January
1999, Issue 102
Wires,
Wires Everywhere
The
RF Solution
TUNING
IN
Getting
on the airwaves design-wise is a relatively simple (at
least conceptually) four-step process, followed by testing
and tweaking to optimize range. But the devil is in
the details, so let’s step through the process focusing
on the areas where gotchas might arise.
The
first step is establishing the basic timing, which is
usually a cart or horse decision. If the MICRF001 listens
to an existing transmitter, that unit’s frequency determines
the clock required on the REFOSC input, which is stepped
up internally. The datasheet shows the defining equations
(e.g., how working with a 315-MHz transmitter calls
for 2.4092 MHz on REFOSC).
If
you have the luxury of choosing the transmitter frequency
(i.e., tuned LC), you can start with a standard value
(3.000 MHz) for REFOSC and work backward. This technique
can save you the cost associated with a nonstandard
clock source and the hassle of having to deal with any
fractional errors in subsequent calculations.
Another
timing decision involves jumpering the SEL0 and SEL1
pins. Though the configuration inherently dictates the
baseband (i.e., out the DO pin) data rate, it’s not
simply a matter of choosing the fastest or most convenient
digital connection. Rather, you have to find a baseband
filter bandwidth that depends on the minimum pulsewidth,
not the data rate.
Here,
for the first (but not the last) time, the issue of
the data coding comes into play because it impacts the
relationship between data rate and pulse width. As an
extreme, NRZ coding, which represents a 1 or 0 as a
corresponding level, won’t work because a 0 is equivalent
to silence. Instead, a PWM type coding is required,
in which each bit consists of high and low levels.
Consider
typical codings like the 33/66 shown in Figure 3 and
50/50 (e.g., Manchester coding). Even though the base-band
data rate may be the same, the former has a shorter
minimum pulse width (33% vs. 50%) calling for a higher
filter bandwidth.
|
(Click
here to enlarge)
|
Figure
3—The way data is coded impacts performance
with minimum pulse width, a key parameter. One popular
way to avoid narrow pulses, which are hard to distinguish
from noise, is to use a 33/66% PWM scheme. |
Because
more bandwidth lets in more noise and reduces range,
50/50 schemes are preferred and although 66/33 are OK,
something like 90/10 should be avoided. The datasheet
contains equations and tables to calculate the optimal
SEL0 and SEL1 setting depending on transmit frequency
and minimum pulse width.
