No quality control system is perfect. Just think “Toyota” in the context of today’s headlines and that should be abundantly clear.
Taking Toyota’s predicament to heart—for any quality manager in any industry—means thinking through where the inherent weaknesses are in production processes and product design. Are there inherent weaknesses in any electronic system? Yes. Are mechanical systems failure-proof? No. Any and every system can fail—and it is up to quality management to accurately understand the real failure thresholds that they must manage.
For example, few of us ever question the correctness of calculations made on a computer but, as experts in the field of computer science and numerical analysis know, computers are also subject to their own quality scrutiny. The finite precision of any microprocessor and the assumptions used in computer language compilers, to cite two examples, can have a profound effect on the correctness of calculations. Indeed, whether on your laptop or the fastest supercomputer, speed and accuracy often compete with each other. Or, as Rob Meyer, CEO of the Numerical Algorithms Group, one of the world’s oldest software organizations, wryly comments on those who seek performance at all costs, “Just how fast do you want the wrong answer?”
When it comes to data management, the Food and Drug Administration (FDA) certainly knows better. That’s why the extensive 21 CFR Part 11 regulations spell out protocols for proving that automated systems record keeping are accurate and tamper proof.
As the manufacturer of what are arguably the most accurate relative humidity (RH) monitoring and alarming instruments in the world, Veriteq takes to heart the inherent real world limitations on RH measurement accuracy and urges all quality managers whose products or processes are affected by RH to do the same. With so many instruments available in the marketplace that state similar specifications for RH measurement accuracy, it's easy to think they are all the same. Nothing could be further from the truth.
Following are some plain facts about what you need to understand about RH sensor accuracy.
First, RH sensors are “air breathers,” vulnerable to a wide range of pollutants. Like tiny sponges, they absorb water vapor from the air. To function properly, RH sensors must maintain intimate contact with the environment. Unfortunately, this exposure leaves sensors vulnerable to air-borne contaminants such as chemicals and cleaners, which can coat or permanently damage a sensor’s surface and thereby prevent it from properly absorbing water vapor. Ultimately these pollutants can distort an RH sensor’s signal.
Second, even simple condensation can affect an RH sensor’s accuracy. If the door to a high humidity environment is opened, for example, condensation may form on the RH sensor inside. Long after the RH reading appears normal, the sensor can remain wet internally causing an offset in value and it may need to be removed and dried before it can once again provide accurate readings.
Third, and the most common occurrence, is sensor drift. The accuracy of all sensors drift from their initial values until the next time the device is calibrated. Understanding the amount of drift is more important than the initial accuracy. If you are out of spec when it’s time to calibrate then all the preceding data could be called into question; and calibration is the only way to know if the humidity sensor has drifted.
Not all measuring systems are created equal because they don’t necessarily manage the above mentioned factors affecting the accuracy of RH sensors. For example, if an RH sensor has been damaged or contaminated, it might send a signal of 4 V, indicating 40 percent humidity—and the system display would reflect this—when in reality the relative humidity might be just 38 percent. As this sensor drifts over time, it may continue sending a 4 V signal, when RH is just 36 percent, then 34 percent, and so on until the RH is well outside its pre-defined range of 35 percent to 45 percent. If the system includes an alarm for when RH goes out of spec, it relies on the sensor signal. If the signal remains at 4 V, the alarm system will not be activated and the system display and recorder will look normal. The two ways to combat RH drift is to measure RH independent of the sensor used to control humidity and use a device where accuracy is known over time.
This should give you insight into why the best-in-class RH monitoring instruments available in the marketplace always consider sensor drift in their specifications. If controlled environments involve products that are expensive to replace or to repeat testing, then you should consider accuracy over time: initial accuracy (as left) and one-year accuracy (as found). For a more detailed discussion, please see www.veriteq.com/download/whitepaper/catching-the-drift.htm.
The point is that every system has vulnerabilities and you either know the exact limitations of your underlying systems or you need to “go to school” so you better know exactly what you are dealing with. To paraphrase Meyer’s quote, “How long before you find out you had the wrong answer?”
Any quality manager taking Toyota’s current troubles to heart, should be thinking along these lines. Learn the inherent limitations in a measurement system. Perform a due diligence investigation to see if the manufacturers of the instruments and systems that you use to safeguard quality have taken these inherent limitations into careful consideration in their product designs.
Although we don’t know the exact cause of Toyota’s quality failures, or the eventual price tag that this design or quality control failure will be, we cansurmise that the true limitations of one or another electro-mechanical system were not truly understood. Take Toyota’s woes to heart by taking a look with new eyes at real-world limitations on the systems and processes that your organization has in place.