KH563
DATA SHEET
With this total value derived, the required external C is
Gain and Output Impedance Range
x
developed by backing out the effect of the internal 10pF.
Figure 7 shows a plot of the recommended gain and
output impedances for the KH563. Operation outside of
this region is certainly possible with some degradation in
performance. Several factors contribute to set this range.
At very low output impedances, the required value of
feedback resistor becomes so low as to excessively load
the output causing a rapid degradation in distortion.
This, and an expression for the external C without the
x
intermediate steps are shown below.
10 C
t
C =
x
10 − C
t
or
The maximum R was set somewhat arbitrarily at 200Ω.
o
1
2
C =
pF
x
This allows the KH563 to drive into a 2:1 step down
transformer matching to a 50Ω load. (This offers
some advantages from a distortion standpoint.
R
o
− 0.08
300 1−
R
g
100
The plot in Figure 6 shows the required C vs. gain for
Low Rf or Rg Region
x
90
several desired output impedances using the equations
80
70
60
shown above. Note that for lower R ’s, C can get very
o
x
large. But, since the total compensation is actually the
series combination of C and 10pF, going to very high
Recommended
Region
50
x
C ’s is increasingly ineffective as the total compensation
40
x
is only slightly changed. This, in part, sets the lower
30
20
limits on allowable R .
o
10
High Noise Region
20
0
Maximally Flat Response
into a Matched Load
18
0
20 40 60 80 100 120 140 160 180 200
Output Impedance (Ω)
16
14
12
Figure 7: Recommended Gain and
Output Impedance Range
Ro = 50Ω
10
8
Ro = 75Ω
6
4
2
0
For a given R , the minimum gain shown in Figure 7 has
o
been set to keep the equivalent input noise voltage less
than 4nV/√Hz. Generally, the equivalent input noise volt-
age decreases with higher signal gains. The high gain
Ro = 100Ω
5
10 15 20 25 30 35 40 45 50 55
limit has been set by targeting a minimum R of 10Ω or a
g
No Load Voltage Gain
minimum R of 100Ω.
f
Figure 6: External Compensation Capacitance (C )
x
Amplifier Configurations
The KH563 is intended for a fixed, non-inverting, gain
configuration as shown in Figure 1. The KH560 offers the
better pulse fidelity with its improved thermal tail in the
pulse response (vs. the KH563). Due to its low
internal forward gain, the inverting node does not present
a low impedance, or virtual ground, node. Hence, in an
inverting configuration, the signal’s source impedance
will see a finite load whose value depends on the output
loading. Inverting mode operation can be best achieved
using a wideband, unity gain buffer with low output
impedance, to isolate the source from this varying load.
A DC level can, however, be summed into the inverting
node to offset the output either for offset correction
or signal conditioning.
A 0% small signal overshoot response can be achieved
by increasing C slightly from the maximally flat value.
x
Note that this applies only for small signals due to slew
rate effects coming into play for large, fast edge rates.
Beyond the nominal compensation values developed
thus far, this external C provides a very flexible means
x
for tailoring the frequency response under a wide variety
of gain and loading conditions. It is oftentimes useful to
use a small adjustable cap in development to determine
a C suitable to the application, then fixing that value for
x
production. An excellent 5pF to 20pF trimmer cap for this
is a Sprague-Goodman part #GKX20000.
When the KH563 is used to drive a capacitive load, such
as an ADC or SAW device, the load will act to compen-
Accuracy Calculations
sate the response along with C . Generally, considerably
Several factors contribute to limit the achievable KH563
accuracy. These include the DC errors, noise effects, and
the impact internal amplifier characteristics have on the
signal gain. Both the output DC error and noise model
may be developed using the equivalent model of Figure
5. Generally, non-inverting input errors show up at the
x
lower C values are required than the earlier develop-
x
ment would indicate. This is advantageous in that a low
R would be desired to drive a capacitive load which,
o
without the compensating effect of load itself, would
otherwise require very large C values.
x
REV. 1A January 2008
9