Datasheet

E =E +(I R ) +4kTR NG +(I R ) +4kTR NG
O NI BN S S BI F F
2 2 2 2
( (
E =E +(I R ) +4kTR +
N NI BN S S
2 2
4kTR
NG
F
( (
I R
BI F
NG
2
+
OPA694
180W
2.86kW
20W
+5V
-5V
V
O
Power-supply
decouplingnotshown.
OPA237
-5V
+5V
V
I
18kW
2kW
1.8kW
OPA694
SBOS319G –SEPTEMBER 2004–REVISED JANUARY 2010 www.ti.com
The total output spot noise voltage can be computed xx = ±(2 × 3mV) ± (20mA × 25Ω × 2) ± (402Ω × 18mA)
as the square root of the sum of all squared output
xx = ±6mV + 1mV ±7.24mV = ±14.24mV
noise voltage contributors. Equation 4 shows the
general form for the output noise voltage using the
A fine-scale, output offset null, or DC operating point
terms shown in Figure 40.
adjustment, is sometimes required. Numerous
techniques are available for introducing DC offset
control into an op amp circuit. Most simple
adjustment techniques do not correct for temperature
(4)
drift. It is possible to combine a lower speed,
precision op amp with the OPA694 to get the DC
Dividing this expression by the noise gain [NG = (1 +
accuracy of the precision op amp along with the
R
F
/R
G
)] will give the equivalent input-referred spot
signal bandwidth of the OPA694. Figure 41 shows a
noise voltage at the noninverting input, as shown in
noninverting G = +10 circuit that holds an output
Equation 5.
offset voltage less than ±7.5mV over-temperature
with > 150MHz signal bandwidth.
(5)
Evaluating these two equations for the OPA694
circuit and component values (see Figure 31) gives a
total output spot noise voltage of 11.2nV/√Hz and a
total equivalent input spot noise voltage of 5.6nV/√Hz.
This total input-referred spot noise voltage is higher
than the 2.1nV/√Hz specification for the op amp
voltage noise alone. This reflects the noise added to
the output by the inverting current noise times the
feedback resistor. If the feedback resistor is reduced
in high-gain configurations (as suggested previously),
the total input-referred voltage noise given by
Equation 5 will approach just the 2.1nV/√Hz of the op
amp itself. For example, going to a gain of +10 using
R
F
= 178Ω will give a total input-referred noise of
2.36nV/√Hz.
Figure 41. Wideband, DC-Connected Composite
DC ACCURACY AND OFFSET CONTROL
Circuit
A current-feedback op amp like the OPA694 provides
This DC-coupled circuit provides very high signal
exceptional bandwidth in high gains, giving fast pulse
bandwidth using the OPA694. At lower frequencies,
settling, but only moderate DC accuracy. The
the output voltage is attenuated by the signal gain
Electrical Characteristics show an input offset voltage
and compared to the original input voltage at the
comparable to high-speed, voltage-feedback
inputs of the OPA237 (this is a low-cost, precision
amplifiers. However, the two input bias currents are
voltage-feedback op amp with 1.5MHz gain
somewhat higher and are unmatched. Whereas bias
bandwidth product). If these two do not agree (due to
current cancellation techniques are very effective with
DC offsets introduced by the OPA694), the OPA237
most voltage-feedback op amps, they do not
sums in a correction current through the 2.86kΩ
generally reduce the output DC offset for wideband,
inverting summing path. Several design
current-feedback op amps. Since the two input bias
considerations will allow this circuit to be optimized.
currents are unrelated in both magnitude and polarity,
First, the feedback to the OPA237 noninverting input
matching the source impedance looking out of each
must be precisely matched to the high-speed signal
input to reduce their error contribution to the output is
gain. Making the 2kΩ resistor to ground an adjustable
ineffective. Evaluating the configuration of Figure 31,
resistor would allow the low- and high-frequency
using worst-case +25°C input offset voltage and the
gains to be precisely matched. Second, the crossover
two input bias currents, gives a worst-case output
frequency region where the OPA237 passes control
offset range equal to:
to the OPA694 must occur with exceptional phase
±(NG × V
OS
) ± (I
BN
× R
S
/2 × NG) ± (I
BI
× R
F
)
linearity. These two issues reduce to designing for
pole/zero cancellation in the overall transfer function.
where NG = noninverting signal gain
Using the 2.86kΩ resistor will nominally satisfy this
space
space
16 Copyright © 2004–2010, Texas Instruments Incorporated
Product Folder Link(s): OPA694