Multiple FX combined for DJ FX device - Suggestions needed

[Spatz]

Hello everyone,

I am a (techno) DJ and not a guitar player, but I still need you help.

I want to make a device that offers all (or at least most) effects a proper techno DJ needs.
Namely, that is a delay, a reverb, an overdrive, a low-cut filter, a high-cut filter with resonance, and a gate, and everything in stereo.
For the delay/reverb, I want two use two or three PT2399 per channel, that are controlled by a microcontroller. This design is loosely based on the Electrosmash Time Manipulator, and the user can select different styles of delays with one rotary encoder, and tap in the tempo and adjust it with another rotary encoder. If you choose delays times that are short enough, you can get some kind of (cheap) reverb effect.

The overdrive should have only one knob, add some moderate to medium distortion to the signal, and the knob should go to 11, at least.

The two separate filters are used to cut out a certain part of the track played (like the snare) from the rest of the signal. For some sounding, the high-cut filter should have an adjustable resonance. Altogether, the filter section should have three knobs (freq-low, freq-high and resonance).

The gate should also have only one knob, and could be used to cut out the remaining unwanted signal.

Do you have suggestions for schematics I can use for the overdrive, filters, and the gate? They should have as few knobs as possible, run on +9V single rail, and have as few parts as are necessary for some decent sounding device. I do not want to make this too complicated, as this is only a hobby project.

Best Regards,

Bastian

[Fancy Lime]

Hi Spatz,

welcome to the Nuthouse!

For the overdrive I assume you'll want something that will produce usable results with a large variety of of input signals like (electronic) drums, vocals and even complete mixes, right? In my experience, relatively mild softclipping with a top boost works best (to my taste) for this sort of thing. I would strongly suggest a Tube Screamer Clipping Stage:
https://www.electrosmash.com/tube-screamer-analysis
Simple, flexible, low part count, easy to do in stereo with a single opamp. You might want to increase C3 to taste. Try leaving out one of the diodes. This will give you a less distorted sound that still has noticeable added harmonics but retains much of the dynamics of the original signal, which may be good for your particular use case.

For the Low- and High-Pass Filters, I tend to go with Sallen-Key topology. Low parts count, flexible, and one of the easiest to design. For a resonant filter you need at least a second order filter (like the Sallen-Key). Also, I think for the use you describe first order filters won't cut the mustard. If you find that second order is also not steep enoungh (you want sharper filtering), you can cascade two identical S-K's for (effectively) a forth order filter. The Wikipedia article is a good start:
https://en.wikipedia.org/wiki/Sallen%E2%80%93Key_topology
In designing the high-pass filter you probably want to shoot for Butterworth characteristic:
https://en.wikipedia.org/wiki/Butterworth_filter
A good filter design tool to pick the right component values can be found here:
http://sim.okawa-denshi.jp/en/Fkeisan.htm
You will need to vary both resistances of the filter in unison, meaning you need one stereo potentiometer per channel or one quad pot for both (and I wouldn't know where to get those.
For the low-pass you'll want Butterworth "or higher". The problem there is, that tuning cutoff frequency and resonance is not easily done independently in Sallen-Key filters. So ultimately, if you want something that is easiest to use, at the expense of being more difficult to develop, you might want to look at State Variable Filters of the Biquad variety:
https://en.wikipedia.org/wiki/State_variable_filter
Those allow independent control of resonance and frequency and can be used as high, low, or band pass (and with little extra effort also as band-stop). You'll still need one stereo pot per channel for the frequency, though. To get around that problem, you would need a voltage controlled filter, of which there are about 1.73 metric s#it-tons in all possible topologies. Have a look at Tim Escobedos 9V MS-20 Sallen-key LP VCF for a starting point on an electronic music classic (scroll down about 2/3 of the page):
http://www.diale.org/escobedo.html
Replace the LM13600 with a LM13700, though.

The gate is a bit more tricky. You can make that super simple with just a pair of anti-parallel diodes but to know if that will suit your purpose at all or what level of difficulty you will have to put up with, we need some more info on what exactly the gate is supposed to accomplish.

Cheers,
Andy

[Mark Hammer]

Andy has some good suggestions, but I'll ask one very important question: what sort of signal level do you anticipate sending to such FX?

