English
Language : 

ICS2495 Datasheet, PDF (2/7 Pages) Integrated Circuit Systems – Dual Video/Memory Clock Generator
ICS2495
Reference Oscillator & Crystal Selection
In cases where the on-chip crystal oscillator is used to generate
the reference frequency, the accuracy of the crystal oscillation
frequency will have a very small effect on output accuracy.
The external crystal and the on-chip circuit implement a Pierce
oscillator. In a Pierce oscillator, the crystal is operated in its
parallel-resonant (also called anti-resonant mode). This means
that its actual frequency of oscillation depends on the effective
capacitance that appears across the terminals of the quartz
crystal. Use of a crystal that is characterized for use in a
series-resonant circuit is fine, although the actual oscillation
frequency will be slightly higher than the value stamped on the
crystal can (typically 0.025%-0.05% or so). Normally, this
error is not significant in video graphics applications, which is
why the ICS2495 will typically derive its frequency reference
from a series resonant crystal connected between pins 1 and
16.
As the entire operation of the phase-locked loop depends on
having a stable reference frequency, the crystal should be
mounted as close as possible to the package. Avoid routing
digital signals or the ICS2495 outputs underneath or near these
traces. It is also desirable to ground the crystal can to the
ground plane, if possible.
Power Supply Conditioning
The ICS2495 is a member of the second generation of dot clock
products. By incorporating the loop filter on chip and upgrad-
ing the VCO, the ease of application has been substantially
improved over earlier products. If a stable and noise-free power
supply is available, no external components are required. How-
ever, in most applications it is judicious to decouple the power
supply as shown in Figure 1.
Layout Considerations
Utilizing the ICS2495 in video graphics adapter cards or on
PS2 motherboards is simple, but does require precautions in
board layout if satisfactory jitter-free performance is to be
realized. Care should be exercised to ensure that components
not related to the ICS2495 do not share its ground. In applica-
tions utilizing a multi-layer board, V SS should be directly
connected to the ground plane.
Frequency Reference
The internal reference oscillator contains all of the passive
components required. An appropriate crystal should be con-
nected between XTAL1 (16) and XTAL2 (1). In IBM compat-
ible applications this will typically be a 14.31818 MHz crystal,
but fundamental mode crystals between 10 MHz and 25 MHz
have been tested. Maintain short lead lengths between the
crystal and the ICS2495. In some applications, it may be
desirable to utilize the bus clock. If the signal amplitude is
equal to or greater than 3.5 volts, it may be connected directly
to XTAL1 (16). If the signal amplitude is less than 3.5 volts,
connect the clock through a .047 microfarad capacitor to
XTAL1 (16), and keep the lead length of the capacitor to
XTAL1 (16) to a minimum to reduce noise susceptibility . This
input is internally biased at V DD/ 2. Since TTL compatible
clocks typically guarantee a VOH of only 2.8V, capacitively
coupling the input restores noise immunity. The ICS2495 is
not sensitive to the duty cycle of the bus clock; however, the
quality of this signal varies considerably with different moth-
erboard designs. As the quality of this signal is typically outside
of the control of the graphics adapter card manufacturer, it is
suggested that this signal be buffered on the graphics adapter
board. XTAL2 (1) must be left open in this configuration.
EXTFREQ
FS0
FS1
STROBE
FS2
FS3
MS0
ICS2495
10
C3 C2
.1 22
5.0V
VCLK
XTALOUT
N/C
MCLK
MS1
NOTES:
FS3-FS0, MS1-MS0, EXTFREQ, and STROBE inputs are all equipped with pull-ups and need not be tied high.
Mount decoupling capacitors as close as possible to the device and connect device ground to the ground plane where available.
Mount crystal and its circuit traces away from switching digital lines and the VCLK, MCLK and XTALOUT lines.
Figure 1
2