Parametric Eq Theory

[opacheco]

Hi,

Where I can find information (pdf, doc., Website address) about the Parametric Eq Theory and Design???

I am interest in built a really good one!!!........BUT I would like to know the theory behind of these equipments.


Thanks a lot,
opacheco.

 

[SSLtech]

"really good one" and "really bad one" are hopelessly subjective, and no text that you can find will ever point you in those directions.

None the less you ask a lot, but the simplest answer to provide information would be to research the state-variable filter, then know that if you add or subtract its output with a full bandwidth signal, you get a single-band parametric EQ.

There are other approaches, but this covers a lot: (Neve 'Formant' EQ, Amek console EQ, Calrec EQ etc.)

Keith

 

[burdij]

Fred Forsell has an interesting article on designing single opamp equalizers with a bit of information about gyrators and other types of frequency dependent elements. It is here:

http://www.forsselltech.com/downloads/design_discussions/Evolution of an EQ Design2.pdf

 

[JohnRoberts]

http://www.groupdiy.com/index.php?topic=26427&highlight=parametric

Here is a discussion and link to the article I wrote back in late '70s about parametric.

A search turns up many hits on this forum.

JR

 

[bcarso]

There is a long white paper by Rane that I will try to find---maybe someone else has it at hand. It seemed pretty comprehensive, and in particular confirmed that there were two different definitions for "Q" in connection with dip circuits, which had been concerning me when I was mapping from specs from a DSP-based box to an analog-domain EQ under design.

 

[JohnRoberts]

[quote author="bcarso"]There is a long white paper by Rane that I will try to find---maybe someone else has it at hand. It seemed pretty comprehensive, and in particular confirmed that there were two different definitions for "Q" in connection with dip circuits, which had been concerning me when I was mapping from specs from a DSP-based box to an analog-domain EQ under design.[/quote]

FWIW, I approached the AES standards committee last year about that and they agreed to review it for a possible standard definition/clarification. (I probably need to follow up since I haven't heard anything since the initial agreement to pursue).

It is only an issue for the Q of peaking (boost-cut) type EQ sections. There has long been poor conformance between sundry analog GEC, as in not all 1/3 octave or 2/3rd octave EQ are created equal or give similar response. This issues also exists for parametric, but they are even less common so differences between variants not as obvious.

This has come to a head with the widespread use of DSP speaker management systems, and the need for speaker manufacturer's to effectively communicate specific corrective speaker EQ to end users so they can dial in their sundry DSP platforms. All the HPF and LPF track fine, just the EQ fall into a small handful of different variants. The amount of peak/cut is also OK, just the width of the region affected.

Dennis Bohn et al at Rane are well aware of this problem and IIRC they even have a conversion spreadsheet calculator on their website for Q conversion between a few different MFRs.

Hopefully the AES will put a stake in the ground so Q can mean the same thing for all EQs, in our lifetime. :roll:

JR

 

[opacheco]

SSLtech, burdij, JohnRoberts, bcarso; thanks a lot for yours comments!!!

SSLtech, where I can find information(theory) about the "state-variable filter"????

Thanks a gain!
Opacheco.

 

[opacheco]

SSLtech, burdij, JohnRoberts, bcarso; thanks a lot for yours comments!!!

SSLtech, where I can find information(theory) about the "state-variable filter"????

Thanks a gain!
Opacheco.

 

[bcarso]

[quote author="opacheco"]SSLtech, burdij, JohnRoberts, bcarso; thanks a lot for yours comments!!!

SSLtech, where I can find information(theory) about the "state-variable filter"????

Thanks a gain!
Opacheco.[/quote]

This wiki isn't bad, although very brief, and goves some links to other filter stuff:

http://en.wikipedia.org/wiki/State_variable_filter

 

[pucho812]

[quote author="RogerFoote"]That EQ that John R designed was a really sweet but reasonably simple EQ. I took it to a transferring / mastering session years ago and the owner wouldn't let me back in to take it home. He was kind of picky too.

Shoot, that was way back when the TL07X was considered a high quality audio device. That EQ proves too that it isn't how much the parts cost but how they are joined in concert.

I remember how satisfying it was to build and listen to. :thumb:

There was a mod on the schematic for using it on single supplies for automotive use. Always wanted to do that but never got to it....

rf[/quote]

is there a schematic floating around would be interested in studying that

 

[clintrubber]

[quote author="pucho812"]is there a schematic floating around would be interested in studying that[/quote]
While I can't provide you a link I can tell you that it is there somewhere.

I should give this EQ a re-listen... I built it end of the eighties, my first 1RU 19" DIY, complete with various added fancy stuff like 'automatic' stereo 2-band or mono 4-band switching and some other gizmo's. Had PCBs etched, for the main boards & the added utility stuff. Spent days wiring the thing, the making vs using ratio so far all wrong on this box

Bye,

Peter

 

[SSLtech]

[quote author="pucho812"]is there a schematic floating around would be interested in studying that[/quote]
It's already posted above...

Follow John's thread link, the pdf is in there.

Keith

 

[JohnRoberts]

Yes the schematic and article were posted.