Most existing guitar-pedal designs anticipate a guitar-level signal, which is noticeably greater than dynamic-mic level, but noticeably lower than line level.

[Spatz]

Hello Andy, hello Mark,

thank you for your answers!

First of all, my preferred DJ mixer, the Allen&Heath Xone:92 has an insert for external FX, and the send output is balanced with 47 ohm impedance and -2dBu max level, which is ca. 0.6V RMS or 1.7V peak-to-peak. As the return input is unbalanced, I'd like to have an input in my FX device that works both with balanced and unbalanced inputs, converts the balanced signal into unbalanced in the input stage, and gives out an unbalanced signal. I hope that I can use a simple opamp to convert the signal from bal to unbal.

Andy, you assumptions about the type of input signals are right, most of the time it will be complete songs that are already properly mixed. The clipping stage of the tube screamer seems like a very good choice for my needs. I will add one to my schematic, and add solder jumpers for the two diodes, so you can choose the amount of distortion you want.
Just one question: I plan to have the tube screamer always in the signal chain (no bypass), so the amount of distortion should be negligible when the knob is at its lowest position. Of course, a very slight amount of distortion is okay, as the device will only be an insert with variable send and return levels, and the other FX (delay etc.) will add some distortion anyways.
I guess I will have to make C3 quite a bit larger if I want to have distortion at lower frequencies, too, right? Four times the capacitance should lower the filter frequency to 180Hz, which seems more reasonable for techno music.

For the filter, a VCF seems like the best solution, as, as you already noted, quadruple potis are very hard to find, if they exist at all.
For the high-pass, Butterworth seems reasonable, I do not want any added volume in the bass area, as it tends to get muddy easily.
For the low-pass, I have looked at the classic Moog filter and the filter for the TB-303. Both seems to be transistor ladder filters. Are there other filter topologies for VCFs that are less complicated?

The gate would more or less a "nice-to-have", and could maybe be ditched if it does not work as intended. I found this schematic for a 1-knob-gate, but the opinions are quite mixed, as it seems to cut off rather abruptly: https://effectslayouts.blogspot.com/2016/07/one-knob-noise-gate.html

I gues it would make sense to change the sequence of the FX in the device to: Low-Cut --> High-Cut --> Gate --> Delay/Reverb --> Distortion, or what do you think?

Best regards, and thank you,

Bastian

[Mark Hammer]

It's not like I follow EDM religiously, but my understanding is that many commercial filter units aimed at DJs will have momentary switches that will trigger transient generators to feed the filter and dictate its corner/resonant frequency, so that one can generate "bursts" of filtering in time with the music itself by repeatedly tapping the momentary switch ("pyoo-pyoo-pyoo").  For instance, the Electrix Filter Queen had such a big momentary button (though I think it was simply for inserting or defeating the filter momentarily):

[Spatz]

Hehe, EDM and techno are two totally different things...  Even though for someone who does not follow it, there might not seem to be big differences. But I personally love techno and despise EDM.
Basically, EDM is about getting from one big "drop" to the next big "drop" in a timeframe that is as short as possible (kind of hyperactive music), while when mixing techno, you try to make it as seamless as possible, and maybe have a "drop" every 5 to 10 minutes. So the pyoo-pyoo-pyoo would not be needed here, this device (and the filters) would rather be used to select a certain part of the next track (like the hihats or snares), and slowly fade it into the current track, with just a bit of delay or overdrive for some variance.
So, two frequency potis and a resonance poti (for emphasis) should be plenty for the filter section, as the user should not really play around with this anyways... just set it so the desired instrument is singled out, and the leave it like that.

[Fancy Lime]

Hi Bastian,

for the effects order I would go: Low-Cut --> Distortion -->  Delay/Reverb --> Gate --> High-Cut.
Reasoning: High-cut last and Gate just before (or the other way round, does not matter much) it gives you the best noise performance. Usually (in the guitar world) we want the reverb or delay to affect the already distorted signal, not the other way round. But for your use case, the other way round may also sound  interesting. I suggest to try both or even make the order of those two switshable. Low cut before distortion makes the distorted signal less muddy, which is even more important with your kind of signal than with a guitar.

