Unmanned Systems Technology Dec/Jan 2020 | Phoenix UAS | Sonar focus | Construction insight | InterGeo 2019 | Supacat ATMP | Adelan fuel cell | Oregon tour | DSEI 2019 | Copperstone Helix | Power management focus

89 accuracy. This is resolved by using the measurements in different ways. The battery state is determined by voltage to within ±1 mV at 400 or 800 V for the performance measurement, and ±15 mV for the secondary safety measurement. The accuracy required for safety is to prevent overcharging or over- discharging, which has to be known to 10- 20 mV. To get the most performance out of the pack meanwhile needs an accuracy of 1-2 mV to maximise the vehicle’s range. The range or operation time depends on how much energy can be safely stored and extracted from the battery pack, so the more accurate the measurement electronics, the more energy that can be safely stored and extracted. Both measurement paths use a delta sigma analogue-to-digital converter (ADC) with 16-bit resolution. This encodes the change, or delta, of the signal, rather than the absolute value, with a 1-bit sigma converter repeating the process 16 times to get the final value. That means the reference point is vital, as the difference in accuracy between the two approaches comes from the voltage reference. For the performance signal chain, a Zener voltage reference provides the highest accuracy, but that is more difficult and costly to design and implement on the chip. For the secondary, safety signal chain a delta sigma ADC with a different layout and design uses a bandgap voltage reference. Using completely different designs means that an environmental effect such as a change in temperature or humidity has a different effect on the primary and secondary chains. These are expected to be roughly in the same 15 mV range, and if the measurements do not agree to that level then the host controller must decide what to do. That then leads to the system-level software architecture. If there is a critical fault in a driverless car for example, do you warn the passengers, then bring the vehicle to a safe resting space? If so, there is a series of events at the system level to determine the next step. This is a key part of the development of the power management system. Driverless taxis in ride-sharing applications without an operator cannot be allowed to catch fire during use, as has happened with other electric cars. Ensuring the correct and safe operation of the battery pack is an essential part of this. When it comes to safety, the decision has to be made instantly. The fault-tolerant time interval is usually a few hundred milliseconds once a fault is detected, so for guaranteeing safety those decisions have to be in the local control unit. If a power management chip fails, all is not lost. As long as a failure is detected then the situation is still safe, but the question then becomes whether the system is operational. Fail safe, fail operational – that mandates the second piece of independent hardware. Voltage is not the only measurement for power management. Before a battery fails, high pressure can develop in the battery, or it can rupture and emit gas, so designers are looking at gas detection to measure the cell pressure. One way to do that is to use ultrasonic waves to look inside the battery for swelling and delamination, as an air gap will have a different reflection from a normal battery. The challenge then is how to economically design a large multi- channel ultrasonic sensor for 20 cells to give a reliable picture of the state of the battery. That would provide a lot of information that needs to be analysed. Extending it to 100 or 200 cells potentially means having to handle 10 times the data, which would be difficult to transfer and process. Power management | Focus Driverless taxis in ride-sharing applications can’t be allowed to catch fire during use, as has happened with other electric cars Unmanned Systems Technology | December/January 2020 The state-of-charge value is critical, as it is used to determine vehicle range and cycle life. A typical automotive application will cycle a battery from 30% (minimum) to 80% (maximum) state of charge to maximise the cycle life (Courtesy of Renesas Electronics)