We've discussed this before. It is a pretty straightforward SVF to generate a bandpass then add or subtract that bandpass to make the peaking section.

One thing I did in that design which is unique (AFAIK) is that instead of making the adjustments completely independent, I designed in an interaction between Q and boost cut, so as you dial in a broader Q the total amount of boost/cut is scaled back, and narrow Q were allowed much more boost/cut. This sounded very natural when equalizing full range material.

I didn't use this interaction in any of the professional designs I did because professional users want a Q knob with numbers on it that that don't change.

In that kit published 30 years ago, I used TL074 opamps and polystyrene capacitors which I suspect are still good by modern standards.

JR

 

[clintrubber]

[quote author="JohnRoberts"]One thing I did in that design which is unique (AFAIK) is that instead of making the adjustments completely independent, I designed in an interaction between Q and boost cut, so as you dial in a broader Q the total amount of boost/cut is scaled back, and narrow Q were allowed much more boost/cut. This sounded very natural when equalizing full range material.[/quote]

First of all, I don't mean: uhh, this isn't unique.

But to understand that Q-behaviour well, is it functionally the same then as what is done in those 550-API-EQs ?
If I understood those well, the peaks get sharper for higher boosts/cuts. Thought the term Proportional-Q or something was used.

Regards,

Peter

 

[JohnRoberts]

I didn't think the API 550 was parametric but that aside, I'm not talking about an EQ whose Q changes some with boost/cut, instead the boost/cut range on my EQ scales a lot when you vary the Q. This was dialed in empirically to give a first order correction for the change in loudness (or apparent boost/cut) when you varied the bandwidth of frequencies involved (from 1/6th of an octave to 2 full octaves). From memory, and this design was 30 years ago so I may not have the numbers exact, the boost/cut range at 2 octaves was something like only +/- 6 dB while at 1/6th octave the boost/cut adjustment range was around +/- 20 dB.

While not perfect on all sources it did cut down on the amount of fiddling with boost when you change Q or vice versa when EQing full range program material.

As was noted elsewhere in this thread there is not even full agreement between different manufacturers for how to specify Q in peaking type EQ sections, so I don't doubt different EQ topologies behave differently.

I still have not seen a "parametric" EQ that varied the boost/cut with Q the way I did, and I didn't even do that myself on my other professional designs because those customers want real numbers on their knobs, not just tick marks... If somebody is aware of a parametric EQ that intentionally varies the boost/cut with Q (published before 1978) I'll gladly retract my claim of originality. Changing the boost/cut more than a factor of 4:1 was not done casually or AFAIK by anybody else..

JR

 

[clintrubber]

I recall that automatic feature from the article back then and it sure sounds (& works) as intended: less knob fiddling.

You're right, what I brought up was related I think, but not the very same. First of all let's call the 550 a semi-parametric (and a switched one at that) since it doesn't have an explicit Q-control.

But for my understanding and NOT to debate your claim which imo still fully stands, say we increase the gain on a 550 from the first to the max boost setting, wouldn't that result in the same idea* for a set of curves when increasing the Q-control on your design - while its boost-control was already set at max ?

Best regards,

Peter

*: note I'm not saying exactly the same shape of curves etc, but the general idea

 

[JohnRoberts]

I don't know that this deserves so much ink (or bits) but I suspect the API apparent Q shift is idiosyncratic of the topology used. Since the definition of Q in boost cut/section is not concisely defined we will need to wait until the AES comes up with a standard definition before we can say what it does precisely. It may indeed change relative to whatever the standard definition becomes.

In any case this is still apples and oranges. I designed an EQ that can deliver either broad or narrow corrections, in small amounts or large amounts.

The 550 is a well regarded EQ, so whatever they did worked.

JR

 

[clintrubber]

It's sure apples and oranges, didn't claim them to be the same. All I said was that it made me think of that non-constant Q behaviour of said non-parametric.
Like you said, too much ink already about this non-confusion that seems to be a confusion.

End of OT folks, no go & build this parametric.

Regards,

Peter

 

[Animatic]

opacheco
I have been getting some good learning from this page.
Including state variable filter designs.

http://sound.westhost.com/articles/dwopa3.htm

Can't vouch for the author, but it seems clear enough.
Maybe not the deepest,
but I ain't ready for that yet anywhos.

 

[JohnRoberts]

Thats a good start but I can't fully agree with his statement "To make gain and Q individually adjustable requires the use of two-opamp summing stage (the first opamp)."

You can vary the Q independently of the BP gain with minor tweaks to that basic 3 opamp SVF. Instead of using R1 for signal input, apply the audio input to R2 (ground old R1 input). Now the BP gain is roughly unity inverting (I'd make R2 and R3 the same for exactly unity gain.)

Adding a resistance to ground from the junction of R2 and R3 will shift the Q to narrower bandwidth, without changing the BP gain. Alternately applying the resistance to ground at junction of R1 and R4 will shift the Q broader, also without changing BP gain.

To vary the center frequency without changing the gain or Q, vary R6 and R7 with a dual pot. Any pot mistracking can cause a slight Q change if the poles shift relative to each other. Note: intentionally offsetting the two poles is another way to vary the Q of that simple circuit without changing the gain.

JR