Datasheet
DIFFERENTIAL LOW-PASS ACTIVE FILTERS
0
−
1
−
2
−
3
−
4
−
5
−
6
−
7
−
8
−
9
−
10
−
11
−
12
Frequency (Hz)
Differential Gain (dB)
10
2
10
4
10
5
10
3
10
6
HIGH-PASS FILTERS
1500
Ω
506Ω
0.1
µ
F
506
Ω
238Ω
5k
Ω
5k
Ω
238
Ω
0.1
µ
F
750Ω
750
Ω
V
O
V
I
+5V
1/2
OPA2830
1/2
OPA2830
238
Ω
238
Ω
BUF602
150pF
150pF
100pF
100pF
0.1µF
+5V
2.5V
OPA2830
SBOS309D – AUGUST 2004 – REVISED AUGUST 2008 ..................................................................................................................................................
www.ti.com
Implementing the DC bias in this way also attenuates
the differential signal by half. This is recovered by
The dual OPA2830 offers an easy means to
setting the amplifier gain at 2V/V to get a net
implement low-power differential active filters. On a
unity-gain filter characteristic from input to output. The
single supply, one way to implement a 2nd-order,
filter design shown here has also adjusted the
low-pass filter is shown in Figure 79 . This circuit
resistor values slightly from an ideal analysis to
provides a net differential gain of 1 with a precise
account for the 100MHz bandwidth in the amplifier
5MHz Butterworth response. The signal is
stages. The filter capacitors at the noninverting inputs
AC-coupled (giving a high-pass pole at low
are shown as two separate capacitors to ground.
frequencies) with the DC operating point for the
While it is certainly correct to collapse these two
circuit set by the unity-gain buffer — the BUF602. This
capacitors into a single capacitor across the two
buffer gives a very low output impedance to high
inputs (which would be 50pF for this circuit) to get the
frequencies to maintain accurate filter characteristics.
same differential filtering characteristic, tests have
If the source is a DC-coupled signal already biased
shown two separate capacitors to a low impedance
into the operating range of the OPA2830 input CMR,
point act to attenuate the common-mode feedback
these capacitors and the midpoint bias may be
present in this circuit giving more stable operation in
removed. To get the desired 5MHz cutoff, the input
actual implementation. Figure 80 shows the
resistors to the filter is actually 119 Ω . This is
frequency response for the filter of Figure 79 .
implemented in Figure 79 as the parallel combination
of the two 238 Ω resistors on each half of the
differential input as part of the DC biasing network. If
the BUF602 is removed, these resistors should be
collapsed back to a single 119 Ω input resistor.
Figure 80. 5MHz, 2nd-Order, Butterworth
Low-Pass Filter
Another approach to mid-supply biasing is shown in
Figure 81 . This method uses a bypassed divider
network in place of the buffer used in Figure 79 . The
impedance is set by the parallel combination of the
resistors forming the divider network, but as
Figure 79. Single-Supply, 2nd-Order, Low-Pass
frequency increases it looks more and more like a
Sallen-Key Filter
short due to the capacitor. Generally, the capacitor
value needs to be two to three orders of magnitude
greater than the filter capacitors shown for the circuit
to work properly.
26 Submit Documentation Feedback Copyright © 2004 – 2008, Texas Instruments Incorporated
Product Folder Link(s): OPA2830