Baxandall Equalizer with Adjustable Frequency (Active Shelf Equalizer)

[YouAre]

I've been searching for an active filter that is able to boost or cut highs/lows wit a shelf response.

Without wanting to get into the technology that goes into analog mixing consoles, I started searching an adaptation of the Baxandall circuit. I wanted continuous control of the frequency response with a potentiometer. I could have easily switched the caps with a toggle/rotary switch set to useful frequencies, but where's the fun in that?

I found this article, which you guys might find interesting (if you haven't already seen it).

http://research.microsoft.com/pubs/102099/Malvar_ParmEq_ElMag0381.pdf

It's not exactly an adaptation of the Baxandall, but it claims to have a similar response. I plan on testing it by simulating it in spice vs. a Baxandall, and by breadboard. Anyone have experience with this? If not, enjoy!

[Bill Mountain]

This looks pretty cool.

What a great find.  I might sim it myself when I get some time.

[YouAre]

This looks pretty cool.

What a great find.  I might sim it myself when I get some time.

Let me know if you get a chance to do this.


I got a chance to sim this, and had some issues. I did add input and output caps. But I realized that I may not have the biasing resistors in the correct spot. I put a 470k resistor to 4.5 volts at the non-inverting inputs of all the buffering op amps (A1, A3, and A4 on the document I posted). But I'm not sure if A2 requires the same biasing.

I'm not sure if A2 counts as a differential amplifier, and even then I'm not familiar with differential amplifiers in AC Circuits. I'm not sure if the non-inverting input would require us to apply a 4.5V bias voltage.

Looking at an established parametric EQ from Geofex ( http://www.geofex.com/article_folders/eqs/paramet.htm ), it seems that we don't need to apply 4.5V to A2.

I will post screenshots of my circuit and simulation tonight.

In the mean time, can anyone also comment on what changes to make to the posted circuit to make it work with guitar or confirm my assumptions about biasing?

Thank you.

[earthtonesaudio]

I wonder why the lowpass buffer is post-filter and the highpass is pre-filter? 

What happens when the treble, bass, and depth controls are turned to max?  How does the signal get in? 

[YouAre]

I wonder why the lowpass buffer is post-filter and the highpass is pre-filter? 

What happens when the treble, bass, and depth controls are turned to max?  How does the signal get in? 

That did seem funky to me. I figured the supposed frequency response was a result of the odd buffer placement. Some cleverness must be involved to  turn a simple passive low pass/high pass "roll-off" filter into an active filter with a shelf response.

[earthtonesaudio]

I did a quick simulation in Falstad (click "import" and paste code to view sim), and the Depth control is indeed funky.  I suspect a real-world implementation might be better/worse in this regard because real pots don't (always) go to zero ohms...

At maximum depth settings the controls will sometimes invert their function, and generally the controls are quite interactive.

code to import:

Code Select Expand

$ 0 5.0E-6 5 38 5.0 43
% 4 28853.998118144256
a 592 512 688 512 1 15.0 -15.0 1000000.0
a 784 352 880 352 1 15.0 -15.0 1000000.0
a 720 176 832 176 0 15.0 -15.0 1000000.0
r 544 96 720 96 0 22000.0
r 720 96 832 96 0 22000.0
w 832 96 832 176 0
w 720 96 720 160 0
w 720 160 640 160 0
w 784 368 784 400 0
w 784 400 880 400 0
w 880 400 880 352 0
w 592 528 592 560 0
w 592 560 688 560 0
w 688 560 688 512 0
r 880 352 976 352 0 470000.0
w 976 512 896 512 0
r 896 512 896 608 0 180000.0
g 896 608 896 624 0
c 688 512 800 512 0 3.899999999999999E-9 0.0
r 800 512 800 576 0 5600.0
r 800 512 896 512 0 470000.0
174 640 160 592 224 0 47000.0 0.15350000000000003 Depth
g 592 192 592 224 0
w 544 96 544 304 0
174 592 304 624 368 0 10000.0 0.005 Bass
174 544 464 592 528 0 47000.0 0.9950000000000001 Treb
w 544 528 512 528 0
w 544 304 592 304 0
w 544 304 544 464 0
w 592 368 512 368 0
w 512 368 512 528 0
r 624 336 688 336 0 56000.0
174 688 336 752 368 0 470000.0 0.09410000000000002 freq_low
c 720 368 720 416 0 3.899999999999999E-9 0.0
g 720 416 720 448 0
w 720 368 752 368 0
w 752 368 752 336 0
w 752 336 784 336 0
174 800 576 848 640 0 47000.0 0.9158000000000001 freq_hi
g 848 608 848 640 0
w 976 352 976 512 0
w 720 192 720 272 0
w 720 272 976 272 0
w 976 272 976 352 0
w 832 176 1008 176 0
w 1008 176 1008 656 0
w 1008 656 512 656 0
w 512 656 512 528 0
170 544 96 480 96 2 20.0 4000.0 5.0 0.1
O 1008 176 1040 176 0


