As next-generation vehicles grow increasingly reliant on radar technologies to enhance driver and passenger safety, the margin of error for these advanced safety systems grows ever smaller. Yet the radar microcontrollers (MCUs) at the heart of these systems are servicing increasingly complex subsystems and applications, often doing so in harsh environmental conditions that push device tolerances to their limits.
Imagine, for example, the thermal stress on a radar MCU governing an autonomous emergency braking (AEB) system in an idling car in extreme summer heat—when the vehicle accelerates, there is no allowance for AEB latency while onboard cooling mechanisms dissipate the heat. The system must function instantaneously with real-time responsiveness to changing/challenging driving conditions. Even putting aside the safety implications of a latency incident or system failure, a radar system malfunction would at the very least require a burdensome vehicle repair, negatively impacting customer satisfaction.
To avoid issues like this, the highest possible levels of quality must be established at the MCU itself, the MCU fabrication process and the subsystems these MCUs are designed into. Let’s take a look at some of the factors in play here and the opportunities for improvement that leading companies like NXP embrace.
As the “brain” of the radar safety system, the MCU is called upon to provide consistently correct data under every circumstance. At the device level, NXP has demonstrated how intermittent system failures can be effectively neutralized at the MCU layer, where redundant cores and memory repositories help to ensure the utmost data integrity. Computations executed on one core are verified by other cores and decision trees are arbitrated accordingly. Similar redundancies are typically built into the subsystem where the MCU resides, helping to ensure, at multiple layers, that errors are identified and eliminated before they can be propagated to other onboard vehicle systems.
The key to this effort is our ability to understand our customers’ use cases in the most granular detail possible, inferring and re-creating the parameters of their test cases and software capabilities and ultimately validating the MCU for those precise use cases. The more we understand, the better we can build validation and test suites that fully test our systems under all conditions. What is the customer’s boot up sequence? How much do they exercise the individual cores? What sort of instructions do they run? Our ability to pinpoint these answers gives us the intelligence we need to fully test the MCU, analyze the resulting data and feed this data back into our own development workflows to ensure continuous improvement for the next generations of MCUs that follow.
When it comes to managing processes with our fabrication partners to assure and enhance MCU quality, NXP subscribes to the old proverb: “Trust, but verify.” Our own heritage with fab operations gives us the ability to discern the suppliers demonstrating the highest levels of process discipline. This innate expertise also affords us the ability to guide our suppliers through any needed process refinements.
This requires tight collaboration and communication from senior management to engineering, with the understanding that NXP will be heavily involved in the testing, validation and technology characterization phases—and we will intervene wherever we see the need. Long before we initiate production processes, we’ll test our products on the targeted fab flows, characterize the variability, customize processes as needed and do the hard work required to achieve optimal process uniformity so as to minimize any variabilities that could compromise MCU product quality.
NXP’s fabrication expertise also comes into play when quality issues arise with our suppliers that must be remedied ASAP—holistically and transparently. Our close collaboration with our suppliers enables us to move quickly to identify an issue when it occurs, rapidly contain it and perform the necessary root cause analysis. When an issue arises, we have a well-defined process in place that will shut down the line and limit the issue to the smallest subset of materials as possible.
Equally importantly, we proactively communicate these fabrication issues with our customers in a timely fashion. When a quality failure is detected, we don’t deliberate on an ‘acceptable threshold’ for the number of failures they might experience before we raise the red flag. The goal is to communicate and rectify the issue long before any of the affected devices might find their way to our customers’ customers. We’re aided in this capability by an extensive, highly-detailed traceability system that allows us to track each individual MCU from production to customer hands.
As most in our industry know, quality performance is typically measured with a parts per million (PPM) metric, whereby we calculate the number of defective devices among a batch of one million. Zero defects per million is the goal, of course, but difficult with high volume manufacturing.
NXP drives a culture of continuous improvement across all internal and external factories. This is a reflection of our desire to meet and surpass our customers’ quality expectations when we say that quality excellence is a journey, not a destination. As part of this journey, we’re proud to say that our PPM metric has improved >60% since ADAS high volume manufacturing began in 2013. This is the continuous improvement that we strive for.