Skip to content
March 27, 2020 / consort3

Active elliptical crossover filter

Working on a new project with non-ideal drivers, I realised the best crossover for my case was the Jason Cuadra elliptic one since it has the fastest cut-off . I will implement this with passive components, but what about an active solution? Eschewing Jasons gyrator circuit, I found a circuit in Leachs’ filter pot pourri, the biquad which indeed modelled as Leach specified. However I failed to implement a high pass version. Leach had used a version of the biquad which is not amenable to high pass filters. The Tow -Thomas type would work and is described here:

https://hasler.ece.gatech.edu/Courses/ECE6414/Unit1/Discrete_01.pdf

Lancasters Active filter cookbook recommends a state variable filter but is hazy on implementation. I noticed that the Microcap filter designer could design elliptical types and used the Kerwin twin T circuit. All these implementations have 2  poles and 2 zeros. I went for a 20dB re-entry as if you make the requirements too onerous the designer adds stages. I had to offset the frequency slightly to get a smooth amplitude and phase curve. The tweeter is reversed.

ellipticA

ellipticB

Scroll down for a crossover with better integration. I seem to have invented a new value in the E24 range! The circuit has high value resistors which might give noise issues so you could halve the resistor values and double the capacitor values to compensate. Because the exact values required do not map to E24 values, C4 should be750p and R17 should be 68k for the doubled scheme. BOM (bill of material): C1, 5.6nF C2, 2.7nF C3, 3.6nF C4, 750pF C5, 1.8nF C6, 2.7nF C7, 1.2nF C8, 3.9nF C9, 1.8nF C10, 8.2nF R1, 6.8K R2, 47K R3, 22K R4, 10K R5, 10K R6, 10K R7, 10K R8, 30K R9, 20K R10, 20K R11, 20K R12, 10K R17, 68K R18, 15K. Components should be close tolerance, 1% resistors are readily available.

Close tolerance reasonable price capacitors are a problem. The classic solution was the polystyrene type but these are difficult to find and often quite expensive. However I found 3 possible suppliers Reichelt who sell the Styroflex type with 2% tolerance and Mouser who sell the Xicon type with 5% tolerance. Tayda do the Mylar type with 5% tolerance. The other solution is to buy a lot of cheap film 10% or 20% capacitors and select closer tolerance ones from the batch. What if none are the right value? Say you want a 100nF and they are all 95nF. Then you parallel a 95nF with a 4.7nF to give 99.7nF. Do not use cheap ceramic types. NPO ceramics are fine and surface mount 2% types are reasonably priced and avaliable in smallish values.. As an amateur you have the time and hopefully a capacitor measuring device. My multimeter has one, however it is worth knowing the error of the meter by measuring with a known 1% or better capacitor. Knowing the error you can compensate your readings.

For the elliptical filter application you need 2% tolerance or better capacitors. I have specified some E24 values which can be made by paralleling E12 values so for instance a 3.6nF can be made by paralleling 2 1.8nF capacitors. A 750 pF can be made by 2 1.5nF in series.

ellipp

ellipperfect

Such a smooth phase curve deserves a soothing piece of music and here is the Jacques Loussier trio with the Hamlet ad.

https://www.youtube.com/watch?v=14AhD3xdoMk

BOM for 1.75kHz crossover C1, 13nF C2, 6.2nF C3, 8.2nF C4, 1.8nF C5, 4.3nF C6, 6.2nF C7, 3nF C8, 9.1nF C9, 4.3nF C10, 20nF R1, 4.7K R2, 33K R3, 16K R4, 10K R5, 10K R6, 10K R7, 10K R8, 22K R9, 15K R10, 15K R11, 13.6K R12, 6.8K R17, 47K R18, 10K

This is the third of a family of notch crossovers I have presented starting with the so-called elliptical filter on 11/9/18, continuing with the TT inverse Chebyshev filter on 16/1/20 and ending with this one which is a true elliptical sort. All of these have 3rd order characteristics.   i.e. -3dB at crossover, phase quadrature etc. with progressively sharper cut-offs.

More circuitry is required if you want the best results, you need to address the acoustic centre misalignment and baffle step issues as well. The various tricks used on passive crossovers  to sort these problems cannot be used here.  Linkwitz’s filter pages are a good introduction, section 4 Delay correction and section 5 Shelving low pass here:

http://www.linkwitzlab.com/filters.htm

Leave a Reply

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

WordPress.com Logo

You are commenting using your WordPress.com 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: