Integration challenges user-interface designers
Dara Sariaslani, Agilent Technologies- June 25, 2012There once was a time when measurement instruments did one thing and one thing only. When Bill Hewlett and Dave Packard designed their first product, an audio oscillator, it needed a simple user interface: three knobs and an on-off switch (Figure 1). The user needed to change frequency (two knobs) and amplitude (one knob). That's all.
Figure 1. The HP 200A Audio oscillator needed just three knobs, two for frequency and one for amplitude, plus an on/off switch.
Basic RF/microwave instruments such as spectrum analyzers initially performed one task; they provided a plot of amplitude versus frequency, but nothing else. Early VNAs (vector signal analyzers) measured S parameters only. Today's instruments, however, resemble Swiss Army knives in that they can perform many unrelated tasks. VNAs, for example, do much more than measure S parameters. Plus, wide range of engineers and technicians now use them. Versatile instruments, however, have complex internal designs. Their human-machine interactions require considerable research and design efforts.
Designing and developing new measurement science still garner the lion's share of R&D money from instrument manufacturers, but UI (user interface) design is gaining in importance. There are numerous examples of companies who excel at industrial and UI design, Apple in particular. Unfortunately, the RF/microwave test-and-measurement equipment aren't stellar examples of "ease-of-use." The problem arises because of the variety of uses and the diversity of users that each instrument must serve. A UI that appeals to R&D engineers might not appeal to production and test-automation engineers and technicians.
UI design challengesToday's instruments include much more hardware and software than single-function instruments of the past. Compare the Agilent PNA-X Series VNA in Figure 2 to the audio oscillator and you understand the challenge of designing a user interface. Instruments that perform many tasks broaden the instrument's user base and expand the measurement applications. Engineers routinely use VNAs to test simple ceramic filters and RF cables. But, did you know that VNAs are also used to measure moisture-absorption levels of diapers, to evaluate material properties of paleolithic ice core samples, and to test functional parameters of satellites? Such a wide range of uses results in a user base that includes every educational level from high school to PhD.
Figure 2. The Agilent N5245A PNA-X has front-panel controls for basic and advanced measurements.
Furthermore, some measurements take several days to run while in other applications, several hundred of parts are measured each day or even each hour. In some applications, the instrument is on a bench, used by many people for many different purposes. In others, the instrument isn't physically accessible and can only be monitored and controlled remotely. The user base also consists of engineers whose first exposure to personal computing and test equipment was an Apple I and an HP 8410. In contrast, younger engineers are now entering the workforce who have grown up with Nintendo and Xbox and are used to iPhones and iPads.
When measuring the S-parameters of a 1,2, or n-port device, you need to set start/stop frequencies range, power level, number of points, and IF Bandwidth. Next, turn on the actual required measurements (i.e. S11, S21, S22, ...) and then perform a calibration. At this point, the DUT can be connected and measured. An instrument whose sole purpose is to make such limited measurements can have a very simple front panel and UI. In reality, it isn't very economical to build such a single purpose benchtop instrument. This is why an instrument such as the PNA-X (Figure 2) has so many buttons and controls. But, if you compare the PNA-X to its predecessor, the 8510 network analyzer, you'd find that the 8510 performed fewer measurements, yet it had four more front-panel buttons than the PNA-X. In the PNA-X, we shifted functions from hardware (front panel keys) to software (menus, dialogs, and soft keys).
Addressing the UI challengeThere are many models and philosophies for capturing user requirements. One effective method, especially for RF/microwave instrument development, is having "Customer Evangelists," people who gain detailed knowledge about requirements and processes from one or more market sectors. Evangelists design requirements that address their customers' needs and they validate designs created by the development team against those requirements.
Customer evangelists may focus on the needs of just one market segment. Thus, there will be many competing requirements from other customer evangelists. In that case, a successful design/development team will need an architect who serves as the gatekeeper and mediator. For complex instrumentation, there may be two or more architects, specializing in hardware, software, and UI. Their job is to prioritize the requirements and apply a consistent set of design principles for addressing each requirement.
PrioritizationPrioritization is perhaps one of the toughest parts of UI design. It requires applying technical, financial, strategic, and market information all at the same time. When applied to UI design, prioritization translates to visibility and ease of use. One approach to prioritization and classification is to apply the "Traffic" model. Based on information from the user base (i.e. talking to the customer evangelist), the architect assesses how often and by what type of users a particular requirement or feature is used. The features accessed by most users, and those most often used by the least technical users, must be the simplest functions to operate. As an example, most modern VNA's can measure much more than simple S-Parameters, but measuring the four S-Parameters of a simple two-port network is still the most common task that a VNA performs. Agilent's PNA-X Network Analyzer is practically a whole rack of RF/microwave test instrumentation encapsulated into a single box. Yet, all four S-Parameters (S11, S21, S12, S22) are on the screen after pressing just three buttons.
CompartmentalizationTaking advantage of customer evangelists and applying prioritization is a key part of any effective UI design process, but the broad user base and varied applications that some RF/microwave instruments need to address create some unique challenges for UI designers. The main challenge is to prevent the features and capabilities designed to address one application or one particular set of users from overwhelming and confusing others. This is where the concept of compartmentalization and isolation comes into play.
