BMS Current Sense Research

The current sensing system is vital in our solar car. As the batteries can discharge a large amount of current in the case of fault, we need a real-time current measuring circuitry to ensure that the battery operates within the safe current limits (overcurrent protection). Sensing the current also helps us make an accurate state of charge model. There are mainly two types of current sensing methods: shunt-based and magnetic-based approaches.

Shunt-based Sensing

Shunt-based sensing is based on Ohm’s law, where the voltage drop across the resistance is used as a proportional measurement to the current flow. A shunt sensor is an inherently closed loop because the signal on the output is directly proportional to whatever currents are flowing through the device. The shunt resistance can be introduced into the current-conducting path in two ways: low side current sensing and high side current sensing. In the former, the shunt resistor is placed on the retune path between the load and the ground. This connecting method faces some problems. As we place the resistor in the ground of the load, its ground has a potential higher than the system ground, which may affect some analog circuits. Another serious problem is that low side sensing fails to detect some fault conditions like that between the supply and the load. Therefore, we use a high-side current sensor where the shunt is placed between the positive terminal of the power supply and the load. In this case, as the common-mode voltage is almost the supply voltage, we need a high common-mode rejection ratio (CMRR) to prevent such a large CM voltage to appear on the output side. So, the CMRR range is an important specification we need to have in our amplifier. Moreover, the shunt-based method is inherently not isolated. In our system, the ground for the resistance is almost the supply voltage. Such high voltage can damage sensitive components in the controller side which are a reference to the system ground level. Therefore, we need to implement isolation between the two systems.

The shunt-based method is straightforward and not expensive. More importantly, it has very high accuracy due to its low offset, offset drift over temperature, linearity error and limited sensitivity to external magnetic fields. However, as the shunt resistance causes power losses based on Ohm’s low, we need to use very small resistance to limit the loss and the heat dissipation. As a result, we end up with a very small voltage. Thus, we need to use an amplifier to be able to process the voltage by the system downstream. The amplifier should be able to support very small input voltage ranges and be bidirectional. In general, the shunt-based method is recommended for the low and medium current levels to avoid power loss and heat dissipation.

A concerning issue with the shunt-based method is that it does not have isolation; the monitored system and measuring system are coupled. Therefore, we need to introduce an isolation circuit. We use an isolated DC-DC converter between the supply from the carrier board and the current sensing system. We also introduce an isoSPI transformer between the ADC (Analog-to-digital converter) and carrier board. Such circuitry can complicate the system and increase the cost. To solve this problem, we consider using hall effect-based current sensing.

The current system we have:

Open

To replace the isolation circuity, there are two isolation methods: isolated shunt-based sensing and magnetic-based sensing (hall effect).

• Isolated Shunt-based Sensing

The isolated shunt-based sensing method combines the features of the non-isolated shunt-based system with those normally associated with much more expensive and bulky closed loop Hall effect current sensors. It uses either an isolated amplifier or an isolated modulator where the output is separated from the input circuitry by an isolation barrier. The galvanic isolation is usually achieved by capacitive isolation, which has low thermal profiles, long lifetime operation and is highly resistant to electromagnetic interference (EMI) caused by external sources. Both isolated amplifiers and isolated modulators feature a delta-sigma modulator that uses the internal reference voltage and a clock generator to convert the analog input signal to a continuous 1-bit digital stream that’s then transferred across the isolation barrier. However, isolated amplifiers have an analog signal on the input and an analog signal on the output. After the modulated output passes over the capacitive isolation barrier, the modulated output is then recovered, lowpass-filtered, and buffered to the output as a differential analog output signal. The isolated modulator produces a digital output where the analog signal undergoes only one analog-to-digital conversion reducing the number of components and solution size with higher accuracy. Taxes Instruments introduces AMC1301-Q1 as an isolated amplifier and AMC1305M25-Q1 as an isolated modulator.