"Unbalancing" is easy. You can do it passively without an opamp by simply using one of the "hot" wires as signal and connecting the other to ground with a resistor (100k or so). But you could also build your effects processor balanced if you have a balanced signal to feed it anyway. "Theoretically" that would be better (at least from an engineering standpoint), although it is debatable if it would make any practical difference.

You can configure a tube screamer type distortion stage such that it will not distort at all in the lowest knob setting by simply leaving out (meaning replacing with a wire) R6 in the Electrosmash schematic.

Yeah, ladder filters are weird. Honestly, I have no idea how actually to design one after reading the original patents, the Stinchcombe papers and a lot of other stuff about them. The "best" way (by some meaning of the word) might be a State Variable Filter (SVF) with switched capacitors as the control element. Switched capacitors essentially act like resistors, with the switching frequency controlling the resistance. Look here:
https://en.wikipedia.org/wiki/Switched_capacitor
For audio use, you'll want a lower limit of the switching frequency of about 40kHz minimum, so you'll need a fairly fast oscillator to control the switches. CMOS logic is probably the way to go here, I would try something like the CD4069 based oscillator in the MXR Envelope Filter (but careful the MXR looks a little like a switched capacitor but is in fact something slightly stranger). For the switches, CD4066 seems a good choice. You could also straight up copy the MXR design, but that is not precisely easy to tweak to taste unless you are comfortable thinking in time domain and frequency domain simultaneously. An easier thing to design but more difficult to get consistent parts for, would be the SVF from the Mutron III envelope filter. One thing you should definitely read before moving on is Marks article on envelope controlled filters:
http://www.geofex.com/article_folders/ecftech/ecftech.htm
Of course, you'll only want the filter section and can replace the envelope detector with a simple pot to control the cutoff frequency.

So much for the moment,
Andy

[ElectricDruid]

Lots of interesting ideas on here. I'd like to chuck a few more in too.

I agree that state variable filters (SVF) would be a good idea, mostly because it's easy to have separate resonance and filter cutoff controls that don't interact. Each filter needs two variable elements to control it. That can be as simple as a dual-ganged pot (if indeed those are simple). Personally, I wouldn't go for a switched capacitor filter - too much circuitry required to build the clock, and who wants to chop their audio up at 40KHz anyway? An alternative solution would be a VCA chip like the V2164. This has four VCAs which can be used to make two separate SVFs - ideal for your low-cut and then high-cut filters. The typical SVF design that we're talking about has a -12dB/Oct slope. The 2164 has another advantages in that it has an exponential control response, so if you increase the control voltage linearly the frequency goes up in octaves - very musical.

I also would suggest using a differential pair for the distortion. This configuration isn't used much for guitar pedals (sometimes, but not much) because it's a very soft approach to clipping. But this makes it ideal for full range signals. The circuit I've been using for my experiments is the final part of this "sine shaper" (if you feed a triangle wave into it, it rounds it off and makes something reasonably close to a sine). The only change I made was to stick a op-amp gain stage in front of it to allow me to vary the drive.

https://electricdruid.net/wp-content/uploads/2017/06/SineSubOsc.jpg

HTH,
Tom

[Spatz]

Hi Andy,

if think I will add wire connections between all individual devices, so you can change the sequence afterwards.

If I think more about it, I think the "unbalancing" will already happen in the cable/adapter, as I plan to use RCA inputs.

For the tube screamer, I assumed leaving out the resistor gets me where I want, but now I am sure, thank you!

For the filter, I found Tim Escobedo's variant of the Korg MS-20 VCF. There's a 9V version, it uses few parts, has frequency and resonance control, and you can easily adapt it for HP and LP. If I want, I could set the resonance for the low cut to a reasonable amount later on, or leave it as a poti, for more control.

Tom, I will read further into your suggestions tomorrow...

Cheers,

Bastian

[Spatz]

I am currently working on the schematics in KiCAD, and I have some questions about the virtual ground:

In the schematic for the delay and the schematic for the tube screamer, the virtual ground is done by two resistors (and a cap), while in the schematic for the filter, the virtual ground is done by one resistor and two red LEDs. I assume that the idea behind the two LEDs is that the voltage drop over one red LED is 2.2V, so the drop over two LEDs is more or less 4.5V. But is there another reason to use LEDs instead of resistors?

And another thing: Should I create a virtual ground for every sub-group, or is one virtual ground for all circuits enough?