Barry Porter "Net EQ"

[john12ax7]

Constant Q sees BW increasing with the amount of B/C, not so much with constant BW.

BW as defined how? If you are calling something constant BW you still have to measure it from somewhere, so you are just introducing other new problems in definition. Though flawed, I don't find it unreasonable to define Q with -3dB points after you've exceeded some nominal gain.

The idea of constant BW also seems misleading, as you need BW to change with Fc. So in practice constant BW would not be constant in either B/C or Freq sweep.

Imo proportional vs fixed Q are better practical descriptors.

 

[abbey road d enfer]

BW as defined how?

That's why I wrote; "you're invited to take it with a pinch of whatever, because no proper definition of BW for a bi-quad has ever been agreed upon"

If you are calling something constant BW you still have to measure it from somewhere, so you are just introducing other new problems in definition.

Exactly; several people have submitted ideas about it, but a consensus could never be reached.

Though flawed, I don't find it unreasonable to define Q with -3dB points after you've exceeded some nominal gain.

It is IMO. What "nominal gain" should be exceeded? Again a subject of debate. How would it work for a "mastering EQ" with +/- 3dB B/C?

The idea of constant BW also seems misleading, as you need BW to change with Fc.

BW should be in relative units, of course; that's why octave and decades have been invented.

So in practice constant BW would not be constant in either B/C or Freq sweep.

Neither is Q.

Imo proportional vs fixed Q are better practical descriptors.

What are the characteristics of both? Since I don't know what Q is. I know what BW is, my ears tell me.
Proportional Q? Proportional to what? Proportional to B/C would be in contradiction with the ofetn claimed assertion that Q is a given parameter of the native filter.
Often, constant Q has been described as "non-symmetrical boost and cut". This is a peculiarity of some EQ's where it is the result of the topology, although using the same filter.
It is a can of worms. My peeve is that people use improperly a known concept, and draw conclusions from there.
Words (for that matter, symbols) are important in science.

 

[karloff70]

For the gain

4 x dual gang center tapped 10K

Shelving filters are in series with the parallel bands section, and while its set up to be either/or, it doesnt have to be.
The addition of a layer on the high/low band is an interface related/space saving decision.

If you separate the shelving filters, you will need.

4 x 4 gang 10K for the frequencies for all bands.
2 x 2 gang 10K for the shelving

If you dont, you will need

2 x 4 gang 10K for the low mid/high mid bands
2 x 6 gang 10k for the low/high bands

I cant remember the value of Q pots, but probably 100K?

4 x 2 gang 100K

Gustav

Awesome, thank you!!

 

[karloff70]

Actually, I don't see it at all. The only difference is the Boost/Cut taper is slightly different (more cramped at extremities) and probable noise penalty with the non-tapped pots. But there is no interaction, in the sense that the response resulting from two adjacent bands is the sum of the response of each band. At least, that's what the simulation says. Maybe someone can check on the actual product.


FWIW...
What is the benefit of constant Q? In practice, it means that the higher the boost/cut, the larger the BW. Is it desirable? My personal taste if for constant BW.

Just to point out, I am not capable to hold that conversation. lol. I merely quoted what Barry's own notes say, that are included in the pdf folder going round here. I can't argue either way from any knowledge of mine.

 

[weiss]

Just to point out, I am not capable to hold that conversation.

Me neither. I just thought, the issue that the bands in my eq influence each other could be due to the fact that i'm using untapped pots. I gave up searching the right potentiometers and switches for B/C don't fit my frontpanel. But honestly it doesn't matter to me anymore, as long as it sounds fine.

 

[abbey road d enfer]

Me neither. I just thought, the issue that the bands in my eq influence each other could be due to the fact that i'm using untapped pots.

Can you show graphs of these interactions? That is interesting because I just don't see it in the simulations.
I know simulation is not real life, but still I expect some kind of convergence.

 

[john12ax7]

BW should be in relative units, of course; that's why octave and decades have been invented.

Oops. I read it late and was thinking of Hz, neglecting octaves and decades.

What are the characteristics of both? Since I don't know what Q is. I know what BW is, my ears tell me.
Proportional Q? Proportional to what?

People can hear narrow vs sharp, but it's hard to quantify exactly what that is as there are not precise definitions. I'm just not seeing a great distinction between calling something constant Q vs constant BW.

If you go from say 6dB to 15dB B/C,,the -3dB BW would vary with B/C for proportional but stay somewhat consistent with fixed.

 

[JohnRoberts]

That's why I wrote; "you're invited to take it with a pinch of whatever, because no proper definition of BW for a bi-quad has ever been agreed upon"

Exactly; several people have submitted ideas about it, but a consensus could never be reached.

It is IMO. What "nominal gain" should be exceeded? Again a subject of debate. How would it work for a "mastering EQ" with +/- 3dB B/C?

BW should be in relative units, of course; that's why octave and decades have been invented.

Neither is Q.

What are the characteristics of both? Since I don't know what Q is. I know what BW is, my ears tell me.
Proportional Q? Proportional to what? Proportional to B/C would be in contradiction with the ofetn claimed assertion that Q is a given parameter of the native filter.
Often, constant Q has been described as "non-symmetrical boost and cut". This is a peculiarity of some EQ's where it is the result of the topology, although using the same filter.
It is a can of worms. My peeve is that people use improperly a known concept, and draw conclusions from there.
Words (for that matter, symbols) are important in science.

I have whined about this before (probably even in this thread). Over a decade or two ago I wasted time and brain cells trying to get the AES Standards Committee to work on a formal definition (or definitions) for q/bandwidth in boost/cut EQ sections. There are well accepted definitions for basic bandpass filters (based on half power points), but once you configure them into boost/cut EQ sections different topologies deliver different results. At max boost/cut the Q/bandwidth gets close to the underlying BP sections that we do know how to define.

