Datasheet

SZZA016B
7–242
Basic Design Considerations for Backplanes
The dc analysis can help provide the designer with best-case low-level voltage (V
OL1
) and
worst-case (V
OL2
) signal levels expected at the receivers on a backplane when the termination
resistance has been determined.
The V
OL
levels affect the noise margins at all receivers as shown in Figure 10. The signal at
point B is at the last receiver at the end of a 10-in.-long transmission line. The low level of this
signal is higher than that of point A (less lower noise margin). Smaller values of R
TT
or longer
backplanes (higher values of R
line
) will reduce point-B noise margin even more. The drive
capability (characterized by R
device
) of the transmitter also will affect the waveforms V
OL
.
Effect of Changing Stub Length on Backplane Signal Characteristics
The effects of changing stub length are most noticeable in the stub associated with the
transceiver that drives the signal and, to a lesser extent, on the stubs at the transceivers that are
receiving the signal. These effects are in two categories:
• Flight time – The longer the stub, the longer it takes for a signal to propagate through it, and
results in increased flight time from the driver to the backplane line (stub delay).
• Rise time [also known as slew rate (V/ns)] – One of the interesting effects of the stub is the
faster rise time observed at the driver circuit as stub Z
o
, or stub length, increases.
The inductance of the stub and connector form an L-C-R network between the driver and the
load (backplane). Figure 12 shows a simplified equivalent circuit.
R
TT
/2
C
o
+ C
d
V
TT
L
o
+ L
conn.
Figure 12. Thevenin Equivalent of Load
The longer the stub length, or the higher the stub Z
o
, the larger is the inductance seen by the
driver [the sum of the stub line inductance (L
o
) and the connector inductance (L
conn.
)] and, thus,
the faster the rise time of the driving waveform. The faster rise time causes increased ringback
(increased reflections) and worsens the signal integrity of the system. Therefore, it is best to use
a low stub Z
o
and keep the length as short as possible, preferably less than 1 in.
The following analyses are based on HSPICE simulations of the backplane model (see
Figure 2). Figure 13 shows the results of simulation data taken on rise time when only the stub
Z
o
was changed. The termination resistance used in the calculation was also changed with each
new value of stub impedance, because this changes the effective characteristic impedance on
the backplane. S1 is the rise time at the driver measured from 20% to 80% steady-state low and
high levels. S2 is the rise time measured at the beginning of the backplane. S3 is the measured
rise time when the signal leaves the backplane at the last receiver slot. S4 is the measured rise
time at the last receiver.