[Design Application]
Accelerated Design Verification And VXI Make A Perfect Match
Need To Account For Unexpected Operating Environments? The VXIbus Is The Way To Go For Hardware-In-The-Loop Testing Techniques.
To paraphrase a famous quote, electronic devices, and embedded systems in particular, sometimes boldly go where no other device has gone before. Simply put, the devices find themselves in a multitude of unexpected operating environments—some of which may not have been intended by the design engineer.
How, then, do design teams verify a new device's operation over the breadth of situations it may encounter? To answer this question, it's beneficial to understand the interdependence between the fault spectrum and the use model. One potential technique, adapted from software design, is hardware-in- the-loop, or accelerated-design-verification, testing. Thanks to recent advances in computational and measurement technology, the VXIbus (VME eXtensions for Instrumentation bus) lends itself to hardware-in-the- loop testing as a means to gain sufficient information regarding the fault spectrum in a dynamic environment.
Figure 1 shows a map of the size of the fault spectrum as it relates to the complexity of an operational environment for an electronic device. In this case, as the operational environment becomes more complex, the fault spectrum increases. An example might be the use of a personal computer in the confines of a home, versus one used as an operational component onboard a jet aircraft. The noise, shock, and cooling issues of the jet may precipitate more faults than the PC used in the home. This is depicted here as a larger circle when the temperature increases. This assumes, of course, a benign home environment rather than one where a two-year-old places a peanut butter sandwich in the fan on the back of the PC.
An excellent example of accelerated testing is illustrated in the automotive industry. Every July, auto racers from around the world arrive in Colorado Springs, Colo., to face the challenges of a race whose venue is the beautiful, but rugged, terrain of the Rocky Mountains. The challenge for both car and driver is the Race to the Clouds, a grueling hill climb to the summit of 14,110-ft Pike's Peak. The course is a twisting and winding gravel road called the Pike's Peak Highway. In just over 12 mi., drivers will experience a 10,000-ft change in elevation, and must manage their cars through 156 hairpin turns.
Casual observers might wonder about the race team's concern for the design of their car's electronic control module (ECM). Each second, this module relates the oxygen sensor input to the torque, and mixture of gas and air as the altitude increases. An engine that starts to lose power as the finish line approaches reflects poorly on the module and its designers.
A highly complex and dynamic environment such as this precipitates incipient failures in engine control modules. After talking with a number of ECM designers, we made a proposal regarding a design-verification scheme that provided greater fault coverage. The idea was to subject the ECM to electronic emulation of the race's environmental conditions. The system would emulate the signals for the ECM with very-large changes in altitude, temperature, pressure, and throttle, along with dynamic changes in the revolutions per minute. This accelerated test process would precipitate incipient failures. The goal was to test design alternatives before moving to manufacturing.
One other example points out that this style of verification can cross industry segments. Designers of telecom amplifiers have reached quality production rates with faults in the parts-per-million (PPM) range. It began when members of an engineering team were confronted by an experienced technician who had the reputation of "knowing" which amplifiers would fail first. Much to the designers' amazement, the track record for this technician was indisputable. After much cajoling, the technician let the engineers in on the secret. The technician would put the amplifiers into operation, then loosen the connectors and monitor which amplifiers provided the best output.
In this case, the technician would make a measurement of the outgoing signal on an oscilloscope with a degraded signal on the input. If the amp gave an output that was sustained within the grease marks on the scope, the technician was fairly certain the amplifier would stand up better than the other amplifiers.
This gave the designers an idea about how to improve the quality of the amplifiers. They used the more-difficult, degraded-signal test as part of their design verification when selecting alternate amplifier designs to pass on to manufacturing. By doing so, the designers were able to precipitate failure modes early in the design process, and not in the field.
In an eye diagram, for example, amplifiers within the light-green area yield better longevity (Fig. 2). This saves time, money, and resources. Now, instead of failure rates measured in the 1% range, designers were boasting of failures in the parts-per-million range.
The idea of accelerated testing is not just focused on temperature or the signals found in a mechatronic device. Another example is accelerating the crystal frequency of an oscillator to higher and lower frequencies. The idea is to uncover hidden faults by pushing and pulling the system into a metastable state. This may mean changing the clock frequency in a dynamic way, rather than with simple static steps. That is, the designer might consider increasing and decreasing the frequency as if it were modulated by 60, 120, or 400 Hz. Designers could vary the power to the board by oscillating the supply voltages, or injecting larger surges as might be found on a "poor-quality" power line.
Reprints
