For alerting an operator, all beeps weren't created equal
Most people think of embedded designs as incorporating microprocessors, memory and
maybe a display of some type. Another feature common to most designs but that most people
probably give little thought to is an audio indicator. Consider for a moment all the tones
and noise you encounter every day-from automobiles, to kitchen appliances, telephones,
elevators (not even counting Muzak), automated-teller machines, fever thermometers and
even your PC. Who decides how devices use tones, their volume and what information they're
trying to convey? From a designer's point of view, we don't need to answer all these
questions. Sometimes we can rely on others to indicate what the customer needs and wants.
However, it behooves us to understand some of the rationale behind this ever-present noise
because we don't design products for our own gratification but to solve a customer's
It's just a beeper
How many times have you been told to put an audio alarm into a product or add a beeper
to indicate each time the operator presses a key? If your experiences are anything like
mine, these requests come up in just about every product design. However, when the boss
tells you to add an audio indicator I'll bet that most times he doesn't even tell you how
loud to make it-let alone specify its frequency, duration, attack or decay. The only
guidance given is to make it cheap. So following the boss' directions, the design now has
an audible indicator. The remaining question is how does it meet marketplace needs?
As an example of how badly a design can fail, consider what happened when I recently
repaired a sailboat autopilot for a friend's father. During the repair process, I switched
on the device's alarm, and the owner remarked that he didn't realize that the autopilot
had an alarm. As I completed the repairs and reinstalled the unit back into its housing
the reason why became clear-the designer missed two vital facts: First, most people who
can afford sailboats with autopilots are older folks. Second, one consequence of getting
older is high-frequency hearing loss.
As a result, the manufacturer made two significant errors-the frequency was wrong, and
the designer buried the audio transducer deep inside a sealed weatherproof box. The
instrument uses a piezoceramic audio transducer with a frequency somewhere near 2000 Hz.
While the autopilot was out of the box, the alarm was loud enough that the owner heard it
easily. Once I closed up its weatherproof enclosure, the sound became so muted that I
could barely hear it at my workbench never mind someone sailing in a storm! The proper
approach to designing the alarm would be to either use a lower frequency that propagates
well through obstacles or incorporate a weatherproof audio transducer into the
While on the topic of frequency, have you ever noticed how difficult it is to localize
a high-frequency noise? Try doing it sometime when visiting a movie theater, and at the
top of the hour try to figure out whose watch just beeped. High-frequency noise is almost
impossible to localize compared to a lower frequency. For example, one day I was working
on my sailboat at its dock when I heard a piezoceramic transducer turn on. It sounded as
if it was coming from my cabin, but I didn't own any device that could make that noise.
With the beeper still sounding, I wandered up to the parking lot (about 150 yards away)
because I could've sworn the noise was coming from there. It wasn't. It turns out that it
came from another dock. From my experience, keep tones below 1000 Hz if you want someone
to find your instrument in a crowd.
And the meaning is...
To prevent such problems, remember that an audible indicator's purpose is to convey
information. If users can't hear it, if it makes noise at inappropriate times or if it
conveys the wrong information, the indicator is worthless. Back in the '70s, whether right
or wrong, people frantically disconnected their seat-belt warning alarms because the
beepers never stopped sounding off. The alarm's purpose was to remind people to buckle
their seat belts and not to stop them from driving. I too disconnected my alarm but for a
slightly different reason. In the winter, my diesel vehicle takes several minutes to warm
up the glow plugs before I can start the engine. When I get in the car, I immediately turn
on the ignition to start warming the plugs. Then I put on my seat belt. I use this
sequence to minimize the time sitting in a freezing car before I can start the engine.
An example of a good audio annunciator is the old-fashioned mechanical telephone ringer
because it properly conveys its information. The designers purposely created it to be
annoying and essentially to cause panic attacks in people so someone would answer the
phone. I haven't met many people who can ignore a mechanical telephone ringer. Not only
does it elicit a Pavlovian response to pick up the handset, I've heard that the slight
audible differences due to the mechanical system allowed people in a crowded office to
tell from across the room when their phone rang. That's a lot of information for a simple
bell to convey.
A call to arms
A lot of the work my company does is in the medical instrumentation field, which
requires that all instrumentation somehow inform the user if something serious is
happening or if the instrument needs attention. Unfortunately, it's gotten to the point
that no one really trusts these beeps, bells and whistles. All these instruments are
clamoring for attention while the poor hospital worker is trying to help the patient. The
clinician is left with the choice of either caring for the patient or servicing the
instrument's perceived problem. For this reason and many others, most clinicians try to
permanently silence these alarms.
With all the debate going on about national health care, you'll see plenty of news
reports taped in hospitals. Forget the dialog and leave politics aside, listen to all the
alarms going off in the background and then notice that no one is responding to them. It's
not the clinicians' fault-they've become desensitized to the urgency because the alarms
aren't conveying the right information.
The medical industry conducts many studies to help alleviate this onslaught of noise in
hospitals. I'm sure other areas, such as the nuclear industry, must have similar problems.
Does your industry have any standards or proposed standards? In an effort to address the
alarm-overload problem in hospitals, the European Economic Community has proposed a
standard that goes into effect January 1995.
Aimed at standardizing the audible alarms medical
instrumentation use, a European Economic Community
standard defines the tone attack/decay times and
In summary, it outlines four different "alarms" that instruments must use to
audibly indicate information and either a low, medium or high priority. Each alarm has a
different pulse and repetition sequence as well as a specific attack and decay times so as
to avoid a startle response (especially from the patient). Once instruments adopting this
standard reach the market, clinicians should theoretically be able to know how serious an
alarm is without looking at the instrument, and long repetition rates won't fray their
It sounds like it should be easy to implement audio indicators that conform to this
standard-just plug in a sound-generator chip. However, getting everything working right is
a task that's far from trivial. Next time I'll describe how we solved this problem,
including all the gory details of the wrong turns we took. PE&IN