Shelving EQ

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

Gold

Well-known member
Joined
Jun 23, 2004
Messages
3,670
Location
Brooklyn
I want to build a four band "fully parametric" shelving EQ. In other words  I'd like to have a shelving EQ that has variable bandwidth (slope but I dislike that term as applied to shelves) in addition to frequency select and gain. I like shelving EQ's and use them quite often. I've played around a little with the DMG Equilibrium EQ plugin which can have this feature. I think it is worth pursuing.

The first question I have is what topology would you choose? From what i can gather a State Variable Filter wouldn't make this straightforward as to get a shelf you have to combind the filter output(s) with the bandpass output.

The T-Filter (Sontec) type shelf sounds great but I'm not sure is variable bandwidth is practical for a shelf.

What would you do?
 
I put parametric EQ with optional shelving on the highest and lowest bands, in a  Loft console input strip back in the early 80s.  I used basic SVF topology. It could switch between peaking and shelving for the two extreme bandpasses. For the shelving sections I summed the BP + LP (or BP+ HP) to generate the shelf. It involved inverting one passband of those two and Q behavior was pretty unconventional... IIRC I necked back the Q adjustment range during shelving so it didn't get too extreme...  The first rule of product design is make it hard for the customer to make bad sounds come out...

Sorry no schematics, and only fuzzy recollection of precisely how I did it all those decades ago, but definitely based on state variable topology.  I probably included HP and LP poles in series with boost circuits so high and low shelving did not provide too much boost at frequency extremes. Back in those days the results were printed to magnetic tape so had limited frequency response.

JR
 
fazer said:
Night eq is is kind of like what you describe but no q control .  Just a thought.

Thanks, I didn't know this was an all shelving EQ. People generally seem to like it. It sounds worthwhile to build a pair and see if I like it. It looks like Gustav offered the boards in the past but they aren't on the current list. Do you know if anyone is selling PCB's? If the Gerbers are available I guess I could do a run and sell what I don't want?
 
JohnRoberts said:
I put parametric EQ with optional shelving on the highest and lowest bands,

I have plenty fully parametric bells but no shelves. This will be shelves with no peaking EQ bands. This is only for me so it doesn't need to be totally idiot proof. Given that would you still have chosen SVF?

The major design issues I see are getting the faceplate text to be meaningful. With a high shelf, using the center frequency, which is the top of the shelf doesn't make sense to me. It makes more operational sense to label the "bottom" of the shelf as the target frequency. If you do that then changing the bandwidth would change the faceplate frequency label. So changing the bandwidth is kind of like changing the frequency if you want to use the bottom of the shelf as the frequency label. I
 
john12ax7 said:
Using the center crossover frequency fur the labeling might make the most labeling sense if doing variable Q shelving.

It makes sense from an easy labeling perspective. It makes less sense from a user perspective.  It’s not how most perceive a high shelf.

I think Abbey said in one of the SVF threads that he defines a shelf bandwidth as where the bottom of the skirt is with 15dB of gain applied.

A shelf with a center frequency of 10K and a bandwidth that lands the bottom of the skirt at 1K will be perceived as a 1k high shelf not a 10K low shelf.

I think that if this were ever to become a commercial product the vast amount of time would be spent tweaking the high shelves to conform to the labeling that makes most sense to the ear.
 
Gold said:
It makes sense from an easy labeling perspective. It makes less sense from a user perspective.  It’s not how most perceive a high shelf.

I'm not necessarily disagreeing,  but with changing Q the bottom and top parts will be changing,  while the center frequency should remain unchanged.  So there is a technical basis for it,  I think that's what Jakob and also Dangerous do in some of their products.

Gold said:
A shelf with a center frequency of 10K and a bandwidth that lands the bottom of the skirt at 1K will be perceived as a 1k high shelf not a 10K low shelf.

For a high shelf,  labeling tends to be above the crossover frequency.  For a low shelf it's below (not sure if that's what you mean).
 
john12ax7 said:
For a high shelf,  labeling tends to be above the crossover frequency.  For a low shelf it's below (not sure if that's what you mean).


I’ve seen it labeled both ways.
The Sontec calls the high shelves by their center frequency. The Maselec calls them more from where they are perceived. There must be some trickery going on to make that happen because you get the “10K shelf” when you have the frequency set to 1K and you switch the Q switch from bell to shelf. Pretty slick IMO.
 
Where are you getting the extreme 1k and 10k? I looked at the Masselec manual and the sweeps suggest they are using a typical knee type definition.  At +8dB boost they are defining the frequency at +6.5dB.
 
The graph in the manual doesn’t show the whole picture. It doesn’t link the front panel labeling to specific curves in the graph.

What I’m saying is that if you have frequency select set to say 1K. In bell the center frequency is 1K but in shelf mode it sounds like the shelf is rising from 1K. That means the center frequency of the shelf is something like 8K-12K.
 
Gold said:
Thanks, I didn't know this was an all shelving EQ. People generally seem to like it.

The Night EQ is not an all shelving EQ. It has a 2 high shelf curves, the other are Peak with fixed Q.
On is fixed at 2.5khz and can cut and boost.
The other is frequency switchable (2.5, 5, 10, 20 and 40khz) and it's boost only.

iyt3sYJ.jpg


zI5b2nI.jpg


I really like it, and I use it a lot on the mixing bus and in Mastering.

