Bidirectional Current Sense Amplifiers (CSA) are capable of measuring current flowing in either direction. The quiescent output level corresponds to zero current. One, sometimes two output reference pins are available for configuring the zero current voltage. A bidirectional CSA can function in unidirectional mode by setting the quiescent output at or close to either of the supply rails. There are many ways of driving the reference pins. The most common methods include using a reference IC, a voltage divider, or a voltage divider followed by a buffer. Output impedance of the driving source impacts the CSA output and may cause significant error if not designed properly. A voltage divider has the advantage of being versatile and inexpensive. It also tends to occupy less area. As a result, it finds adoption in many applications.
All trademarks are the property of their respective owners.
One way to classify a Current Sense Amplifier (CSA) is based on whether it is able to measure current in both directions. Then it can be put into one of two categories - unidirectional or bidirectional. A unidirectional device only linearly responds to current flowing in one direction, with its output moving in one direction in proportion to the input differential signal. A current flowing in the opposite direction causes the output to collapse to one of the supply rails, normally ground. Figure 1-1 illustrates such a scenario.
For a bidirectional CSA, current flowing in either direction is allowed. The output of the device moves off of a quiescent output level, in proportion to the input differential signal. The fact that bidirectional CSA output is able to move up toward supply or down toward ground implies that the quiescent output level corresponds to zero current. In these devices, there is typically one or two output reference pins. The output is level-shifted by driving the reference pins with a suitable source. Figure 1-2 shows the same bidirectional input is accurately reproduced. A bidirectional CSA can be configured as unidirectional by setting the quiescent output at or close to either supply rail.
This report reviews TI’s bidirectional CSA then examines different ways of configuring the output reference, associated performance tradeoffs, and the reasons behind these tradeoffs. Next, the impact of the resistor divider driving reference pins in common CSA architectures is explored. The goal is to help designers make an informed decision when choosing a reference driving circuit that meets performance requirements, and is economical at the same time.
A CSA measures target current by deriving a small differential signal (voltage or current) that is proportional to the magnitude of the current. Signal conditioning circuity then turns this small differential signal into a stable and noise-free output for further processing down the signal chain. For shunt-based current sensors, either non-isolated or isolated, the input differential signal is created by inserting a shunt resistor in the path of the target current. Magnetic sensors work without making physical contact between the sensor IC and the target current. For example, the magnetic field generated by the load current can be sensed by a Hall sensor, which is then conditioned and amplified by a low-noise amplifier.
Figure 2-1 shows a block diagram of a typical bidirectional CSA with a single reference pin. The input stage is responsible for extracting the differential input signal while rejecting the typically very high input common-mode voltage. The input stage can take on many forms, including but not limited to, voltage feedback, current feedback, and isolated technology. The output stage takes care of output drive capabilities to interface effectively with downstream circuitry.
The output stage is typically a classic difference amplifier. To enable bidirectional measurement capability, the output stage is equipped with a reference pin. By providing a positive reference voltage to the reference pin, the output is level shifted to a desired quiescent output voltage. Typically, when a positive differential input is applied, the output moves away from the quiescent voltage, toward the supply. Conversely, when a negative differential input is applied, the output moves away from the quiescent voltage toward ground.
Matching of the resistor network is important. One of the parameters that reflect how well the resistors match is Reference Voltage Rejection Ratio (RVRR). This parameter measures net change (relative to Vref) in output voltage for a given amount of change in reference voltage. If RVRR is listed in the data sheet and is input referred, the device gain should be used as a multiplier in calculating the corresponding change in output.
Some bidirectional devices come with two reference pins which are connected internally to form a voltage divider. Figure 2-2 shows such an arrangement.
As an example, for INA240, the reference divider is made up of two equal-value resistors. Real-world differences directly influence the reference voltage. For this reason, the divider accuracy is specified in the data sheet, However, if the two reference pins are shorted together and driven with a voltage source, then the divider function is not used. The divider accuracy specification is not a concern in this situation.
A common scheme of creating a reference voltage, called splitting the supply, is shown in Figure 2-3. One of the reference pins is connected to the device power supply, while the other connected to ground. This results in a reference voltage that is half of the supply. In similar fashion, this scheme can be used to create customized references, with voltage rails at different potentials.
The two-pin arrangement brings flexibility without incurring additional error compared with external resistor dividers. When the two reference pins are shorted together, they function exactly the same as a single pin and can be treated as such.
Based on conventional difference amplifiers, this topology achieves input and output functionalities in a single stage. Compared with a conventional difference amplifiers, the most significant distinction is the capability to withstand common-mode input voltage that is significantly higher than device supply voltage while maintaining high gain. For example, the working input common-mode voltage could be 28 V while the device supply is only 3.3 V. Shown in Figure 3-1 is the simplified diagram of such a device. In this diagram, Ri and Rf represent the ideal values of the input and feedback resistors respectively.
The resistor network must be closely matched to achieve good Common Mode Rejection Ratio (CMRR) and Gain Error (GE) performance. The resistor network is often trimmed for matching. Although sometimes it is possible to meet relaxed design goals by optimizing circuit design and layout without trim.
When driving the reference pin of a single-stage difference amplifier, ideally the voltage source should have zero impedance to maintain the balance of the resistor network.The difference amplifier transfer function is the familiar equation:
As Figure 3-2 shows, by attaching a noninverting output stage to the single-stage difference amplifier, a two-stage CSA with potentially very high gain is developed. Aside from flexibility in gain configuration, the intermediate output can be brought out as an optional output pin. The overall performance is dominated by the first stage, therefore much of the same performance limitations remain as with single-stage difference amplifier. Compared with other multistage alternatives discussed in the next sections, the added output stage does not significantly improve critical electrical performance.
The device gain is the product of the gains of all stages, and the transfer function is written as:
The front stage is usually a fully differential amplifier dedicated to rejecting high input common-mode voltage so that a clean and gained-up version of the differential input is passed to the output stage, which is effectively isolated from the high input common-mode voltage. This is a huge advantage in achieving exceptional CMRR.
Matching resistor networks is important in both stages to achieve excellent CMRR and GE performance. This is especially true for the input stage, where the resistor network is sometimes in-package trimmed at final production. The output of the input stage is fully differential, and the common-mode voltage is defined by the Common Mode Feed Back (CMFB) circuitry. It is worth noting that this common-mode output voltage is the same as the common-mode input voltage of the output stage. For compliance, the common-mode voltage is set between ground and supply voltage. It is normally set to somewhere close to mid-supply. With this arrangement, the output stage is decoupled from the high input common-mode voltage seen at input of the device.
The transfer function of the amplifier shown in Figure 3-3 can be written as:
Figure 3-4 shows the block diagram of the INA170. Bidirectional current measurement is accomplished by output offsetting through amplifier A2. The offset voltage level is defined by an external resistor and voltage reference. This setup permits measurement of a bidirectional shunt current while using a single supply. Applying a positive reference voltage to reference (pin 3) causes a current to flow through ROS, forcing output current to be offset from zero. The transfer function is written as Equation 4.
Driving the reference pin is straightforward. A simple voltage divider will suffice due to the high input impedance of amplifier A2. A filtering cap close to the reference pin may help limit noise contribution. Resistor sizing and VREFVref voltage should fit the operating range of amplifier A2.