Parametric EQ with State Variable Filter, why which topology?

[JohnRoberts]

* As I wrote earlier, I don't recommend using "Q" as a parameter of EQ's. I know I'm rowing against the current because that's the current usage, but I stand that "Q" is valid only for response that falls asymptotically to -infinity, which is not the case of an audio EQ, where the response tends to unity.
I'd rather express this as "(relative) BW", i.e. BW expressed in octaves, set at max boost/cut, measured at half max efficiency. That makes more sense operationally and perceptionally. There have been tentatives by the AES to standardize this, but never found completion.

Q and Bandwidth are used kind of interchangeably to represent the same thing (how wide or narrow is the effect of the EQ).  To repeat myself too, in boost/cut EQ sections  the Q or BW of the EQ does not strictly agree with the Q of embedded BP filters. I petitioned the AES to come up with a standard definition or set of definitions for Q/BW in boost/cut EQ.  I was unsuccessful and life is short so I gave up. 

"Relative BW" makes as much sense as any, but we all all need to be talking about the same objective measurements when labelling controls or specifying product performance.  Making measurements at maximum boost/cut will more closely resemble the Q of the related BP filter, but may not represent boost/cut characteristic for more modest EQ settings.

In case it doesn't sound like it, I am agreeing with you.

JR

 

[abbey road d enfer]

In case it doesn't sound like it, I am agreeing with you.

No problem, that's what I figured out.

 

[NickApparently]

Thanks for all the replies!

I finally found some time to try the SSL style EQ with a SVF. (with a phase-inverter right after the bandpass)
https://www.dropbox.com/s/44luiw2qh6rot1k/Schermafdruk%202018-08-09%2020.58.22.png?dl=0

Two things stood out to me: First of all, the gain adjust was just about +/- 6dB. Bandwidth adjust was great.
After some messing around I removed the phase-inverse from the Band-Pass and gain was insane, which was expected as the signals will cancel each other out. Sadly the width was very narrow and barely adjustable.

My question now is, what am I doing wrong? Why is in the 'normal' way limited to only 6dB, and why is the Q (bandwidth) barely adjustable  if I remove the phase inverter?

EDIT:
I used the SVF from the schematic in the attachment

 

[JohnRoberts]

Thanks for all the replies!

I finally found some time to try the SSL style EQ with a SVF. (with a phase-inverter right after the bandpass)
https://www.dropbox.com/s/44luiw2qh6rot1k/Schermafdruk%202018-08-09%2020.58.22.png?dl=0

Two things stood out to me: First of all, the gain adjust was just about +/- 6dB. Bandwidth adjust was great.

Yes, panning the BW/Q control between the + and - inputs of the HP SVF section give wide BW range with modest noise gain, vs hanging the BW pot to ground from one or the other.

After some messing around I removed the phase-inverse from the Band-Pass and gain was insane, which was expected as the signals will cancel each other out. Sadly the width was very narrow and barely adjustable.

unclear... how did you remove the phase-inverse?

My question now is, what am I doing wrong?

can't say, don't know what you did

Why is in the 'normal' way limited to only 6dB,

topology and values define boost/cut... there are different ways to get more.

and why is the Q (bandwidth) barely adjustable  if I remove the phase inverter?

still don't know

JR

EDIT:
I used the SVF from the schematic in the attachment

 

[NickApparently]

Yes, panning the BW/Q control between the + and - inputs of the HP SVF section give wide BW range with modest noise gain, vs hanging the BW pot to ground from one or the other.

Thanks, I'll try BW control with a variable resistor between the + and - input of the HP section, without a tie to ground.

unclear... how did you remove the phase-inverse? can't say, don't know what you did, topology and values define boost/cut... there are different ways to get more. still don't know

JR

Okay, I'll try to explain better. Check the Attachment ^^
With the Phase invert section the gain was only +/-6dB, BW control was great.
Without the Phase invert section (Remove R26, R27 and U6A, tie R18 to R28) the gain was unlimited, but barely any BW control.

Unlimited gain is cool, but not usable. And I think that the BP signal should be phase-coherent with the incoming signal. As you mentioned earlier: "One of my primary rules about circuit design is make it hard for the customer to get bad sounds out of any path"

 

[JohnRoberts]

Thanks, I'll try BW control with a variable resistor between the + and - input of the HP section, without a tie to ground.

no.. perhaps I wasn't clear... the R to ground at the - input drives the Q narrower IIRC. R to ground at the + input drives Q wider.  If you just use a single variable R on one input, you also need a R on the other end to provide a useful adjustment range. This results in higher noise gain (more noise).  Panning between the + and - input with R to ground at the wiper, reduces the ground load for one input while increasing it at the other, for lower total noise for the same Q/BW range.

Okay, I'll try to explain better. Check the Attachment ^^
With the Phase invert section the gain was only +/-6dB, BW control was great.
Without the Phase invert section (Remove R26, R27 and U6A, tie R18 to R28) the gain was unlimited, but barely any BW control.

Unlimited gain is cool, but not usable. And I think that the BP signal should be phase-coherent with the incoming signal. As you mentioned earlier: "One of my primary rules about circuit design is make it hard for the customer to get bad sounds out of any path"

Don't do that... there are many workable topologies around the WWW, keep looking.

Back in the 70's I published a list of equations for standard SVF EQ. Lots of my customers were college kids, so i was doing their homework for them (to sell kits).  I don't know if I still have that sheet around.

