Understanding Crosstalk - Understanding Signal Integrity - page 114

amplitude. NEXT will always be present because, unlike with FEXT, the mutual
inductance and capacitance terms cannot cancel each other.
As shown in the Problems, when more than one trace aggresses a victim, L m
and C m represent the sum of the mutuals from all the aggressing traces.
Equations (10.2b) and (10.2d) show that, when the coupled length is small,
NEXT grows by increasing the length of the coupled region ( tpd ), by increasing
voltage swing of the aggressor ( dV ), or by reducing the aggressor rise time ( dt ).
It is apparent from (10.2c) that for long lines (defined as where the signal rise
time is less than twice the coupled length) the NEXT amplitude is only affected by
Kb and by the aggressor's voltage swing. Therefore, NEXT is combated by lower-
ing the coupling between traces since that lowers Kb , keeping the coupled length
short, reducing the voltage swing of the aggressor, and lengthening its rise time.

10.4.2 What Are Typical Kb Values?
Reverse crosstalk is shown in Figure 10.6 for a pair of 65

5-mil
(0.13-mm)-wide, half-ounce thick striplines and solder mask-covered microstrips
on FR4 (

Ω and 50 Ω

ε r = 4.2). The separation is given as multiples of the trace width: a separa-
tion of 3 represents 15 mils since the trace is 5 mils wide (or 0.39 mm for a 0.13-mm
trace).
It is evident that Kb (and reverse crosstalk) is greater for higher-impedance
traces, and reverse crosstalk may be larger in stripline than in microstrip.
To estimate the reverse crosstalk amplitude without using a field solver to de-
termine L m and C m , Kb is found from the figure and is used in either (10.2b) or
(10.2c), depending on the length of line. The Problems present this in more detail,

0.13

0.12

0.11

0.10

65 Ω

0.09

0.08

0.07

0.06

50 Ω

0.05

0.04

0.03

0.02

0.01

0.00

2 4 6 8 10

Separation (trace width multiples)

Figure 10.6 Kb to an adjacent trace for 5-mil (0.13-mm)-wide half-ounce stripline (solid lines) and
solder mask-covered microstrip (dotted lines) on FR4. (Raw data from Linpar [12].)