[YouAre]

I did a quick simulation in Falstad (click "import" and paste code to view sim), and the Depth control is indeed funky.  I suspect a real-world implementation might be better/worse in this regard because real pots don't (always) go to zero ohms...

At maximum depth settings the controls will sometimes invert their function, and generally the controls are quite interactive.

Thank you! I'll check this out when I'm home from work.

I was planning on leaving the depth control out of this anyway. Not really sure what its purpose is, based on the claimed behavior of the control.

[earthtonesaudio]

Oh, it does work as a depth control, it just gets crazy when set to max.  That might be an artifact of the sim, but generally grounding the feedback node of an opamp is a recipe for ...unpredictable... behavior.

A resistor in series with that control would probably tame it.

[YouAre]

Oh, it does work as a depth control, it just gets crazy when set to max.  That might be an artifact of the sim, but generally grounding the feedback node of an opamp is a recipe for ...unpredictable... behavior.

A resistor in series with that control would probably tame it.

I'm having trouble getting my LT spice to mimic the results of the sim you posted. Your sim seems to indicate that the filter does in fact work as described, but the boost/cut range is closer to 8db as opposed to the 15db shown in the original document.

I assume my simulation has some biasing mistake. Falstad seems to not need us to apply any bias voltages. Would you happen to have any insight as to wear I would/wouldn't need to apply 4.5v?

Thanks again. I love the sim!

I will edit this post to included simulation results at a later time.

[YouAre]

Also! Alternative method for an adjustable low shelf.  http://sound.westhost.com/project28.htm

It resembles RG's parametric EQ located here: http://www.geofex.com/article_folders/eqs/paramet.htm

If you short C2 or C4, you get a frequency adjustable low shelf! The first link does not include RG's Resonance pot. My simulations showed that you should definitely leave in the Resonance pot, and setting it to 5k gives the most balanced response.

I will edit this post later to show my simulation results.

[earthtonesaudio]

Oh, it does work as a depth control, it just gets crazy when set to max.  That might be an artifact of the sim, but generally grounding the feedback node of an opamp is a recipe for ...unpredictable... behavior.

A resistor in series with that control would probably tame it.

I'm having trouble getting my LT spice to mimic the results of the sim you posted. Your sim seems to indicate that the filter does in fact work as described, but the boost/cut range is closer to 8db as opposed to the 15db shown in the original document.

I assume my simulation has some biasing mistake. Falstad seems to not need us to apply any bias voltages. Would you happen to have any insight as to wear I would/wouldn't need to apply 4.5v?

Thanks again. I love the sim!

I will edit this post to included simulation results at a later time.

First thing to note is that Falstad's filter applet does NOT do nonlinear stuff.  No clipping, no diodes, etc.  So offsetting the voltages will not change the response at all.
But if you want to rearrange the circuit to show DC bias you can right-click the opamps to change their power supplies from the default +/-15V.  Then you can swap out all the GND points for 4.5V. (voltage sources are available under "inputs" in the right-click menu).
But the usual caveats apply: the sim is just a sim, and if your real "4.5V" is coming out of a resistive divider with a capacitor to ground, it's not going to behave the same as an ideal voltage source.  But you already knew that. 

Beyond that I didn't look into whether the sim and the schem match.  But being a practical person, I would recommend to trust neither until you've at least breadboarded it.

[PRR]

That schematic has problems (even before you get to biasing).

Can anyone find the Wireless World article in the footnote? WW's editors were sometimes sharper.