
AD1317
–9–
REV. A
FUNCTIONAL DESCRIPTION
The AD1317 is an ultrahigh speed window comparator designed
for use in general purpose instrumentation and automatic test
equipment. The internal connections for windowing operation
keep the capacitance at the critical common input (VINA/B)
well below what could normally be obtained using separate input
pins.
Another key feature is that the front end circuitry may be dis-
abled, decreasing input bias currents to 50 pA (typical). This
enables sensitive dc current testing without having to physically
disconnect the AD1317’s input from the circuit. The comparator’s
outputs would normally be latched to maintain absolute logic
levels prior to inhibiting the input.
High speed comparators using bipolar process technology usu-
ally have input bias currents in the 1
μ
A to 20
μ
A range, and the
AD1317 is no exception in this regard. This occurs because the
input devices usually have low current gain but must be oper-
ated at high currents to obtain the widest possible bandwidth.
Careful design minimizes variations in the AD1317’s bias cur-
rent with respect to both differential and common-mode input
variations. This translates directly to a high equivalent input
resistance, the minimum of which occurs with zero differential
input. The typical input resistance of the AD1317’s common
input under this condition is on the order of 4 megohms.
Many ATE applications have required input dividers/buffers to
reduce standard logic voltages to levels which can be processed
by “687” type comparators. These dividers have also reduced
the slew rates at which the comparators must properly function.
The AD1317’s 9 volt differential and common-mode input
ranges and 2.5 V/ns slew rate capability make these buffer cir-
cuits unnecessary in most applications.
Separate, complementary latch inputs are provided for each
comparator. These may be driven by differential or single-ended
sources ranging from ECL to HCMOS logic. When using the
comparator’s transparent mode, the latch inputs may be tied
anywhere within their common-mode range with a maximum
differential of 4 V. Symmetrical hysteresis may also be generated
by applying a small differential voltage to the latch inputs (see
HYSTERESIS).
The AD1317’s outputs are standard emitter followers with ECL-
compatible voltage swings. The recommended output termina-
tion is 50
to –2 V. Larger value termination resistors connected
to –V
S
may be used, but will reduce edge fidelity. Typical
output rise and fall times (20%–80%) are 1 ns with a 50
,
10 pF load. The maximum output source current is 40 mA.
THERMAL CONSIDERATIONS
The AD1317 is provided in a 0.450"
×
0.450", 16-lead (bottom
brazed) gull wing, surface mount package with a typical
θ
JC
(junction-to-case thermal resistance) of 17.5
°
C/W. Thermal
resistance
θ
CA
(case to ambient) vs. air flow for the AD1317 in
this package is shown in Figure 25. The improvement in thermal
resistance vs. air flow begins to flatten out just above 400 lfm
1, 2
.
NOTES
1
lfm is airflow in linear feet/minute.
2
For convection cooled systems, the minimum recommended airflow is 400 lfm.
Figure 25. Case-to-Ambient Thermal Resistance vs. Air
Flow
DISPERSION
Propagation delay dispersion is the change in device propagation
delay which results from changes in the input signal conditions.
Dispersion is an indicator of how well the comparator’s frontend
design balances the conflicting requirements of high gain and
wide bandwidth. High gain is needed to ensure that small over-
drives will produce valid logic outputs without an increase in
propagation delay, while wide bandwidth enables the compara-
tor to handle fast input slew rates. The input signal criteria used
to determine the AD1317’s dispersion performance are ampli-
tude, overdrive and slew rate for both standard CMOS and
ECL signal levels.
HYSTERESIS
The customary technique for introducing hysteresis into a com-
parator uses positive feedback as shown in Figure 27. The major
problems with this approach are that the amount of hysteresis
varies with the output logic levels and that the hysteresis is not
symmetrical around zero.
The AD1317 does not use this technique. Instead, hysteresis is
generated by introducing a differential voltage between LE and
LE
as shown in Figure 28. Hysteresis generated in this manner
is independent of output swing and is symmetrical around zero.
The variation of hysteresis with input voltage is shown in Figure
29; the useful hysteresis range is about 20 mV.
LAYOUT CONSIDERATIONS
Like any high speed device, the AD1317 requires careful layout
and bypassing to obtain optimum performance. Oscillations are
generally caused by coupling from an output to the high imped-
ance inputs. All drive impedances should be as low as possible,
and lead lengths should be minimized. A ground plane should
be used to provide low impedance return paths. Care should be
taken in selecting sockets for incoming or other testing to mini-
mize lead inductance, and sockets are not recommended for
production use.