3 band stereo EQ w/ fixed frequencies and boost/cut

[wmtunate]

I've been asked to build a custom stereo 3 band EQ circuit for mastering: 315Hz Low-pass Shelf, 315 to 3.15K Bandpass, 3.15k high-pass shelf. Boost and Cuts on all bands. I've been looking over the Night EQ circuit and it seems simple enough to mod the high shelf frequency, but I'm having trouble wrapping my head around the low shelf and the bandpass. How do I do low-pass shelf with a gyrator circuit like these? How do I do a bandpass? I assume I can chain a low-pass shelf and a high-pass shelf to get a bandpass in this fashion.

Also, I need boost and cut +/-5dB in 1dB increments for each channel using 2-deck 12 position rotary switches so that both channels are controlled by one knob. Is there a formula I can use in this case to calculate resistor values for various dB boosts/cuts? I've looked over the Night EQ threads and I've seen posts about using rotary switches over the rev log 500k pots, but I haven't seen a post about replacement resistor values.

Now that I think about it, if I'm chaining a high-pass into a low-pass to get a bandpass, I'd need to control the boost/cut of the high-pass and low-pass at the same time. For a stereo unit, that would require a 4-deck rotary switch, wouldn't it? Is there a way to control the boost/cut for the bandpass section with a 2-deck rotary?

Sorry for all the questions, but this is my first custom EQ circuit and I'm trying to figure it all out. If there's an easier way to get what I need, I'm all ears. If at all possible, I'd like to avoid a passive design with a makeup gain stage at the end.

The circuit in question:
http://www.khstudio.us/DIY PICs/3D EQ - FULL.bmp

 

[PRR]

> How do I do low-pass shelf with a gyrator circuit like these?

There are no gyrators in the link you posted.

> How do I do a bandpass?

Look at U2a "650Hz". For a first-stab, assume all big caps are short, all little caps are open. Gain would be 1+(R14/R12) or 11. At infinite frequency all caps are short, so C16 forces gain to unity (not zero; also U2a now sees just 127 ohm load). For 0.2uFd and 1,270 ohms the corner is near 627hz. Now open that and insert the 2uFd at the 127R. Bass will fall-off to unity and the corner is near 627Hz. So everything below 627Hz is rolled-off, and everything above 627Hz is rolled-off: bandpass! (however without infinite rejection at unfinite frequency.) Because the two corners are at the same frequency, gain will never get to the full "11" calculated without caps. Being lazy I'd allow 3dB for each corner, assume 627Hz is down 6dB, peak gain is 5.5. My idiot assistant SPICE, given an unrealistically good "opamp", gives peak gain of 6.000 at 626.9Hz. The -3dB points are 250Hz and 1.5KHz.

Simple capacitor-scaling should give you 300-3K.

The shelf networks are the same minus one cap.

I do wonder why the impedances are SO low. It makes the op-amps strain at high frequency, and the big caps cost more (or may not be available in delicious flavors). Not for "noise"; look at the ~~50K nominal summing network. It looks like a truck of 1uFd caps overturned on the freeway, and the filter resistor values adapted to this windfall.

> I assume I can chain a low-pass shelf and a high-pass shelf to get a bandpass

Yes; but the broad "band" we want for mastering can usually be done in a single stage, as shown above.

And note that -this- EQ foes not cascade filters. It splits the range below 1.5KHz into four broad bands, and sums them. While you can boost a band, cut is somewhat limited: even if you snip the 160Hz pot, 650Hz and 40Hz and "Sub" (seems to be 10Hz?) will bleed 650Hz in their unity-gain zones. OTOH, with this kind of summing-EQ, not taking a zero (not shelving to unity-gain) will probably give enuff phase-shift to give deep nulls at some settings.

> if I'm chaining a high-pass into a low-pass to get a bandpass, I'd need to control the boost/cut of the high-pass and low-pass at the same time.

Why? Hi-pass, low-pass, then whack the combined signal.

I would not call this a "cut" EQ. The audio band is split, each band gained-up about 6, then those bands are summed with gain of 0.1/1/10 in the 500KRA+5K7 against the output amp's 56K2 resistor. There is a "bypass" which gives unity gain. Seems to me, with all controls centered, it gives gain of about 5 over the whole band. To get equal-level EQed or bypassed, the band pots have to be set well below centered. Then the amount of "cut" in any band is very limited. This seems to be a "boost everything, but not all the same amount" EQ.

 

[PRR]

me> There are no gyrators in the link you posted.

Ah, I just noticed how strange U1a is drawn. Yup, the 1K54 and 0.1u do resemble a Gyrator.

But ain't.

The cap is a power rail bypass; why this is shown when NO other power details are shown is odd.

The resistor REALLY belongs at R25 because....

me> Seems to me, with all controls centered, it gives gain of about 5 over the whole band.

Wrong. R25 R24 knock the signal down by about 12. (ALL of it, not just the 2.5KHz++ shelf, so why put it here? And share a "ground" with power-crap?) Then each band filter gains-up almost 6. The overlap between bands may be enough to give "unity" overall.... I could tell you but I'd have to charge you for labor. (Audio EQ design is VERY labor intensive... for one-off jobs, never from-scratch if there is any hope of stealing.)

And this 1/12 loss makes the super low impedances less of an issue. I suspect it is "possible" to have the 650 band straining at cymbals while U4a is not strained, but I admit it would need an extreme program. (But isn't this why mastering desks have extreme EQ, clients with extreme sonics?)

 

[tv]

I find C1 (470uF) into 10k way huge? Is there a reason for that?

 

[wmtunate]

Thanks for all the help PRR. That's the kind of info I needed.

However, there's gotta be an easier way to get what I need. I've been screwing around with this thing all day in simulation and nothing's working. I can't get a flat frequency response for the life of me with the gain pots at zero and changing cap values isn't having the intended results. This shouldn't be that hard.

I guess I'll go scour the metas again...

 

[wmtunate]

Damn, I should have searched harder earlier. This schematic looks promising:

And to think, it came from an old thread titled fixed frequency 3 band stereo eq anyone?

UTFSF next time Nate! :roll:

 

[wmtunate]

I've been working with this circuit for a few days now and it seems to perform pretty well in simulation. I was getting interaction between the low and high shelf due to them being on the same opamp, so I split them up and gave them their own section. This also made my capacitor calculations much easier.

I'm going to substitute OPA134s for the OPA604s as supplier stock seems to be pretty low across the board for the 604s.

I'll be using THAT1240 as the balanced line receiver and THAT1646 for the balanced line driver output. I'm also planning on putting Cinemag CMLI-15/15PCA trannies on the inputs and CMOB-2S on the outputs and having them switchable on the front panel for a little different flavor (in-line with the THAT chips).

Is there any issue I should be aware of with regard to switching between the transformer and IC balanced inputs and outputs? I'll be using relays, and I think that I should switch both the primary and the secondary of the transformers out of circuit when not in use so I don't have one winding in parallel with the signal coming into or leaving the THAT chips.

 

[clintrubber]

[quote author="wmtunate"]I was getting interaction between the low and high shelf due to them being on the same opamp, so I split them up and gave them their own section.[/quote]
Just curious, what kind of interaction ? Or overlap of EQ-activity ? That could be, but would it be interaction ?

It's pretty common to 'wrap' lots of bands 'around' one and the same opamp (API-EQs, SWR-bassamp-pre, ..., ... (endless list)), it can be done.

 

[wmtunate]

[quote author="clintrubber"][quote author="wmtunate"]I was getting interaction between the low and high shelf due to them being on the same opamp, so I split them up and gave them their own section.[/quote]
Just curious, what kind of interaction ? Or overlap of EQ-activity ? That could be, but would it be interaction ?

It's pretty common to 'wrap' lots of bands 'around' one and the same opamp (API-EQs, SWR-bassamp-pre, ..., ... (endless list)), it can be done.[/quote]

(In simulation, mind you) When set at max boost on the high band (I had it at about +8dB), the low band would cause the high band to drop in level as its boost was increased. +8dB on the lows dropped the highs down to about +3dB max. Same issue with cuts. The lows were not affected by the highs at all. The problem was mitigated somewhat by bumping up the 6.81k resistors on either side of the low band pot to around 15k each, but then I had to compensate by increasing the overall gain of that opamp up, which bumped up the max boost of the highs. I couldn't find a happy medium.

As soon as I separated them to their own opamps, the problem went away. I doubt it was EQ overlap, as the shelf corner frequencies were set about 3k apart.