Shelf/bell topologies in SVF parametric EQs

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
Paul Gold of Salt Mastering (who’s on here somewhere but I don’t know his user name) was talking for a while in the mid-teens about developing an all-shelving eq, two lows and two highs. That sort of reinforced my idea of using a four-band Net EQ topology and having the two lows and two highs both shelve. I’m sure it’s been done before.
There was a time when there was a craze about adding infra bass and ultra high controls to HiFi rigs. It was a cascaded Baxandall with very low LF and very high HF turnover. The idea was to recreate the spectrum lost in vinyl mastering. I think it ended when people were fed up with replacing woofers and tweeters.
 
@Gold what ever happened with your stunt eq idea?
It hasn’t happened. I still think it’s good idea. The lowest shelf would act more like an HPF. I would probably use it as cut only. Same with the top high shelf. I would use it as cut only. There is a design in the works that may make the four shelf idea moot.
 
Last edited:
It hasn’t happened. I still think it’s good idea. The lowest shelf would act more like an HPF. I would probably use it as cut only. Same with the top high shelf. I would use it as cut only. There is a design in the works that may make the four shelf idea moot.
It’s strictly for lathe cutting, right? Something like .5 or .75dB steps? 12dB/oct lowest shelf I presume?

The thing I’m working on centers around 650Hz, similar ranges as an MEA-2 but fewer frequencies, and steps starting at 1dB. More console EQ than transfer EQ, but very bus-applicable. Four bell/shelves, cut/boost, shelves all 6dB/oct. Funny thing is, while I’ve read your various thoughts about this, I never caught what sort of program material inspired the idea.
 
Last edited:
I'd be curious to retro-engineer this EQ.
I think it could be done in a fairly low noise way. I’m not sure if I like the high-q / 12dB shelf combo, though modular synthesis people would adore the feature. Maybe a pot section could auto-dampen Q range to restrict that shape. I can see people who might gravitate towards a Zahl desk being into a UTA desk. The UTA EQ is definitely an accomplishment in that channel strip width, sure looks like 35mm.
 
Last edited:
It’s strictly for lathe cutting, right? Something like .5 or .75dB steps? 12dB/oct lowest shelf I presume?
It would be a general EQ. I guess a filter slope can sometimes define a shelf shape but I don't like to describe shelves using a filter shape. The deluxe version of the EQ would have a bandwidth control to alter the shelf shape.
 
It would be a general EQ. I guess a filter slope can sometimes define a shelf shape but I don't like to describe shelves using a filter shape. The deluxe version of the EQ would have a bandwidth control to alter the shelf shape.
Right…I’m just thinking in poles for purposes of translation to circuitry. That bandwidth control would fade between two filters set to the same -3dB point, but one is two pole and one is single pole?
 
Right…I’m just thinking in poles for purposes of translation to circuitry. That bandwidth control would fade between two filters set to the same -3dB point, but one is two pole and one is single pole?
That hurts my head just thinking about it..

For TMI Back in the 70s I sold a kit parametric aimed at the consumer audio business. I engineered in a feature where the Q changed with boost/cut to generate a first order correction for the apparent loudness change with bandwidth. From memory the EQ delivered something like +/- 20 dB at 1/6th octave, and a more reasonable +/- 6dB at 2 octaves. Professional parametric EQs I designed around the same time had fully independent Q controls, ASSuming the customers knew what they wanted.

JR
 
I’m looking at the subtext of these traces. The fc of the shelf is the f0 of the bell, right? So with a separate filter for the shelf that replaces the bell in shelf mode, you would set the fc differently. By how much — an octave? The goal being to match fc of the shelf to f2 of the bell (for low shelf) and to match fc of the shelf to f1 of the bell (for high shelf)? I’m trying to think of the clearest way to ask this and failing.
 
That hurts my head just thinking about it..

Well, you wouldn’t need a SVF. In Porter topology, the same signal would feed a two-pole S&K and a single pole S&K. Then both go into an inverting mix amp to crossfade between the two. Then invert the result and feed it back into the cut or boost point in the main mix network. It’s the same signal though, so I suppose you could just slap a linear pot between the two outputs and attach the wiper to a (non-inverting) buffer, saving an amp section.

Actually yeah that’s better. 1P S&K + 2P S&K mixed into a the NI I/P of a voltage follower using a simple 5K linear pot. Unity gain result. Three op amps per filter.

Same approach in any cascaded loop of a W492, if I understand polarities in that circuit correctly.
 
Last edited:
I'd be curious to retro-engineer this EQ.
Many ways to skin this cat but this is non-inverting and uses the same parts for low or high shelf/bell fading. Simplest way or nah? You wanna sim it? If so I can add values and give it a mix amp (well, two mix amps…)

612CD3CF-31EF-4BF3-B711-6827CF0C462E.jpeg
 
Last edited:
Okay, so I found an example of BP+LP and BP+HP being used for shelving. It’s the Sony MXP-3000 console EQ - the SVF version, not the Wien bridge one - which seems like a lot of people have had their hands on. I did find some older forum posts where users greatly preferred the Wien version.

It’s pretty straightforward — in shelf mode, Q is disabled and the BP output is replaced by a differential amp that sums BP+LP or BP+HP. The integrators do not shift in frequency (which is very attractive), and the outputs are mixed in fixed ratios. For the low shelf, BP is a little lower than LP. For the high shelf, BP is way lower than HP. But they’re both in there and (I assume) sum to the same level of just the BP in bell mode. Maybe that’s the trick, getting these ratios right. It’s clear that the designer started with equal amounts (BP and LP both at 50%, for example) and tweaked from there.

Maybe there’s an MXP-3000 user out there who can comment on this sound or usage?

@Brian Roth might you have the MCI JH-6XX EQ in schematic form somewhere?
 

Attachments

  • sony mxb-3000 eq.pdf
    877.7 KB
Last edited:
The simplest possible way!

What do you think of the sound of these shelves compared to dedicated single pole passive buffered ones? Also, when you have cut or boost engaged and switch from bell to shelf, is the spreading in frequency emphasis significantly noticeable? As in, can you smoothly evaluate the difference between the two settings or do you just have to recalibrate your brain?
 
the Calrec is very intuitive in use, to the degree where I think it might have been a specific design goal.
That’s inspiring enough to pull me away from separate shelf filters. Fewer resistors and muxes, that’s for sure.

Like also the single "press for high-Q" button..
Do you know off hand what the low and hi-q settings are? I’m not sure I understand the effects of modifying both the input attenuation and the ground tie at the same time; usually it’s just one or the other, regardless of polarity, and every once in awhile you see the Soundcraft technique (which I know is John’s preferred Q adjust). Seems to be the same on all bands, the only difference being that the shelf switch forces low-Q. I’d take a link if this has already been addressed elsewhere, but i didn’t see it on your site or anybody’s build commentary.

3BB28051-EED5-4FF6-A75B-9B245260E306.jpeg
 
Last edited:
I didn't know that I had a preferred Q adjust.... :unsure:

From a quick glance at the posted schematic i know that is not my preferred frequency adjust topology. Potentiometers are generally 20% bulk resistance tolerance. Using them as rheostats is not very precise/repeatable for production. Using them ln true potentiometer configuration with bottom grounded and frequency resistor fed from the wiper use the ratio not resistance. An additional resistor feeds around the pot into the integrator, establishing the LF end limit when the pot is fully down. This way the pot tolerance is not a factor when full up (max) or full down (min). For the schemo shown the pot is only out of the picture for max frequency.

JR
 
Back
Top