Skip to content
September 11, 2018 / consort3

3rd order elliptical crossover with one op-amp

This combines 2 ideas, the equal component 3rd order Sallen & Key filter and the S&K with bridged T.



Note that R5 and R8 can be replaced with short-circuits, without them the simulator jibbed. You do not get the valley with reversed tweeter that the 4th order LR filter gives. However this gives the 360 degree phase change that the in-phase 3rd order is supposed to give. Amazingly the reversed tweeter configuration gives the 180 degree change of the standard design which some prefer. One could tune the nulls using C4 and R9 to take out particular breakup or resonance peaks. Comparison with standard 3rd order shown in blue on LHS and green on RHS


Vary the components in the table to get the crossover frequency indicated

FREQ. R2,3,4 R1 R7 R6 R9
3.12K 12K 8.2K 3K 56K 220K
2.84K 13K 9.1K 3.3K 62K 240K
2.53K 15K 10K 3.6K 68K 270K
2.34K 16K 11K 3.9K 75K 300K
2.11K 18K 12K 4.3K 82K 330K

Strictly speaking, this is not an elliptical filter, which has both pass and stop band ripple but it is an inverse Chebychev which has stop band ripple. With the addition of one component to an equal component S&K filter it gives useful extra attenuation to the filter slope.

The reverse tweeter connection shows a bit of a ‘BBC dip’ in the amplitude response. If you lower R2,3,4  it gives a flatter response. You can get an even flatter response with a slightly lowerR2,3,4 than this, but the value is non standard.

New table for reversed tweeter configuration

FREQ. R2,3,4 R1 R7 R6 R9
2.97K 12K 9.1K 3.6K 62K 240K
2.72K 13K 10K 3.9K 68K 270K
2.42K 15K 11K 4.3K 75K 300K
2.24K 16K 12K 4.7K 82K 330K
2.01K 18K 13K 5.1K 91K 360K

Thanks to Andyc and FvM for the inspiration.

How to do this with passive components:



For the addition of L3 and C4 to a standard 3rd order crossover you get improved out of band rejection while preserving the in band 3rd order characteristics The fact that the drive unit impedance is not purely resistive affects the response. You would need zobel networks across the drive units to remedy this.

By the way, going back to the active circuit, you can use the  tracking idea to adjust the circuit values to suit component values you may have in stock. Say you have lots of accurate 3.3nF and the circuit calls for 4.7nF. Then the capacitors are 2 values down in the E12 series. To compensate you go 2 values up in the E12 series for resistors and jump 4 values for the E24 series so if C5,6,7 are 3.3nF, R1 is 12k, R6 is 82k R9 is 330k and R7 is 4.3k

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: