Whats the deal with Ring Modulators?

[Astronaurt]

Hey Y'all!

So I've been doing a some research into Ring Modulator designs as a friend recently asked me about building one to sell him. I've never delved deeply into these circuits before, so thus far it's simultaneously refreshing and perplexing to see an effect where there are so many different lay-outs, topologies, and designs to achieve the same goal! I'm diggin' it.

That being said, I'm curious as to why that is- I'm seeing Voltage controlled Amplifiers, Phase locked loops, Quadrant Multipliers, 555 LFOs, Op-amp based Oscillators; the number of different paths is pretty expansive. Except what ever happened to the original Transformer coupled Diode bridge design?

https://upload.wikimedia.org/wikipedia/commons/c/cd/Ring_Modulator.PNG

First off, is there a particular reason or limitation to explain why it's not typically used in Guitar effects? Excessive current draw for driving X-frmrs? Frequency limitations? Bulky and takes up too much space? etc?

That being said, has anyone else here built or modified a Ring Mod pedal? What are your favorite or preferred circuits to implement the effect, and why?

Ideally I'd like to come up with a design that's relatively low part count without sacrificing functionality and flexibility. One can dream right?

[R.G.]

First off, is there a particular reason or limitation to explain why it's not typically used in Guitar effects? Excessive current draw for driving X-frmrs? Frequency limitations? Bulky and takes up too much space? etc?

All of the above, plus another biggie - guitar level signals are typically about 1/10 to 1/3 of the forward voltage drop of a diode. Guitar level signals won't work or won't work well at turning on those diodes as the signal changes. The changes in diode conduction is what the two transformer ring modulator relies on. It's hard to get lower-drop diodes, although germanium and Schottky get close. So you need amplifiers to raise the signal levels and such.

And then there's the expensive, big, heavy, poor-frequency-reasponse transformers after you've gone and mess with the amplifers. Better to do it with amplifiers some way if you can and ditch the iron.

Ideally I'd like to come up with a design that's relatively low part count without sacrificing functionality and flexibility. One can dream right?

Yep. We all would. Seems like a fundamental law of nature is: good, fast, cheap - pick any TWO.

[garcho]

^ same law applies to music projects. to be worth your time and effort (and money), and to have any hope of being sustainable, they must be 2 of the 3:

1 - artistically rewarding
2 - financially rewarding
3 - fun

have you tried one of the LMC567 faux-ring modulators? it can achieve a lot of what a "real" ring mod can do with a very low parts count

[blackieNYC]

I have the thing modulator and I love it, but I passionately recommend the Corruptor
here's the original thread, you can find the schematic at Freppo's site Parasit
http://www.diystompboxes.com/smfforum/index.php?topic=109237.0  BUT I added a blend knob (a panner between cos fuzz and the ring mod), using one more gate as a mixer.  It becomes very musical.  I used to let complaints of "carrier bleed" keep me from building ring mods.  When you've decided to kick in a ring mod, a little extra noise is usually not of concern IMO.  This carrier is heard between played notes only.  You can address this two ways - tune the carrier to key, or - don't stop playing.
Also,  modulated delays and chorus/flangers/phasers: If you bring your LFO oscillator up into the audio range (I would say at least 40 Hz or maybe up to 200 Hz) you get wonderful ring mod type sounds.  A switch can substitute the cap which determines frequency range in the standard 2 op amp design - if you want both very slow speeds and audio range speeds.  Jenny Greenteeth for one.

[Mark Hammer]

+1 to blackie/alan's comments.

But beyond that, one has to always remember that ring mods have their roots in synthesis, which means that:

• people would feed them *pure* waveforms, like sinewaves, such that a person could actually hear the sideband products as actual notes, rather than the harmonically-complex thing that is a guitar signal;
• the signal fed to both the carrier and modulator of a synth ring-mod is far more controlled, with respect to dynamics, such that one can arrange for the duration and/or intensity of the sideband products

As a result, although ring mod CAN sound musically useful with a guitar, there are a number of things one has to plan around in order to get it to sound optimal.  Certainly creative lowpass filtering of the input signal is helpful to get it to more closely approximate a sine wave (or at least produce fewer sideband products of harmonics).  And control over the depth of the modulation and maintenance of signal level (by compressor) is helpful.