Gold said:
It sounds worthwhile to build a pair and see if I like it. It looks like Gustav offered the boards in the past but they aren't on the current list.
Do you know if anyone is selling PCB's? If the Gerbers are available I guess I could do a run and sell what I don't want?

PeterC designed and sold the pcb's not Gustav. I think I still have some PCB also that I'm not going to use.
But before building it you could try the Maag plugin emulation just to see if you like it and if you think you want to build it.
The plugin does it quite well, I used the original for years and I have a clone I built myself.

https://www.plugin-alliance.com/en/products/maag_eq4.html

MhGulSul.png


I think Fazer idea is quite good,
I just don't know if it's possible to modify the Night EQ shelfs circuit to have variable bandwidth/slope, but it's definitely work a check and if it is possible then I'm sure you'll be happy with it
 
I don’t think the Nite EQ is close enough to what I have in mind.  I definitely want a four band shelving EQ. Two low bands and two high bands. 

In use the first low band will be used mostly cut at very low frequencies. The second low band will be mostly boost and start just below 1K . The two high frequency sections have the first one start just below 1k and is mostly boost. The second high frequency section is for very high  frequencies and is mostly cut.

It’s kind of like a tilt EQ or a Baxandall with separate control of low and high. With high and low cut.

I’ve thought about very simple like a swinging input. The super deluxe version would be with variable bandwidth.
 
You could easly do two cascaded Baxendall's with different turnover frequencies; it has been done before. IIRC commercial realizations limited B/C to 6dB or so. You don't want to have a section with too close LF and HF because of impredictable interaction, so one has the lowest LF and the lowest HF, the other has the highest. It works.
Making a variable-order (slope) shelving EQ is almost against nature in the analog domain. Mathematically it has limitations you ought to be aware; anything with a Q>1.4 will result in over and undershoots, so actually a boost will start with a cut and finish with an overshoot. Not very user friendly after all. Cascading 1st-order sections seems like a good idea but turns out to provide very little enhancement to the slope, you would have to cascade 4 sections to start to see an improvement.
 
abbey road d enfer said:
You could easly do two cascaded Baxendall's with different turnover frequencies; it has been done before. IIRC commercial realizations limited B/C to 6dB or so.

That’s an interesting idea. There are plenty of Baxandall style PCB’s available so it’s practical too. I wouldn’t need more than 6dB. I have 12 position switch’s I want to use for gain. I was planning on doing boost only and cut only bands in half dB steps.

Making a variable-order (slope) shelving EQ is almost against nature in the analog domain. Mathematically it has limitations you ought to be aware; anything with a Q>1.4 will result in over and undershoots, so actually a boost will start with a cut and finish with an overshoot. Not very user friendly after all.

Okay I’ll give up on that idea for now. I guess I’ll wait until I’m able to use circuit simulation software to play with variable slope.  I think there is a unique and useful design in there somewhere. 

I plan on starting with cascading the shelf sections of the Neumann 492  because I have the PCB’’s and it should be quick to get it going to check out. If I hate it I’ll look for something else. If not it may be fine until I figure out something  better.
 
The Pultec EQP1A has separate shelving boost and cut controls for it hi and lo bands. It woiuld be quite simple to modify the turnover frequencies of these to give you the separate bands you need.

Cheers

Ian
 
abbey road d enfer said:
You could easly do two cascaded Baxendall's with different turnover frequencies; it has been done before. IIRC commercial realizations limited B/C to 6dB or so. You don't want to have a section with too close LF and HF because of impredictable interaction, so one has the lowest LF and the lowest HF, the other has the highest. It works.
Making a variable-order (slope) shelving EQ is almost against nature in the analog domain. Mathematically it has limitations you ought to be aware; anything with a Q>1.4 will result in over and undershoots, so actually a boost will start with a cut and finish with an overshoot. Not very user friendly after all. Cascading 1st-order sections seems like a good idea but turns out to provide very little enhancement to the slope, you would have to cascade 4 sections to start to see an improvement.

Exactly... as I already mentioned I reduced the Q adjustment range in shelving mode because the underdamped (boost before cut) response was not intuitive for the EQ operator (IMO).

JR
 
JohnRoberts said:
Exactly... as I already mentioned I reduced the Q adjustment range in shelving mode because the underdamped (boost before cut) response was not intuitive for the EQ operator (IMO).

JR
When Steinberg introduced the 1st version of VST, their EQ algo was based on a digital implementation of the hardware SVF, which resulted in the same underdamped response, whatever the actual Q setting. Devs were convinced thye had done the right thing; audio engineers pulled their hair trying to understand what happened. The brain received several conflicting sensations.
 
ruffrecords said:
The Pultec EQP1A has separate shelving boost and cut controls for it hi and lo bands.

I’m not after the Pultec boost cut thing.  I thought the boost and cut were different shapes. One a bell and the other a shelf? I want shelves only.

Isn’t your Helios EQ high and low Bax style shelving with a bell mid band?  I could probably cascade two of those  to implement Abby’s idea. I’m not sure about the slopes I want.  A Bax  isn’t as flexible as some other topologies.

I’m also considering lashing up a swinging Input version because it needs very few parts and is easily done on perf board. Normally there would be unacceptable interaction between bands but with the boost only and cut only arraignment I don’t think it’s a problem.
 

Latest posts

Back
Top