English
Language : 

ISL59530_11 Datasheet, PDF (22/26 Pages) Intersil Corporation – 16x16 Video Crosspoint
ISL59530
For this reason, the ISL59530 must be in DC-coupled
mode (Clamp Disabled) to be compatible with S-video
and component video signals.
Bandwidth Considerations
Wide frequency response (high bandwidth) in a video
system means better video resolution. Four sets of
frequency response curves are shown in Figure 47.
Depending on the switch configurations and the routing (the
path from the input to the output), bandwidth can vary
between 100MHz and 350MHz. A short discussion of the
trade-offs — including matrix configuration, output buffer
gain selection, channel selection, and loading follows.
2
MUX, AV = 2
0
BROADCAST,
-2
AV = 1
MUX, AV = 1
-4
BROADCAST,
AV = 2
-6
-8
-10
1M
10M
100M
1G
FREQUENCY (Hz)
FIGURE 47. FREQUENCY RESPONSE FOR VARIOUS MODES
In multiplexer mode, one input typically drives one output
channel, while in broadcast mode, one input drives all 16
outputs. As the number of outputs driven increases, the
parasitic loading on that input increases. Broadcast Mode is
the worst-case, where the capacitance of all 16 channels
loads one input, reducing the overall bandwidth. In addition,
due to internal device compensation, an output buffer gain of
x2 has higher bandwidth than a gain of x1. Therefore, the
highest bandwidth configuration is multiplexer mode (with
each input mapped to only one output) and an output buffer
gain of x2.
The relative locations of the input and output channels also
have significant impact on the device bandwidth (due to the
layout of the ISL59530 silicon). When the input and output
channels are further away, there are additional parasitics as
a result of the additional routing, resulting in lower
bandwidth.
The bandwidth does not change significantly with resistive
loading, as shown in the “Typical Performance Curves”
beginning on page 9. However several of the curves
demonstrate that frequency response is sensitive to
capacitance loading. This is most significant when laying out
the PCB. If the PCB trace length between the output of the
crosspoint switch and the back-termination resistor is not
minimized, the additional parasitic capacitance will result in
some peaking and eventually a reduction in overall
bandwidth.
Linear Operating Region
In addition to bandwidth optimization, to get the best linearity
the ISL59530 should be configured to operate in its most
linear operating region. Figure 48 shows the differential gain
curve. The ISL59530 is a single supply 5V design with its
most linear region between 0.1 and 2V. This range is fine for
most video signals whose nominal signal amplitude is 1V.
The most negative input level (the sync tip for composite
video) should be maintained at 0.3V or above for best
operation.
FIGURE 48. DIFFERENTIAL GAIN RESPONSE
In a DC-coupled application, it is the system designer’s
responsibility to ensure that the video signal is always in the
optimum range.
When AC coupling, the ISL59530’s Clamp (also called
“DC-restore”) function automatically and continuously
adjusts the DC level so that the most negative portion of the
video is always equal to VREF.
A discussion of the benefits of the DC restoration function
begins by understanding the Clamp circuit shown in
Figure 49. The incoming video signal is typically terminated
into 75Ω, then AC-coupled through C1, at which point it is
connected to the base of the buffer’s differential pair. These
components form the video path.
The Clamp function consists of Q1, D1, Q2, D2, the two
current sources, and the 3 switches controlled by the Clamp
Enable signal. The VREF voltage is level-shifted up two
diode drops (Q1 and D1) to the base of Q2. If the voltage at
the cathode of D2 goes below VREF, Q2 and D2 will turn on,
keeping the INx voltage at VREF. If the voltage at INx is
greater than VREF, Q2 and D2 are off and the INx node is
high impedance. This is how the clamp function forces the
lowest portion of the video signal (the sync tip) to always be
equal to or greater than VREF.
To make sure that the sync tip is always equal to (not equal to
or greater than) VREF; i1 is constantly sinking ~2µA of current
from C1. This causes each sync tip to be slightly lower voltage
than the previous sync tip, causing Q2 and D2 to turn on at
each sync tip and raise the voltage to VREF. The 2µA pull-down
with a 0.1µF capacitor and a 15kHz HSYNC frequency results
in 1.3mV of “droop” across every line, or 0.2% of the video
signal. Because 1.3mV is only 0.2% of a 0.7V video signal, this
droop is imperceptible to the human eye.
22
FN6220.8
April 13, 2011