[garcho]

all it takes is a simple noise gate and ring mod bleed whine is gone, when it needs to be. envelope followers for ring mods don't have to be as sophisticated as they are with VCFs, or comp/expanders, as long as it doesn't ripple too much as the plucked string dies out (1-10Hz), everything is fine. no one uses ring mods for their subtlety. with a clean blend option, you don't notice the decay ripple much anyway. i'm building one right now (LMC567 style) that uses a simple op amp envelope follower, along with a SWTC (thanks Mark!) and has practically no bleed through, and can tame the harshness to a musical degree. it doesn't sound exactly like a classic ring mod, but some of its flaws are fringe benefits.

Moog makes a decent and surprisingly cheap (for Moog, anyway) ring mod guitar pedal

[PRR]

> no one uses ring mods for their subtlety.

+1

The classic audio ring modulator is the scrambler, used to talk secrets on public telephones. (Watch Dr Strangelove.) The scrambled signal is an AWFUL RACKET. However your co-conspirators can de-scramble it to near-normal by injecting the same carrier the other way.

Yes, analog computers (synthesizers) "can" do nice things with ring-mod. Never got anything nice myself (I used to oversee the beast), but it has been done. Getting "music" with complex guitar tone and a ring-mod seems to me like endless frustration, but maybe it depends on your music. And also very-much what faults your particular ring-mod has.

The transformer-diode bridge is a classic in narrow-band RF work but as R.G. says the diode drops are brutal even if you boost guitar to many-volt.

The LM13700 may be as good a ring-mod as anything near its price.

[anotherjim]

I've always liked the name and the idea more than the sound. I can certainly understand the fascination with them. I've always been under a similar affliction with phase locked loops.

As already said, there are ways to add something like ring mod to more generally useful FX.

[ElectricDruid]

Hi Astronaurt,

If you're using a ring mod with an oscillator stuck into one input, then you can use whatever you like for the oscillator. A lot of the options you mention cover that range: LFO, 555, PLL, op-amp osc, etc etc. Something sine or triangle-ish is probably a better bet to keep the harshness under control, but you could go crazy and feed square or ramp waves through it if you're into that kind of thing. KRRROOINNGGG!!!
As far as the actual ring mod part goes, you can build one with VCAs, OTAs like the CA3080 or LM13700, or fancy multiplier chips like the AD633. The only advantage of the passive diode ring mod is that it's passive, and that's not an advantage with guitar since you'd need active circuitry to boost the signal to a sufficient level to drive it anyway. So it's a bit of a waste of time, since other designs have a lot less distortion. I've built a few OTA ring mods, and the actual ring mod part isn't too complicated. These days I'd probably try and use a PIC for the oscillator, but that's just me trying to do everything bar make toast with a PIC.

Tom

[ashcat_lt]

Also,  modulated delays and chorus/flangers/phasers: If you bring your LFO oscillator up into the audio range (I would say at least 40 Hz or maybe up to 200 Hz) you get wonderful ring mod type sounds.

A tremolo running at audio rate is pretty much exactly a ring modulator. 

I personally have never been particularly interested in the typical static-oscillator as carrier.  It can be interesting here and there, but the real fun comes when you start multiplying complex, changing, interesting things by each other.  I like to ring mod my guitar against my bassist quite a lot.  When I'm playing chords, it comes out as kind of just a distorted mess of a mix - like we're playing through a broken radio that's drifting off the station or something.  When I'm playing single-note leads, it's more like a crazy sounding synth playing off-kilter harmonies.  Just a couple of examples, but I think a ring modulator without a way to override the carrier with an external signal is severely crippled.

Some of the other above points are almost more important in this situation, though.

Because we're multiplying, dynamics can become exaggerated so that if both are quiet, it gets really quiet, and if both are loud, it gets REALLY LOUD.  It really benefits from dynamic control on both sides.  Cleaner compression is probably better going in, but I usually like to distort the heck out of the output, to get a bit more square wave synthy sound.

