 difference amplifier, 20, so it can be as high as 156 mV referenced to the AD8603 output.

Also, from this calculation we see that the CMR of the input difference amplifier is very important for low offset. If the current sense circuit is used outdoors, the temperature specifications (initial gain drift, offset voltage drift, and CMR over the entire temperature range) will be important, and the AD629 is ideal for such applications.

The AD8603 will contribute an error caused by its offset voltage (0.3 mV maximum) and bias current (1 pA). This produces a maximum output offset error of about 6.3 mV for the noise gain of 21. The total maximum output offset error is the sum of the error due to the AD629 (156 mV) and the AD8603 (6.3 mV), or 162.3 mV referred to the AD8603 output. Fortunately, this error can be removed with system calibration.

At the same time, if we use the typical specifications rather than maximums, the offset voltage at the AD8603 output will be approximately 45 mV.

The offset errors of the AD629 can be calculated as follows using the maximum specifications:

The circuit must be constructed on a multilayer PC board with a large area ground plane. Proper layout, grounding, and decoupling techniques must be used to achieve optimum performance (see MT-031 Tutorial, Grounding Data Converters and Solving the Mystery of AGND and DGND and MT-101 Tutorial, Decoupling Techniques).

 Initial Gain Error 0.05 mV Offset Voltage 1 mV DC CMR (77 dB) 6.768 mV Total Offset 7.818 mV

### Common Variations

The ADR361 is another option for the voltage reference because of its smaller size, low power, and high precision.

An integrated instrumentation amplifier, such as the AD8223 or AD8226, can be used instead of the AD8603. This eliminates the need for the external resistor matching required by the AD8603 circuit. A difference amplifier, such as the AD8276 with integrated resistors, can also be used instead of the AD8603 if a gain of one is acceptable.

The AD629B has 9 dB higher CMR than the AD629A. Its offset voltage is one-half and its gain error is almost one-half that of the AD629A. This is critical for cases where system calibration is impossible.

If a more integrated solution is needed for the converter, the ADuC70xx series of ARM7TDMI® precision analog micro-controllers with the integrated 12-bit, 1 MSPS ADC is a good choice.