Need an Active 2nd Order Low Shelf Filter aka Low Shelving Filter

elmura

Member
2012-08-21 2:34 pm
Hey Guys,
I've been trying to design a simple, linear, active, second order low shelf filter with around 6dB attenuation without much success. Research on the web has proved a massive time consumer without luck.

The design of the low frequency shelf cut specs are around 5-6dB between shelves and a 1kHz corner. Needs to have low phase shift, and low group delay variation and have low noise & THD.

I've got a high pass shelf working based on Linkwitz Shelving Filter, but I need a low shelf cut.
Thanks
 

Attachments

  • 1st order Shelf.JPG
    1st order Shelf.JPG
    50.1 KB · Views: 331
See if the attached circuit works for you...

You can design this yourself, using the LT designer I wrote:
CFL Linkwitz Transform Designer with Monte Carlo Sensitivity Analysis

There are other ways to make the second order shelving filter. You have to implement a biquadratic filter, and you can do that using various "biquad" circuit topologies.

As others have said, phase shift and group delay are a direct consequence of the frequency response and are independent of which circuit you choose. The distortion is dependent on what amplifier (op amp) you use and the resistances in the circuit, etc. and is something that you can tweak.

-Charlie
 

Attachments

  • LT cut circuit info.GIF
    LT cut circuit info.GIF
    23.3 KB · Views: 321
  • LT cut circuit plot.GIF
    LT cut circuit plot.GIF
    29.3 KB · Views: 316

elmura

Member
2012-08-21 2:34 pm
I've tried the Linkwitz Transform circuit and yes, it works quite well for the response curve. Nice spreadsheet CharlieLaub. Very interesting and looked like some work to put together.

! However ! Regarding the Linkwitz Transform:
- There are too many components in the signal path
- It inverts the signal
- Lots of noise & distortion added by the network

Is there a simpler, cleaner way? Similar to the high pass circuit I posted? Even if it is 1st order.
 
Yes. Non-inverting opamp, with CR in parallel in the feedback arm and R to ground from the inverting input. Make sure the opamp is unity-gain stable. Unity gain at high frequencies; gain set by resistors at low frequencies, rolloff set by C (in conjunction with R's). Should be in any decent electronics textbook, or Google.
 

elmura

Member
2012-08-21 2:34 pm
Interesting work Michael. I've been playing around with it in Tina simplifying it to my requirements. It works great.
Questions:
1. Is this a biquad?
2. What do you adjust to alter corner frequency?
3. Is it possible to do the same thing with one opamp for bass shelf cut? ie. a simpler design?
 
elmura,

If you take a look at the tech-pub by Dennis Bohn at Rane Corporation (here: http://rane.com/pdf/acceler.pdf), it should answer many of your questions, this circuit is based on that work. The original circuit utilized a dedicated opamp for each filter; by scaling the first filter's resistance/impedance level I was able to eliminate one opamp from the filter section.

1) No, it's not a biquad, it's a non-traditional type cascade of two first order filters that, because of the way thet're implimented here, will provide a steeper than normal slope while limiting phase shift.

2) If you read through the whole thread I started (here: http://www.diyaudio.com/forums/chip-amps/199130-usable-tone-control.html), you probably noticed the original version had filters where the corner frequencies were easy to adjust. After evaluating the first circuit I still wasn't happy with the way it affected the mid-band frequencies. That's when I decided to try the accelerated slope type of filter, but in using those I gave up the easy frequency adjustment the original had.
To change the filter corner frequency, a simple mathematical scaling of the cap and/or resistor values works well, just be sure to scale both of the cascaded filters equally.

3) The circuit is already about as simple as it can be. It requires a summing inverter at the input, a summing inverter at the output, and what ever filter stage(s) in between to provide the transfer function desired, a typical topology for many graphic and parametric equalizer circuits.

As a final comment, the circuit is simple, easy to build and tweek, tolerant of less than ideal physical layout, and best of all when done right won't get in the way of the music.

Mike
 
Last edited: