|
|
Post by matt@IAA on Mar 17, 2019 19:05:24 GMT -6
Ive been tinkering with a few different EQs and one that really has intrigued me is a state variable filter design. These are nice because of the wide flexibility - adjustable Q, infinitely variable frequency and gain. The downside is there is some complexity... each band takes at least four op amps.
I laid one out that uses ten IC op amps and THAT chips for balanced in/out. It is inspired by the UREI 545. Low pass and high pass filters, three fixed bands and a fourth with a range switch.
Then I laid out the pcb for an all discrete version using a socketed 2520/990 for the gain amp and three simple discrete op amps for the filters. I limited this design to just the parametric EQ for simplicity sake. Of course it’s easy to add a socketed input and transformerbalanced output stage. The question is... which one sounds more interesting?
Clean and neutral with TL072s and no iron? Or all discrete with some mojo?
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Mar 18, 2019 3:15:42 GMT -6
At least in modular synth land (of which I know a bit, had a 22u monster at one stage!), SVFs are known for being A) extremely versatile (you can have parallel LPF, HPF, BPF and Notch outputs), but B) slightly boring in sound, when compared to more well known/classic designs such as Transistor or Diode Ladder (although I think the original Oberheim SEM filter was an SVF). But I assume modular SVFs are almost always designed with chips, not discreet opamps, for size and cost considerations. A long way of saying, maybe build one, try a few things, and hopefully the trannies and discrete opamps will give you enough colour? That would be my choice, if you want something with a bit more vibe than sterility.  Will obvs add to the cost though! Wish I had the know how and tools and skills to build one! |
|
|
|
Post by svart on Mar 18, 2019 9:26:25 GMT -6
I'm a huge fan of the CAPS type of variable filter, Constant Amplitude Phase Shift. It's a great design, in that Q doesn't change over gain like in state-variable and that you can get much tighter Q's, but it suffers a huge drawback in that it needs ganged pots with both log and anti-log or else it doesn't track well. These are impossible to get, and impossible to fudge with law bending tricks, which leaves only custom pots (read:$$$).
I find that the discrete vs. integrated argument for EQ tends to favor integrated for EQ in general since phase matching is somewhat crucial.
I personally think the MC33079 is a good opamp for EQ, although it's higher BW tends to oscillate in phase circuits if there isn't enough NFB around the gain. Their outputs run hotter, close to class A, which is apparent in the lopsided slewrate specs.
|
|
|
|
Post by matt@IAA on Mar 22, 2019 18:05:44 GMT -6
I’ve kept tinkering and moved onto a parallel arrangement inspired by a Sontec. All discrete, op amp balanced input and transformer balanced out.
|
|
|
|
Post by ppa on Mar 22, 2019 19:10:07 GMT -6
At least in modular synth land (of which I know a bit, had a 22u monster at one stage!), SVFs are known for being A) extremely versatile (you can have parallel LPF, HPF, BPF and Notch outputs), but B) slightly boring in sound, when compared to more well known/classic designs such as Transistor or Diode Ladder (although I think the original Oberheim SEM filter was an SVF). But I assume modular SVFs are almost always designed with chips, not discreet opamps, for size and cost considerations. A long way of saying, maybe build one, try a few things, and hopefully the trannies and discrete opamps will give you enough colour? That would be my choice, if you want something with a bit more vibe than sterility. Will obvs add to the cost though! Wish I had the know how and tools and skills to build one!
the matter is how the state variable filter is made, there are several famous eqs that are based on it that sound good and not boring. BTW the state variable filters used on synths are different vs. those used on studio recording parametric eqs, because very often they are voltage controlled and they use different parts to change their parameters.
|
|
|
|
Post by ppa on Mar 22, 2019 19:42:16 GMT -6
for example the Neumann W492 equalizer is based on state variable filters, the same thing for Barry Porter NEteq, amek angela eq and others.
edit: BTW the Neumann W492 equalizer is very fine with the synths, isn't it? |
|
|
|
Post by matt@IAA on Mar 22, 2019 20:12:22 GMT -6
for example the Neumann W492 equalizer is based on state variable filters, the same thing for Barry Porter NEteq, amek angela eq and others.
edit: BTW the Neumann W492 equalizer is very fine with the synths, isn't it? The W492 is almost a like for like clone of the Urei 545, or vice versus...right down to the peak indicator approach. Just uses much better op amps (5532/4 vs 4741s). |
|
|
|
Post by ppa on Mar 23, 2019 10:15:25 GMT -6
these eqs have all cells in series, I put all cells in parallel to reduce the signal path, just to make a modern version of SVF eq.
|
|
|
|
Post by matt@IAA on Mar 23, 2019 11:26:43 GMT -6
Porter NetEQ is parallel also. Fun part with parallel is you have to either have a boost / cut switch or a center tapped pot to keep the levels right.
|
|
|
|
Post by ppa on Mar 23, 2019 15:39:49 GMT -6
Porter NetEQ is parallel also. Fun part with parallel is you have to either have a boost / cut switch or a center tapped pot to keep the levels right. I designed it for diy and it uses classic linear pots for gains with no problems. It can use the classic 16mm Alpha pots too, and other ones. |
|
|
|
Post by johneppstein on Mar 23, 2019 18:20:51 GMT -6
State variable filters are the type used in the vast majority of chip based consoles that have full parametric EQ. They can sound quite good but the problem is that they can get somewhat phasey sounding on more extreme settings.
|
|
|
|
Post by matt@IAA on Mar 25, 2019 12:09:08 GMT -6
I drew up a PCB that is a schematic clone of a UREI 545 EQ with a few mods.
The original used Harris 4741 quad op amps. I've re-drawn it with duals, and penciled in TL072s or 5532s.
The original also used an output transformer, where I made it use THAT chips, the 1246 and 1646 for I/O. I suspect this'll clean it up a lot. It also has a header for unbalanced in and out points, though, so you could pair this up with a mic pre or an output line amp if you wanted.
I also included two jumpers to bypass the low and high filter sections, leaving a four band parametric design if desired.
I kept the fourth band as a variable design with a switch, which could use either a rotary or a 3PDT (wired in). You could also use it as a fixed band by using one cap and a jumper.
Also, everything uses off-the-shelf parts, particularly 9mm pots (like Bourns PTD or 51 series, or BI/TT P09x) and such. It fits in a 1RU, and you could definitely stack them in a 2RU for a stereo version.
If there's interest, I'll order a batch of PCBs.
|
|
Will obvs add to the cost though!
