|
A client was controlling a fan using
Pulse-Width-Modulated (PWM) signal generated in firmware
on a microcontroller, but the characteristics of that
PWM signal were discovered to be causing sounds which
were unacceptable to customers. A different frequency
would have to be implemented, but the designer wasn't
sure what frequency would accomplish the task best.
Making this change in firmware could be time consuming,
so it would be wise to choose the best signal frequency
before having firmware written, compiled, installed, and
tested. We needed to create a PWM signal
which matched the original, and provide an interface for
trying many different PWM signals. Additionally, we
needed to take some measurements to characterize each
value to see if it would meet the customer's
specifications.
Thus, we would need to be able to generate a digital
pulse train at any desired frequency from 5 Hz to 100
kHz, and to adjust the duty cycle of the signal from 10%
to 100%. After each adjustment, we would need to measure
the frequency of the fan's tachometer, and record either
a sound or accelerometer to be sure the resonances were
minimized.
A benchtop function generator proved the problem to
management, but the process was very slow and time
consuming. To move forward into solving the problem, a
laptop running LabVIEW and a Data Acquisition Card were
brought into the lab. Within about 1 hour a simple
ad-hoc application and some wiring were ready to begin
testing. The test changed frequency incrementally, and
at each frequency tested several duty cycles.
For an inexpensive, portable, flexible test, a PCMCIA
board was chosen for use in a laptop. The DAQ-Card 6024E
offered 2 counter/timers, which allowed for both
generating the PWM and for testing the frequency of the
fan's tachometer. The analog inputs allowed us to sample
a microphone and extract the sound amplitude and
frequency. LabVIEW 7.1 was used to write the user
interface to do some testing manually, and was later
used to cycle through a range of frequencies and duty
cycle. For connectivity a BNC-2120 was chosen and made
setup and changes quick and easy.
The complete test took just about 3 hours, and when
it was complete, it was clear that several resonant
frequencies were very bad choices, and allowed choosing
a frequency which provided smooth fan motion and minimal
noise.
Making and testing these changes in firmware might
have taken weeks to finish. The engineering staff was
unsure what PWM frequency would do the job, and each
guess would require about a day to write, compile,
upload, and try. Now, the complete solution was found
before lunch, and once tested and proven out, the final
implementation could possibly have been done the same
day!
|