JR

 

[NickApparently]

no.. perhaps I wasn't clear... the R to ground at the - input drives the Q narrower IIRC. R to ground at the + input drives Q wider.  If you just use a single variable R on one input, you also need a R on the other end to provide a useful adjustment range. This results in higher noise gain (more noise).  Panning between the + and - input with R to ground at the wiper, reduces the ground load for one input while increasing it at the other, for lower total noise for the same Q/BW range.

Thanks for clearing that up!

Don't do that... there are many workable topologies around the WWW, keep looking.

Back in the 70's I published a list of equations for standard SVF EQ. Lots of my customers were college kids, so i was doing their homework for them (to sell kits).  I don't know if I still have that sheet around.

JR

Good call, felt weird to remove the phase-inverter.
I'll keep looking! Is there any specific term I can use when searching for summing-BP-parametric-SVF?

Which components are affecting the maximum boost/cut? I've noticed that the value of the Boost/cut-pot isn't really affecting the maximum boost/cut. Somewhere I've red that, in a topology like in the attachment, you can add a resistor (2K7 in this example) from the - input of U1 to ground, in order to change the max boost/cut from 9dB to 12dB. Could you help me find any other components?

Thanks for you great info so far!

 

[Bo Deadly]

Which components are affecting the maximum boost/cut?

Reduce the resistor on the wiper (R3 in that ESP schem). Something like 4k7 maybe.

Note that 12dB is considered quite a lot of boost / cut. I would probably do something more like 10dB myself.

 

[boji]

One of my primary rules about circuit design is make it hard for the customer to get bad sounds out of any path

+1

This hit me square in the face years ago when I began to code up simple gui's for endusers. People's ability to exploit flaws/bugs -often out of sheer ignorance- always exceeded my greenhorn coder's ability to anticipate incorrect use. 
Beta testing became infinitely important in my mind at that point, not just to improve the code.

 

[JohnRoberts]

+1

This hit me square in the face years ago when I began to code up simple gui's for endusers. People's ability to exploit flaws/bugs -often out of sheer ignorance- always exceeded my greenhorn coder's ability to anticipate incorrect use. 
Beta testing became infinitely important in my mind at that point, not just to improve the code.

Try designing products for Peavey customers for over a decade...    no offense to Peavey customers specifically but they were often first time, inexperienced users, so the goal was to get decent sound despite that.

JR

PS: Sometimes less is just right... When Peavey designed a nice tube mic preamp, I advised them to limit the boost/cut in the shelving EQ (to something like +/- 6 dB), while still letting the gain knob go to 11...  8)

 

[NickApparently]

Reduce the resistor on the wiper (R3 in that ESP schem). Something like 4k7 maybe.

Note that 12dB is considered quite a lot of boost / cut. I would probably do something more like 10dB myself.

I would settle for 12dB, but mostly I'm curious how the filter works. I understand most parameters of the SVF, but the boost/cut remained mainly a mistery to me.
Am I correct to say that altering R3 works in the same way as a non-inverting op-amp  with gain?

Attached is my test-setup. In this setup I'll try and change R25 from 22K to 10K.
Will it  also be possible to change R28 from 4K7 to 2K7? This will alter the gain of the BP signal.

 

[JohnRoberts]

I would settle for 12dB, but mostly I'm curious how the filter works. I understand most parameters of the SVF, but the boost/cut remained mainly a mistery to me.

The SVF is a specialized filter with HP, BP, and LP outputs.  Boost or cut is generated by taking the BP output and adding/subtracting that from a dry stage .  There are sundry different topologies for how the BP gets combined with dry signal that affect the boost/cut sensitivity.  Further the gain of SVF BP stage output can also be varied to alter boost/cut. General practice is to operate the SVF BP stage at unity gain, since changing that gain also changes the Q/BW.

To visualize boost/cut, imagine the SVF BP output is just another wideband signal input (with inverted polarity), perhaps visualize it as a unity gain inverter. Calculating the wideband gain that way will give you the effective boost cut.

JR

Am I correct to say that altering R3 works in the same way as a non-inverting op-amp  with gain?

Attached is my test-setup. In this setup I'll try and change R25 from 22K to 10K.
Will it  also be possible to change R28 from 4K7 to 2K7? This will alter the gain of the BP signal.

 

[Bo Deadly]

Attached is my test-setup. In this setup I'll try and change R25 from 22K to 10K.
Will it  also be possible to change R28 from 4K7 to 2K7? This will alter the gain of the BP signal.

I don't know. That schematic is fundamentally different from the ESP one that you keep referencing.

 

[NickApparently]

I don't know. That schematic is fundamentally different from the ESP one that you keep referencing.

Okay, thanks. I thought that maybe the boost/cut pot worked pretty much the same way. I'll just test it, we'll see what happens.

 

[NickApparently]

Okay, so I changed the value for R25 to 4K7, which gave an boost/cut of +/-15dB. Next up was the Q/BW, as told, changing the gain of the BP filter also changes the Q/BW, I had to make the BW control a bit wider.  R21 is changed from 2K7 to 1K8. If somebody's interested, the Q is now adjustable from 0,5 - 5. Switching to shelf-mode is very easy, because you don't need an extra Op-amp to sum the BP with either the LPF or HPF.

Now I wonder why someone would choose between either one of the schematics.. With the normal SVF schematic I need 1 opamp more than the other schematic. The rest is details.

 

[atavacron]

If somebody's interested,

Four years later, i’m interested. What was the result of your explorations?