OP179/OP279

–13–

REV. F

3

2

1

U1A

OP279

4

R1

31.6k

C1

0.01 F

C2

0.01 F

R2

31.6k

R5

31.6k

R6

31.6k

R4

49.9

HI

LO

500Hz AND UP

DC – 500Hz

6

5

7

C3

0.01 F

U1B

OP279

C4

0.02 F

R7

15.8k

R3

49.9

0.1 F

0.1 F

100 F/25V

100 F/25V

TO U1

+5V

–5V

COM

Figure 39. Two-Way Active Crossover Networks

In the filter sections, component values have been selected for

good balance between reasonable physical/electrical size, and

lowest noise and distortion. DC offset errors can be minimized

by using dc compensation in the feedback and bias paths, ac

bypassed with capacitors for low noise. Also, since the network

input is reactive, it should driven from a directly coupled low

Figure 40 shows this filter architecture adapted for single supply

operation from a 5 V dc source, along the lines discussed

previously.

3

2

1

U1A

OP279

4

R1

31.6k

C1

0.01 F

C2

0.01 F

R2

31.6k

R5

31.6k

R6

31.6k

R4

49.9

HI

LO

500Hz

AND UP

DC –

500Hz

6

5

7

C3

0.01 F

U1B

OP279

C4

0.02 F

R7

15.8k

R3

49.9

10 F

10 F

100k

10 F

100k

100k

10 F

100k

0.1 F

100 F/25V

TO U1

+5V

COM

+

100k

+

Figure 40. A Single Supply, Two-Way Active Crossover

Bandpass Configurations

The MFB bandpass filter using an OP179/OP279 section is

shown in Figure 38. This filter provides reasonably stable me-

dium Q designs for frequencies of up to a few kHz. For best

predictability and stability, operation should be restricted to

applications where the OP179/OP279 has an open-loop gain in

excess of 2Q

7

6

5

R = R3

0.1 F

GIVEN:

PICK A STD C1 VALUE, THEN:

C2 = C1

R1 = 1/(H*(2*PI*F*C1))

R2 = 1/(((2*Q) –H)*(2*PI*F*C1))

R3 = Q/(PI*F*C1)

EXAMPLE: 60Hz, Q = 10,

IN

R2

1.4k

(1.33k )

OUT

U1B

OP279

R3

530k

C2

0.1 F

C1

0.1 F

R1

26.4k

(264k )

Figure 38. Two-Pole, Bandpass Multiple Feedback Filters

Given the bandpass design parameters for Q, F, and pass band

for C1. Then C2 and resistors R1-R3 are selected as per the

relationships noted. This filter is subject to a wide range of

component values by nature. Practical designs should attempt

to restrict resistances to a 1 k

Ω to 1 MΩ range, with capacitor

values of 1

µF or less. When needed, dc bias current compensa-

Two-Way Loudspeaker Crossover Networks

Active filters are useful in loudspeaker crossover networks for

reasons of small size, relative freedom from parasitic effects, and

the ease of controlling low/high channel drive, plus the con-

trolled driver damping provided by a dedicated amplifier. Both

Sallen-Key (SK) VCVS and multiple-feedback (MFB) filter

architectures are useful in implementing active crossover net-

works (see Reference 4), and the circuit shown in Figure 39 is

a two-way active crossover which combines the advantages of

both filter topologies. This active crossover exhibits less than

0.01% THD+N at output levels of 1 V rms using general pur-

pose unity gain HP/LP stages. In this two-way example, the LO

signal is a dc-500 Hz LP woofer output, and the HI signal is the

HP (> 500 Hz) tweeter output. U1B forms a MFB LP section

at 500 Hz, while U1A provides a SK HP section, covering fre-

quencies

≥ 500 Hz.

This crossover network is a Linkwitz-Riley type (see Reference

5), with a damping factor or

α of 2 (also referred to as

“Butterworth squared”). A hallmark of the Linkwitz-Riley type

of filter is the fact that the summed magnitude response is flat

across the pass band. A necessary condition for this to happen

is the relative signal polarity of the HI output must be inverted

with respect to the LOW outputs. If only SK filter sections

were used, this requires that the connections to one speaker be

reversed on installation. Alternately, with one inverting stage

used in the LO channel, this accomplishes the same effect. In

the circuit as shown, stage U1B is the MFB LP filter which

provides the necessary polarity inversion. Like the SK sections,

it is configured for unity gain and an

α of 2. The cutoff frequency

is 500 Hz, which complements the SK HP section of U4.