Embedded Systems December 2000 Vol13_13

Issue link:

Contents of this Issue


Page 65 of 197

= A product also needs to be compat- ible with the devices that would tend to be used alongside it. If you are sell- ing components of a rack stereo sys- tem, the on/ off button for each one should have similar positioning. If you sell TVs and VCRs, the buttons for changing channels should work in a similar fashion . You mu t also attempt to be com- patible with the surrounding environ- ment. If the device will be used in a noisy environment, very quiet alarm sounds wi ll not be appropliate. On the other hand, a pocket calculator may be used by students in a quiet library, so you do not want it to make loud key- click noises. Sometimes you will not be able to guess the environment in advance. One automatic teller machine lobby that I have used has mirrored walls, similar to an elevator, to prevent customers feeling claustro- phobic. Unfortunately, the same mir- rored surface makes the user's key presses-inclucting their PIN-visible from almost anywhere in the lobby! Directed interfaces Some interfaces strongly suggest a direction. A question-and-answe r ses- ion provides an interaction where the direction is di ctated by the user inter- face. The use r may have the option to break out of the sequence, but the questions themselves suggest to the user that the next appropriate action is to provide an answer. That type of in teraction is considered directed. An example of a less directed interface is a car dashboard. You start the ignition and any numbelĀ· of things are avail- able for you to do. The sequence of putting it into first gear and pulling out of the parking space into traffic involves lots of options-you could have gone into reverse, turned on the lights and/ or wipers, and so on. The in terface does not suggest anyone path more than any other. This is a non-directed interface. You would make your interface more or less directed for a number of reasons. A more directed interface sui ts a device with a single simple goal. It also suits a novice, who will not want to make many decisions. A non-direct- ed interface provides more power to the user who knows how to navigate the device's features. They can go directly to the control that tlley want without having to follow a predefin ed sequence. Sometimes an interface will change from one to the other. An automatic teller machine will give you little or no options until you have entered your card and personal iden- tification number. This is as it should be. The machine should not provide any service to a user who attempts to bypass this step. Once tlle card has been validated, the interface can relax and allow the user a bit more fl exibili- ty. The interface is directed enough that the novice does not get lost, but more than one option is available at each step. Novice users usually prefe r direct- ed interfaces with an obviou path; non-directed interfaces are more pow- erful but more difficult to use. Multithreading Some interfaces allow many indepen- dent paths through different parts of the interface to be active simultane- ously. Consider a hi-fi system with a CD player, tape deck, tuner, and amplifier. The tape deck can rewind while the user is changing channels on the tuner. The amplifier may be instructed to take input from the CD player, so that is the audible source of music. Each piece of equipment is being allowed an independent thread of control. This is analogous to multi- threaded software. In most hi-fi sys- tems you would consider the ability to multiplex an accident of the design of independent components that are not capable of communicating with each other. A smarter system might change the amplifier to be directed to the last device that was used, and disable all others. This would avoid acciden tally leaving a tape playing when you are lis- tening to the radio. However, this also means that you lose the ability to 64 DECEMBER 2000 Embedded Systems Programming rewind a tape while listening to the radio- something users may find use- ful. The conclusion from this investi- gation of the multithreaded hi-fi: mul- tithreading makes use of the device more challenging, but offers more power to tlle user. Multithreading of an interface does not necessarily require the soft- ware to be multithreaded. The state of each thread of control can be stored separately, and events from each thread be handled by tlle same real- time task. Windowing systems typically have a separate tllfead of control for each application. Embedded systems have slightly different needs. Many process control devices consist of a settings area and a monitoring area. Allowing each one a thread of control helps reinforce in the user's mind tllat mon- itoring and control are two separate activities. Consider the interface to a furnace shown in Figure 3. While the use r is in the middle of typing in the new temperature, he may decide tllat he wants to check the average temper- aUlre. He can press the AVERAGE but- ton, and see the display change. The number he was entering on the key- pad is still valid. He can return to tll at activity and accept the new value. If tllere was a single thread of control, operations on the monitoring side would cancel incomplete actions on the settings side and vice versa. As an aside, the ACCEPT key in this control panel was deliberately kept as far as possible from tlle CLEAR key. These keys are often placed next to each other since they have related functions. Unfortunately on many sys- tems, the cost of pressing one instead of the other is quite high. In tllis case the ACCEPT key was placed beside the figure being accepted to encourage the user to look at tlle digits they have just entered before committing them. Modes Some interfaces require navigati ng through a numbe r of modes. Sometimes these modes will corre-

Articles in this issue

Archives of this issue

view archives of EETimes - Embedded Systems December 2000 Vol13_13