Two filter quacker

[strungout]

Oy.

A while ago, I wanted to put (basically) two quackers (Dr. Q/Dr. Quack/NQ) in parallel, and Transmogrifox suggested I just needed to double the filter section. I've decided to give it a go, but I wanna know if I have the right idea heading into this, so:

Sorry the image isn't very clean. I blame soft graphite and an eraser that smears, and an uncooperative Photoshop. The values aren't definitive, I'll of course mess around with em alot. I was wondering if I need another tranny (in series with the wiper of the range pot) to modulate the second filter? Any comments, suggestions will be appreciated.

Ciao.

[Tim Escobedo]

You'll most likely need separate transistors for each filter.

I've never tried MFB filters quite like those.

[puretube]

put in 2 series output resistors in IC1b & IC2 (~2k2), and 1 to ground at the summing point (100k);
add a pulldown R (1M) at the input, too.

[Mark Hammer]

Look at the schem of the Baseballs, which also uses two parallel filter sections and bipolar transistors as voltage-controlled paths to ground.  That will show you how to connect the two resistors as suggested.

I would hike up Ton's resistor suggestion a notch by suggesting you put in 2k2 resistors AND a 10k panpot to blend the two filter outputs.  The wiper of the pot goes to your output cap (where the two op-amp outputs currently sum at), and the two outside pot lugs go to one side of each of the series resistors Ton suggested.

P.S.: Why are the two filters seemingly tuned to the same range?

[Transmogrifox]

Looks good.  It looks like the other guys covered the details I saw like the extra tranny and summing resistors.  I liked Mark Hammer's idea with the blending pots.

You could add some extra fun with perhaps a backward sweep on one of the filters so one goes down as the other goes up.

Also, if you put the filters in series (maybe as an extra DPDT switch option), you will have a 4th order filter like a synth filter.  Slightly stagger tune them and you will have an interesting response.

[strungout]

Thnx for the suggestions

Well, this was simpler than I thought it would be, actually, I didn't have to change much to get it working. As is, I added a second tranny in EF, the resistors from each filter outputs summing to a 10k pot feeding the output of the circuit. I lowered the sweep of the bass section with two 6.8nf caps. Blending between the two filters is pretty interesting, in fact it gave me the basis for another effect idea I had a while ago  

Now I need more experimenting. I used the "bass" and "normal" sections, but the "normal" section has an attenuation effect on the signal, added to the sweep, it sounds too thin... I'll see how it sound if I lower the sweep more, but I have a strong feeling I'll just be adding copy of the "bass" section and tune the sweep higher.

I tried uneven values, that got pretty interesting, in the filters, like 10nf from the OA output and .22uf from the OA's inverting pin, it produces a nice sweep but it creates some distortion (not bad, but not what I'm going for). Changing the 470k feedback to 330K made it cleaner
but it lost some of that depth Oh well. Experiment goes on.

The lack of uniformity on the part of the EF reaction to what note is plucked and range setting and sensitivity and attack and decay is still annoying. So, I was trying to sleep and got this idea: assuming that the EF is set so that the attack on high notes is slow enough to hear the sweep going up, that means that the low notes will fire the sweep up real fast (if you pluck with the same strenght as with the high notes), so: what adding a filter after the sensitivity pot to attenuate the lows? They wouldn't trigger the sweep as easily, right? Anyway, just an idea.

There seems to be losta fun to be had with this circuit (like Transmogrifox's ideas).

And the filter were seemingly tuned to the same range because I just drew the circuit up to get a general idea of the configuration needed.

Ciao.

[Mark Hammer]

When using LDRs for producing sweep, it gets easy to adjust how much sweep occurs for each filter section.  You can sort of do that by having a variable resistor to ground in parallel with the transistor,but it isn't as flexible as being able to stick resistances in parallel and series with an LDR.

But all of that aside, I think the smart thing to do is to have separate envelope followers for each filter instead of playing around with filters after the rectifier.


This provides several advantages.  First, if you intend to do any band splitting (and my sense is that you do), different bands provide different signal amplitudes so being able to set the follower for mids and highs a little hotter than the one for lows would be helpful.  If you want to set the rise/attack time differently for each filter section this also helps because increasing the attack resistor (100R in what you posted) reduces sweep amount, as well as speed, requiring cranking up the sensitivity a bit to compensate.

Setting each follower to be differentially sensitive to different bands is a simple matter of varying the input capacitor to that stage (presently .05uf) and/or sticking a cap in parallel with the feedback resistor (e.g., a 100pf cap and 2.2M resistor rolls off signal above 720hz, producing less "drive" for those notes).  You could very easily stick in a couple of toggles or dipswitch to quickly change the "response band" for each follower/filter combinations.

Of course at this point it starts to move closer to the new E-H Bi-Filter, doesn't it?

Glad to hear the filter panpot worked out.  One of the things it lets you do is deliberately use the distortion produced by increasing the 479k feedback resistor for more gain/Q, but still fade it back, relative to the other filter.  The panpot and output volume pot combination will be very helpful in that regard.

If you haven't seen it yet, take a look at Marjan Urekar's "Fuzzyballs" (http://members.tripod.com/urekarm/synth/fuzzyballs.html) for an example of just exactly how complicated you can take this thing.

Finally, I didn't see the Laval origin at first.  I'm a proud graduate of Chomedy Protestant High School (class of 1969), though I haven't lived there since Belmont Park was beside the Cartierville Bridge.

[puretube]

did you also add a 2nd base resistor for the 2nd transistor?

[strungout]

Oy.

Mark: Small world as they say  Let's see 1969...that would make me -10 years-old...

puretube: yes I did. Figured copying what was already there was the easiest way to start.

And wow. This circuit has alot of possibilities indeed!

So, I tried keeping only one EF, but adding low/highpass filters didn't accomplish what i needed (can you connect low/highpass filters the same way in inverting amps?), actually, when i put a highpass (using the EF's .05 uf input cap), it stopped wah-ing and all I could get was a wobbly-ripple-wah when I upped the sensitivity. Erh, again, interesting, but not it.

Then, 2 EF's. I first started with one controling each filter, but I'll get back to that. I tried two EF's driving one filter. The result is pretty cool, as-is you can set to two EF's for different sensitivty and it works, but there's a definate need for a pan pot between the two, as the EF set for the higher note dominates the other. I didn't experiment any more with it.

Back to one EF per filter. That seemed the best solution and is the one I like best for one reason; filter interactions. I used 6.8nf in one filter and 10nf in the other, and the effect was easy to hear; the sweep of one filter is delayed relative to the other, so you get what sound like two guitars playing at (almost) the same time. The effect is less noticable when the caps are all equal value.

So, with what I tried earlier and these new (to me) findings, I'm thinking about using switches to chose how many and which EF's or filters are on. Along with panpots, and attack/delay mods, you get yourself a nice, monstery, Franken-quack! And alot of knobs...

I haven't settled on what values I'll keep in the filters, but I'm thinking I'm gonna have at least three, each with a pair of caps at 6.8nf, 10 nf and .1uf or .22uf (that's a really deep bassy distorted wah. Lots of values/interactions possible. BUT...I need to save me up some money to build it. And experiment with the filters in series too... hmm, I need a break.

Oh yeah, I did make the second filter in the "bass" configuration, just sounded better. I draw up a schematic with some notes in a few for others to see, this is really a good way to experiment and learn a few things.

Thnx much for the pointers all

Ciao.

[Mark Hammer]

Given how easy the DQ and its variations are to build, you won't be surprised to hear that I have a few of them (I think I've made about 6 so far).  Plugging two in series produces some very nice overdriven filter sounds.  Moreover, cascading two bandpass filters results in a more resonant-sounding filtering.  Think Sherman Filter Bank and Chemical Brothers.

The one problem with it though, is that since the envelope signal in the 2nd one is driven by the audio output of the first one, the smoothness of the sweep leaves something to be desired.  A far better arrangement is to have two cascaded filter sections (both tuned to the same centre-frequency) driven by either a common envelope follower (as in the Baseballs), or by at least two envelope followers driven by the same audio signal.