How measurement systems can improve EV/HEV batteries in noisy environments
Article By : Ivo Marocco, APP-BMS-BAP
By eliminating the need for external noise filtering components, engineers can improve measurement integrity and accuracy, channel-to-channel measurement synchronization and reduce the time for all measurements to return to the host...
Confidence is essential to the mainstream adoption of electric and hybrid electric vehicles (EVs/HEVs), however to boost confidence, we must improve the accuracy of battery cell measurement within these vehicles. In order to achieve higher levels of measurement accuracy, the high-levels of noise that interferes with data acquisition and its transfer to the main processors must be dealt with. Measuring battery cell voltages, temperature and current with high accuracy is not sufficient and requires synchronization.
Figure 1: Example electric vehicle showing power sources that require battery cell monitoring
The noise sources in EVs/HEVs come at different frequencies and with various amplitudes, which makes it challenging to decipher how to best filter them so the measurements of cell voltages, temperature and pack current is not affected. Errors in measurement may lead to various consequences, including erroneous reporting of the battery state-of-charge, possible over-charging and excessive battery cells discharging with possible impact on the safety of the driver, passengers and the vehicle. To help combat these challenges, Texas Instruments’ portfolio of battery monitors and balancers is designed to achieve high-voltage measurement accuracy with integrated noise filtering, which also minimizes the need for additional external components.
The drawbacks of today’s signal noise filtering solutions
For drivers and passengers, the modern automobile is much quieter, even when it’s not an EV/HEV, but there’s a lot of signal noise not heard that affects internal systems, including the measurement of battery voltage, temperature and current and the way this data is communicated to the main electronic control unit (ECU).
This signal noise is coming from different areas of the vehicle, including the heater, inverter motor and charger. These various noises all resonate at different frequencies, ranging from tens of hertz to a few hundred megahertz, affecting the quality of the signal that needs to be monitored. As a result, noise cancellation or at least damping the majority of noise, becomes a ‘must-do’ irrespective of its origin, in order to achieve the highest possible performance. Improper or insufficient noise reduction can introduce harmonic components in the measurement path leading to additional error that the system won’t be able to account for.
Original equipment manufacturers (OEMs) have a major challenge since it is difficult to precisely characterize the noise sources such that a clear component selection would enable their complete filtering. This unknown has repercussions in the way the filtering is done. Often, design engineers will choose discrete RC filters and ICs that are conservatively over-designed in order to be on the safest side, ultimately impacting the total solution cost and effectiveness.
BMS system integrators and designers should also pay attention to the type of data converters that are integrated into the battery monitor. A BMS monitor where there are parallel sigma-delta ADCs with decimation filters per each channel, for example, can help suppress noise, but they come with a longer conversion time for each measurement. This in turn, affects the total voltage measurement speed. On the other hand, a multiplexed SAR ADC converter is much faster, but it comes with a time delta between cell voltages sampled across all channels posing a question on their synchronization.
Overcoming challenges related to cell measurement synchronization
Synchronizing cell voltage measurements will certainly play a major role in how accurate the state-of-charge (SOC) algorithm will be able to determine the battery state-of-charge with the least possible error. These algorithms are different from OEM to OEM, and the result is that there is not really a strong consolidated specification of the minimum synchronization required for the cell voltage measurements. However, there is a common agreement amongst OEMs that this number must be well below 1msec and as close to 0 as possible.
The number of multiple cells that each BMS monitor can measure at the same time also comes into play. As already mentioned above, depending on the BMS monitor architecture and the number of channels, it is possible to obtain perfect synchronization by having an ADC, like a sigma-delta, on each channel so they can all start measurement at the same time.
However it is also important to keep in mind the time delay that occurs on the daisy-chain communication line, as each BMS monitor transmits its data down to the main ECU. Here is where communication speed and frame protocol have to be considered. Also in this case there is a lack of consolidation amongst OEMs as far as this requirement is concerned. The market is assessed around numbers like 10msec, 20msec and sometimes even 100msec. This means that for example the ECU will have to receive the data related to the cell voltages of a 400V system every 10msec and within that time the cells voltages sampled across all 96cells must be aligned within less than 1msec.
Using external components to filter noise doesn’t add up
For an effective and cost optimized solution, filtering noise in a battery management system by minimizing and ultimately eliminating the need for external components is the approach Texas Instruments has taken with its family of automotive battery monitors and balancers.
The BQ79616-Q1 solves the noise problem by integrating front-end filters before ADC measurement, so high frequency noise is suppressed before sampling occurs. The integrated front-end filters enable the system to implement with simple, low voltage rating, differential RC filters on the cell input channels.
Furthermore, post measurement filters are integrated to improve measurement accuracy after the ADC conversion, with a variety of frequency filtering options to choose from. The integrated, post-ADC digital low-pass filters enables DC-like voltage measurements for better SOC calculation. The TI monitor supports autonomous internal cell balancing of up to 240mA at Ta=80C with temperature monitoring, auto-pause and restart balancing to avoid overtemperature condition. This enables the ECU to have less overhead and perform additional processing at faster speed.
To speed up the delivery of all cell measurement results, the BQ79616-Q1 optimizes the communication protocol for fast data return in a daisy chain configuration to better reduce the delay from one device to another. For example, in a 96-cell 400V system with six BQ79616 connected in a daisy-chain fashion, the voltage measurement can return to the system in 2.5 ms with 1 Mbps baud rate where the channel-to-channel cell voltage measurement delta is only 120 microseconds. This reduced communication time frees up the ECU for other operations, and improves overall fault detection time tolerance.
Figure 2: Example of a daisy chain configuration with TI’s BQ796XX family of battery monitors
The inclusion of the isolated, bi-directional, daisy chain ports supports both capacitor- and transformer-based isolation, allowing the use of the most effective components for centralized or distribution architectures commonly found in the EV powertrain system. In addition, an isolated, differential daisy-chain communication interface allows the host to communicate with the entire battery stack over a single interface. In the event of a communication line break, the daisy-chain communication interface is configurable to a ring architecture that allows the host to talk to devices at either end of the stack.
A long-term, cost-effective noise filtering solution
By eliminating the need for external noise filtering components, engineers can improve measurement integrity and accuracy, channel-to-channel measurement synchronization and reduce the time for all measurements to return to the host. This process should also help generate an optimized, cost effective solution to help OEMs achieve 1% error on SOC and state-of-health (SOH) calculation targets. As these improvements continue to permeate the EV/HEV market, we will see confidence grow as more cost-effective and reliable products become available.
