Quality and reliability–they’re not the same

Quality measures the manufacturing process while reliability evaluates the effect of part quality over time. Learn about the units of each and how each measurement contributes to the product performance you seek.

  Quality and reliability are often used interchangeably to describe product failure rates but the two terms actually describe two entirely different aspects of the behavior of a part over its lifetime. In this article, we will discuss these metrics in detail to help you better understand which to apply when researching components for use in your applications.

 

Quality
When an engineering team gets ready to take a product to manufacturing, they summarize all of its characteristics in a specification. The quality of a part is a measure of how closely that part meets those specifications as defined in a product data sheet.   You can think of quality as an assessment of the manufacturing and test process — how well does the facility produce parts that match the specifications? How often does the process produce faulty parts?

 

We define part quality using defective parts per million (DPPM) as expressed by

 

V2N1 Quality_equation 1

 

DPPM can be used to quantify the overall quality level of a specific product or technology, or it can be used to quantify the expected fallout of a specific failure category. Manufacturing line fallout or 0 km fallout are two examples of customer complaints that could be categorized as quality problems.

 

To ensure the effectiveness of its manufacturing processes, the Cypress Flash Business Unit (Flash BU) has developed an outgoing product quality program for quality monitoring. This program tests a random sampling of parts to assess whether they meet specifications. Products undergo testing across all datasheet parameters at minimum, maximum, and room temperatures. A complex sampling plan ensures weighted coverage across the entire yield distribution, and we express failure rates in terms of DPPM. We use this data to drive continuous improvement in the outgoing quality of our products and technologies.

 

Reliability

 

We use the term quality to express how well our parts perform against our specifications, typically at time zero. Reliability can be thought of as the measure of quality over time. Because reliability introduces a time component to the quality equation, we use failures in time (FIT) as the unit of measure. FIT expresses reliability as the number of failures per some established time interval, typically one billion device hours.  We define FIT as

 

V2N1 Quality_equation 2 (i.e. number of failures per 1,000,000 devices operated for 1000 hours).

 

It is important to note is that there is no such thing as an "average" DPPM. If you want to quantify reliability, you should use units of FIT, which can be converted to DPPM for any given point in time. Here is a simple example to illustrate this point: For a device operated with a 100% duty cycle, a fail rate of 2 FIT translates to approximately 17.5 DPPM at one year, about 87.6 DPPM at five years, and roughly 175 DPPM at 10 years.

 

The failure rate for semiconductor devices typically follows the classic bathtub curve (see figure 1). The lifecycle begins with an initial failure rate that drops off rapidly during a time period known as early life. These failures represent defective subpopulations. The failure mechanism in these devices is not common to the general population but is typically a result of error or defects. At the Cypress Flash BU, we classify early life as 4000 field-equivalent hours (FEHs), which is defined as life test hours multiplied by acceleration factor for the mechanism. We report the early life failure rate in FITs.

 

V2N1 Quality_fig1

Figure 1: Plot of failure rate as a function of time assumes a characteristic bathtub curve. This reflects an early life period with a decreasing failure rate; transitioning at 4000 hours (red line) to a low, steady-state failure rate; transitioning to an increasing failure rate (blue line) as it moves into the end-of-life phase.

 

Above 4000 hours, the lifecycle process transitions into inherent life, a period characterized by a low, constant failure rate. Finally, at the end of the lifecycle, failure rates began to climb again as the part enters the end-of-life period.

 

To ensure reliability, the Cypress Flash BU has developed the Qualification Maintenance Program (QMP). This program measures the reliability of all process technologies on a regular basis. Typically, about 20,000 devices per month are subjected to a battery of reliability stress tests. We use QMP to improve device reliability by establishing the root causes of failure in devices that do not pass. The program also provides data on device lifetime and the performance of plastic packaging under variations in temperature and humidity.

 

Qualification maintenance testing is conducted on representative samples of devices from each process technology. Sampling is not limited to process technologies from in-house wafer fabs, but also includes Cypress Flash BU-certified external foundries. The program selects devices on the basis of complexity, production volume and strategic importance. The goal is to assess process, package and product reliability.

 

In closing, let's say it again — quality and reliability are two separate quantities with distinct definitions and purposes. Expressed in units of DPPM, quality measures the effectiveness of the production process and how well the finished parts meet specifications. Reliability provides a measure of quality over time and is expressed in units of FIT. Armed with this knowledge and with documentation from the Cypress Flash BU, you should be able to specify and purchase parts that will serve the application and satisfy your customers.

 

Further reading
1. Spansion Flash BU Quarterly Reliability Report

 

Also in this issue:

Leverage MCUs to meet the wearables challenge, Whistle canine activity monitor tracks doggie doings, Energy harvesting starter kit makes it easy to go battery freeSolving recurrent error codes, Tetra SDK speeds wearables to market


Get More from Core & Code Subscribe
image_pdf

Leave a Reply

Your email address will not be published. Required fields are marked *


Other stories in this issue

Product Spotlight

Energy harvesting starter kit makes it easy to go battery free

With the problem of powering such small devices out of the way, teams can put all their creativity into innovative wearables like a child’s shoe that integrates a location sensor.