Like the non-isolated shunt-based method, the isolated shunt-based method has a very high level of DC accuracy, with the isolated modulator achieving a slightly higher level than the isolated amplifier. The cost of this approach depends on the accuracy requirement of the application. If it is <3%, then this method is less expensive and outweighs other current sensing methods. The shunt-based method has a higher operating temperature than other methods. Unlike non-isolated, this method can provide very high isolation similar to that of the closed-loop method.

Open

Magnetic-based Sensing

This method utilizes Amper’s law. When the current passes through a conductor, it generates a magnetic field. The magnetic field is proportional to the current passing through the primary conductor and produces a voltage across the other conductor. However, this method is vulnerable to magnetic interference which affects the offset and linearity performance. The inherent isolation between the two systems is the main advantage of this method. It also has very low power loss and can support high voltages. Mainly, there are two types: in-package and module sensors.

• In-Package Hall Sensing

In the in-package method, the current to be measured passes through the device package, and the magnetic field generated by the current flow through the lead frame is measured internally with an isolated sensor.

They are smaller and cheaper compared to other hall effect approaches without the need for any external components. However, this method is very susceptible to interference of external magnetic fields reducing its accuracy. Moreover, they have magnetic offset and are drifted over time and temperature. To address this issue, the TMCS1100 family from TI provides zero-drift sensors with real-time sensitivity compensation, improving the accuracy of the measurement. Another concern is that the input current capability is limited due to thermal considerations. When large power is dissipated in the loop, it results in high power dissipation in the loop, and the entire lead frame. For example, TMCS1100 has a maximum continuous RMS current of 30 A. The current capability can be increased by increasing the size of the input conductor like ACS758xCB from Allegro, reaching 200 A. However, this increases the size and temperature of the component and ends up with a higher total error. It also does not have as high an isolation rating as other isolated methods.

• Module Hall Sensing

Module sensors utilize ferromagnetic material in the form of a ring (core) with a gap that contains the hall effect sensor. It allows for high power measurements. By using a magnetic core to concentrate the magnetic fields from the primary current, the influence of the stray external fields is significantly reduced and thus, increasing the measurement accuracy.

There are the open-loop and closed-loop modules. The open-loop approach is simple, but it has magnetic offset error induced by the residual flux of the magnetic core in the transducer. This offset error depends on the previous core magnetization. Due to the eddy current and hysteresis losses, this approach is less accurate. To solve the problem, a closed-loop approach can be used to null the magnetic field within the path. The output voltage is used as an error signal to force current through the winding of the core, which opposes the primary current passing through the primary conductor. Assuming that the produced current compensates for the magnetic flux across the sensor, the produced current is proportional to the primary current and achieves zero flux.
The linearity becomes independent of the magnetic flux sensor. Due to the theoretical zero magnetization in the core, the eddy current and the hysteresis losses are significantly reduced, resulting in reducing the influence of the thermal drift. Therefore, the closed-loop current sensors achieve high measuring accuracy. As the bandwidth is low, it is suitable for DC and low-frequency AC. The disadvantage is that the system needs external components for the closed-loop compensation. It also has large bulk due to the coil. Thus, this method is expensive. The closed-loop modules are higher in cost compared to other solutions.

As per Taxes Instrument design reference, the magnetic field is concentrated through the core and measured through a sensing element placed in the gap of the core. The signal then is passed on to a signal-conditioning stage for filtering and amplification. The compensation process is done through an external drive stage. The output voltage is measured across a shunt resistance. The proposed design is as follows.

Closed-loop Current System Design Reference by TI

Conclusion
The current sensing system we have is non-isolated shunt-based. Therefore, here I explored different isolated sensing methods so we can remove the isolation circuity, simplifying the system and achieving higher safety. The discussed methods are isolated shunt-based sensing, in-package hall sensing, open-loop hall sensing, and closed-loop sensing. Although all methods achieve good isolation, isolated-based and closed-loop sensing seem to be good alternatives. However, the isolated shunt-based has a small size and costs less.