Lithium-ion (Li-Ion) battery packs contain a large number of cells that must be properly monitored to improve battery efficiency and extend battery life. The 6-channel AD7280A in the circuit shown in Figure 1 acts as the master monitor, providing accurate measurement data to the battery management controller (BMC).

The AD7280A has a built-in ±3 ppm voltage reference that provides battery voltage measurement accuracy of ±1.6 mV. The ADC resolution is 12 bits, and it only takes 7 μs to convert 48 units.

The AD7280A is located on the high-voltage side of the battery management system (BMS) and has a daisy-chain interface that allows up to eight AD7280As to be stacked together to monitor the voltage of 48 lithium-ion battery cells. Adjacent AD7280As in the stack can communicate directly, passing data up and down without isolation. The AD7280A master device at the bottom of the stack uses the SPI interface to communicate with the BMC. Only here is high-voltage galvanic isolation required to protect the low-voltage side of the BMS. The ADuM1201 digital isolator and the ADuM5401 isolator with integrated DC/DC converter together provide the required 6 channels of isolation in a compact, cost-effective solution.

The AD7280A daisy chain draws power from the battery cells it monitors. The ADuM5401 integrates a DC/DC converter to power the high voltage side of the ADuM1201, provide V _DRIVE power to the AD7280A SPI interface, and provide a shutdown signal to the AD7280A daisy chain circuit. If the +5 V supply on the low side of the BMS is pulled low, the isolator and AD7280A daisy chain shut down. Likewise, if the PD signal from the BMC goes low, the The ADuM5401 low voltage supply routed by the ADG849 switch will be pulled low, which will also cause a hardware shutdown of the isolator and AD7280A daisy chain.

To optimize the communication performance of the daisy chain under high-noise conditions (such as when encountering battery interference), the daisy chain signals are shielded on an inner layer of the printed circuit board (PCB), with shielding provided above and below by the V _SS power plane. The power plane is connected to the VSS pin of the previous device in the daisy chain. Figure 2 shows the top layer of the EVAL-AD7280AEDZ PCB, which contains the upper shield. Figure 3 shows the inner layer (Layer 2), which contains the shielded daisy chain signals, with the shielding below implemented on Layer 3 shown in Figure 4. Each daisy chain connection is equipped with 22 pF capacitors that are terminated at the V _{SS pin of the previous device or the V DD} pin of the next device , depending on the direction of data flow in the daisy chain. The PD , CS , SCLK, SDI and CNVST daisy chain connections pass data up the daisy chain, so the 22 pF capacitors on these pins are terminated to the V _SS pin of the previous device.

The SDOlo and ALERTlo daisy chain connections pass data down the daisy chain, so the 22 pF capacitors on these pins are terminated to the V _DD pin of the next device. Use a low-impedance trace to directly connect the next device's V _DD to the previous device's V _SS so that these two potentials are as close as possible in a high-noise environment.

The ground guardrail at the isolation barrier is used to surround the low-voltage end formed on the left side of the PCB. This guardrail consists of a guard ring tied together by vias, connected to the digital ground of all layers on the board. Noise reaching the power and ground planes at the edge of the board may be radiated, but with this shielding structure, the noise is reflected back.

Input-to-output dipole radiation may also occur when a current source is driven across the gap between ground planes. To minimize this effect, use a continuous shield across the isolation gap, thereby extending the ground plane to all layers of the PCB to utilize shield overlap for cross-isolation barrier coupling. Isolation gaps on the layers were kept to a minimum, with the test board using a 0.4 mm gap. For more advice on radiation control from iso Power® devices such as the ADuM5401 used in this circuit, see application note AN-0971 .

Test Results

An important measure of the performance of this circuit is the amount of noise in the final output voltage measurement.

Figure 5 shows a histogram of 10,000 measured samples for the VIN3−VIN2 channel. This data was measured using the AD7280 evaluation board connected to the EVAL-CED1Z converter evaluation and development board. See the "Circuit Evaluation and Testing" section of this circuit note for setup details.

Use a resistor divider string driven by the supply voltage to simulate the battery voltage. The captured codeword 2675 represents 3.612 V, which is a typical lithium-ion battery voltage. Note that only a small portion of the codewords are affected by noise and fall outside the main bin.