A user interface must clearly delineate the different capabilities in the instrument to minimize confusion. Most modern RF/microwave instruments have the ability to make a wide array of measurements. A good UI shouldn't appear any more flexible than the underlying hardware. In fact, it is acceptable to limit capabilities to minimize the possibility of minimizing user mistakes and enhancing the overall user experience. Many UIs provide some "advanced" access to the unrestrained capabilities of the hardware for those who want to push the instrument while assuming the risks of unrestrained access. This approach isolates the typical user from the complexities needed to satisfy the more advanced users.
The Agilent PNA-X network analyzer, for example, UI uses compartments to separate groups of related functions. Figure 3 highlights the compartments on the N5242A PNA-X network analyzer. In addition to S-parameters, the PNA-X can measure IMD (intermodulation distortion), gain compression, noise figure, mixer/converter group delay, and many other nontraditional VNA measurements. Prior to the introduction of the PNA-X and other instruments, engineers made these measurements with stand-alone instruments or by customized test systems. There is an industry standard lexicon associated with each measurement and there are historically accepted standards for representing the results and the configuration parameters of each of these measurements.
Figure 3. Compartmentalization of front-panel controls keeps like functions grouped together, enhancing ease of use. Click on image to enlarge
The PNA-X UI design utilizes compartments to avoid confusion and cross-contamination of terms between different measurements that can all coexist in a single instrument setup. We base his compartmentalization on the "Measurement Class" concept (Figure 4) where the user assigns a specific role to a measurement channel. For example, the user can choose to create a new measurement channel as a noise-figure measurement class (Figure 5). With this selection, whenever this channel is the active element in the UI, all the relevant menus and dialogs alter to show only noise-figure-specific information.
Figure 4. The PNA-X's Measurement Class Selection Dialog groups measurements available to an input channel.
Figure 5. The noise figure setup provides access to only those functions needed to make the measurement.
Uniformity and ConsistencyA good UI is easy to learn and a UI designer should employ as many different learning techniques as possible to make the product acceptable to the widest possible audience. One common theory of learning is learning by repetition. If you repeatedly do the same thing, it eventually becomes second nature. Applying uniformity in the UI is one way of enforcing learning by repetition. One example of uniformity might be using the same UI widget for setting frequency any place in the interface, where a user might have to enter a frequency value.
In another theory of learning, called Associative Learning, the individual learns by making associations between common stimuli and responses. In instrument design, consistency matters. How does the system react to common stimuli exercised under different conditions? For example, what happens when a user clicks on a common button like "Cancel?" Do all the dialogs that have a Cancel button behave the same way? More importantly, do all of them behave in a consistent manner with the common expectations associated with the word "Cancel?" Figures 4 and 5 are also examples of consistency and uniformity. All dialogs have OK, Cancel, and Help buttons in the lower left edge of the dialog and Cancel always rolls back any changes made to any altered settings in the dialog.
When It Can't Be Designed From ScratchUnlike some consumer electronics, expensive RF/microwave test instruments can't be completely redesigned with every new technology advancement. One barrier to the sometimes overwhelming desire to go back to the drawing board is simple economics, both for instrument manufacturers and for their customers. These instruments usually represent a very large R&D investment by the instrument manufacturers and a large capital investment by customers. Therefore, all parties expect a long, continuously supported life for these instruments. It is not unusual (especially in the aerospace/defense industry) for engineers to incorporate instruments in test systems that they expect to last at least as long as 15 years. These types of programs usually represent a huge software investment and instrument makers must never break backwards compatibility. When instruments are incorporated into automated test stations, remote programming interfaces are just as important (if not more so) than panel controls. The same principles of prioritization, compartmentalization, uniformity, and consistency apply to the remote interface with the added caveat of not breaking backward compatibility.
When working with legacy platforms, keep in mind that the guiding UI design principle is to make incremental improvements rather than complete renovations. When you consider the longevity of some of the more popular instruments like the HP 8510 VNA or HP 856x spectrum analyzer, you realize that many of the designers that worked on these instruments probably spent most of their careers making incremental modifications. One key to making good incremental design changes is to maintain some level of familiarity with the previous design. Another key is to make the benefits of the new design clearly visible to the users and make sure that the new benefits are relevant and actually improve the users' experiences. It might be counterintuitive, but long-term users of any platform resent UI changes even if the original interface was not very good to begin with. That is, unless the changes translate to concrete benefits for them.
ConclusionsInstrument designers make many tradeoffs between capability and usability and we can all point to examples of successes and failures on our lab benches. Making the right tradeoffs is becoming more challenging as market and customer demands push us to integrate more capability into single instruments.
Instrument interfaces have been steadily evolving since the days of the HP 200A Audio Oscillator, but we're poised for an evolutionary jump. New technologies like high resolution, multi-touch displays and faster, more reliable wireless connectivity solutions may soon change the way we physically interact with RF/microwave instruments. More importantly, new advancements in measurement hardware and science will change our daily interactions with measurement instruments--be they on our lab benches, in our shirt pockets, or inside fully automated test systems. The RF/microwave measurement instruments of tomorrow will be a lot more autonomous and a lot less reliant on expert operators. T&MW
Dara Sariaslani Dara joined Agilent Technologies (Hewlett-Packard) in 1997 as a measurement engineer. He was the technical lead for several satellite payload test systems, responsible for designing the measurement automation infrastructure and several of the measurement routines. He joined the Component Test Division as the lead measurement engineer and was responsible for leading the development of some the earliest PNA applications. His current responsibilities include the requirement and definition planning for all of PNA-X's active applications including the nonlinear application, the NVNA. He holds BS degree in Physics and Computer Science from the California Polytechnic University in San Luis Obispo. E-mail: firstname.lastname@example.org.