JR

PS; I know Bill has told me how easy it is to get the AES standards committee working on something so I must have a personal problem (or this definition is hard).

 

[abbey road d enfer]

I'm just not seeing a great distinction between calling something constant Q vs constant BW.

The distinction I see is that BW can be understood by anyone who has a musical ear, when Q is a malapropism. I bet you can ask a trained listener to evaluate then BW of an EQ and get a decent answer; if you ask the same person to evaluate the Q, he may very well answer huh?

 

[abbey road d enfer]

If you go from say 6dB to 15dB B/C,,the -3dB BW would vary with B/C for proportional but stay somewhat consistent with fixed.

Actually it's the other way around, although BW varies in both cases really.

 

[john12ax7]

Actually it's the other way around, although BW varies in both cases really.

Why other way around? That would go against logical interpretation.

 

[abbey road d enfer]

Why other way around? That would go against logical interpretation.

I have misinterpreted your sentence. Anyway, there can be no "logical interpretation" of a concept that is fundamentally wrong. Q does not apply to EQ"s, period.
And the "constant-Q" circuit used in some Rane GEQ's actually produce bands that go wider and wider as the amount of B/C increases. It's a matter of topology.
And I maintain that nothing is constant in an EQ, neither a supposed Q nor the BW, whatever "standard"metrics used to define it.

 

[Gold]

Using the graph in lost #578 I think of constant bandwidth as 0dB gain as being at the top of the bell. If the X axis line represents gain, as the X axis moves down the Y axis gain increases. The bandwidth stays more or less constant for a while.

For constant Q I picture 0 dB where the bell skirt equals 0. Again if the X axis line moves up the Y axis line gain increases and the bandwidth changes a lot with gain.

I’m nit proposing this is technically accurate. Just how I interpret the meaning of the language.

 

[abbey road d enfer]

Using the graph in lost #578 I think of constant bandwidth as 0dB gain as being at the top of the bell. If the X axis line represents gain, as the X axis moves down the Y axis gain increases. The bandwidth stays more or less constant for a while.

For constant Q I picture 0 dB where the bell skirt equals 0. Again if the X axis line moves up the Y axis line gain increases and the bandwidth changes a lot with gain.

Can you describe it graphically, because I just can't figure out.

 

[JohnRoberts]

I'm not going to search for links but Dennis Bohn from Rane has written extensively about constant Q wrt GEQs. It was an effective marketing push for them (decades ago).

JR

 

[john12ax7]

I have misinterpreted your sentence. Anyway, there can be no "logical interpretation" of a concept that is fundamentally wrong. Q does not apply to EQ"s, period.

You have not given a sound reason for BW being ok but Q is not. BW is traditionally defined at -3dB points a well, so what happens at <3dB B/C?

Claiming that musicians can better understand BW is not sufficient justification for your position. Q and BW have an established relationship, if one is flawed they both are. It is fine too have a personal preference, but there is no justification for arguing one is objectively superior to define an EQ. We are dealing with opposite sides of the same flawed improperly defined coin.

 

[abbey road d enfer]

You have not given a sound reason for BW being ok but Q is not. BW is traditionally defined at -3dB points a well, so what happens at <3dB B/C?

I wrote: "the BW, whatever "standard"metrics used to define it." That means I consider the standard -3dB definition of BW to be inadequate for EQ's, but that it would make sense if we agreed on a consensus. We could call it range or give it the name of a famous audio designer, it would make more sense than Q.

Claiming that musicians can better understand BW is not sufficient justification for your position. Q and BW have an established relationship,

No. Q applies to 2nd-order numerators with asymptotic skirts to minus infinity.

if one is flawed they both are.

No. The relationship between Q and BW is adequate for a certain type of circuit, not all. Do you think it's valid for a 4th-order filter?

It is fine too have a personal preference, but there is no justification for arguing one is objectively superior to define an EQ.

My position is based on the fact that Q is a mathematical entity that does not correspond to any human perception. I believe the concept of width, range, extent... is more adequate.

We are dealing with opposite sides of the same flawed improperly defined coin.

We can agree on that.

 

[john12ax7]

No. Q applies to 2nd-order numerators with asymptotic skirts to minus infinity.

Q applies to many things you are neglecting, it is not only that particular case.

No. The relationship between Q and BW is adequate for a certain type of circuit, not all. Do you think it's valid for a 4th-order filter?

Q can be adequate for a variety of things. When doing RF communication systems when someone says we need a higher Q for a stage, everyone understands it to be a more narrow selective passband. The implementation will not be a strictly 2nd order filter that goes asymptotic to infinity.

My position is based on the fact that Q is a mathematical entity that does not correspond to any human perception. I believe the concept of width, range, extent... is more adequate.

Words are just verbal entities, they still need to be learned, I know musicians (who play well) who don't' know what frequency is. We often need to attach mathematical or other entities to human perception. There is nothing innately special with words like width and range, so I don't see the need to discriminate against other mathematical entities. The main thing would be for the concepts to be conveyed. If there is enough consistent historical use a descriptor can be valid, whether mathematical or not.

 

[john12ax7]

I'm not going to search for links but Dennis Bohn from Rane has written extensively about constant Q wrt GEQs. It was an effective marketing push for them (decades ago).

JR

I looked it up and can be found here Constant-Q Graphic Equalizers

Attached are the graphs of what he is calling Q that varies with boost/cut vs constant Q.

 

[moamps]

......If there is enough consistent historical use a descriptor can be valid, whether mathematical or not...

The story of Q
http://www.collinsaudio.com/Prosound_Workshop/The_story_of_Q.pdf