We call it carrier bleed, but at a certain point it's hard to say which is which.  Really, they both affect the output in the same way, and if one is silent, there should be no output.  But passive guitars never actually get completely silent.  Hiss, noise, hum, etc from one will "keep it open", and especially when you're smashing the signal down like I do afterwards, you end up hearing the "not silent" source pretty clearly.  A noise gate is not a bad idea.  Course, the series diodes do kind of exactly that, if you can get it to where the noise is below the Vf.  Which leads to the idea that you can bias one or the other end of each diode to move the threshold around.  That's how the detector in most comps and gates work, right?

Filtering does different things on either side, also.  I personally am often deliberately shooting for all the crazy sideband noise that you get from running full-bandwidth signals against each other, but as one of the sources gets more complex (and starts to approach broadband noise), the result starts to sound more and more like the other (same as with "noise carrier bleed" above).  So, yeah filtering going in can clean some of that up. 

But, you know, if we're using a simple waveform for one of the inputs (call it carrier, why not?), then we're really talking about aliasing.  Using appropriate filters, you can kind of decide whether you want the sum or difference signals.  High-pass above the "carrier" frequency and you get (mostly) only the sum byproducts.  Low-pass below it and you get difference.  It can also help supress that carrier whine we were talking about.   I did a thing once where I ran a whole mix through a couple of ring modulators.  The "carrier" was somewhere in the middle of the audio band, and the same for both.  In one I low-passed going in and high-passed coming back out and the other was vice versa.  The result, mixed back together, was pretty crazy.  The harmonic content was mirrored around the carrier frequency such that treble became bass and bass became treble.   

IDK if any of this helps anybody here.  I do have the Moogerfooger ring mod, which is kind of awesome, but most of my experience with these things is in software.

[Keppy]

Also,  modulated delays and chorus/flangers/phasers: If you bring your LFO oscillator up into the audio range (I would say at least 40 Hz or maybe up to 200 Hz) you get wonderful ring mod type sounds.

A tremolo running at audio rate is pretty much exactly a ring modulator. 

Note that a true ring mod can remove the input frequencies completely, leaving only those frequencies generated by the multiplier. Chorus/tremolo mods don't do this. That doesn't mean they won't get you what you want, but they won't get everyone what they want.

Part of the reason ring mods have so many types of circuits is that nulling out the input and carrier signals is difficult, and not everyone wants to completely null the input signal anyway.

Another reason is that the multiplier effect multiplies not just fundamental frequencies but also harmonics. Sine wave carriers give you a more mellow, controllable sound and less noticeable bleed but are hard to generate. Triangle and square waves are easy to generate but get chaotic very quickly.

Plus, these aren't circuits associated with classic music, comparatively speaking. It's not like trying to nail a classic fuzz sound. Most of us folks using and designing these are drawn to outside the box, weird stuff, so it makes sense there's not really a standard model for these circuits.

[ElectricDruid]

Also,  modulated delays and chorus/flangers/phasers: If you bring your LFO oscillator up into the audio range (I would say at least 40 Hz or maybe up to 200 Hz) you get wonderful ring mod type sounds.

A tremolo running at audio rate is pretty much exactly a ring modulator. 

Note that a true ring mod can remove the input frequencies completely, leaving only those frequencies generated by the multiplier. Chorus/tremolo mods don't do this. That doesn't mean they won't get you what you want, but they won't get everyone what they want.

Part of the reason ring mods have so many types of circuits is that nulling out the input and carrier signals is difficult, and not everyone wants to completely null the input signal anyway.

The technical difference between those two cases is that a tremolo is an amplitude modulator, a two-quadrant modulator, whereas a ring mod is a four-quadrant modulator. The two-quadrant modulator produces sum and difference frequencies but also produces the original frequencies. The four-quadrant modulator only produces the sum and difference frequencies.
"Two quadrant" means that one input is bipolar +/-, and the other is unipolar +ve only. That's your tremolo - The audio input is the bipolar one. The control input goes from zero through quiet to loud, but the volume is only 0 to positive. Audio-rate modulated phasers/flangers/chorus/filters give similar effects.
"Four quadrant" means that both inputs are bipolar, which means that a ring mod can not only control the amplitude of the input, it can also invert it. So volume now goes from -loud, to -quiet, to zero, to +quiet, to +loud. Both inputs are bipolar, and both could be the carrier or the modulator.

In practice, they both sound pretty "clangy", but ring mods are even more clangy!

HTH,
Tom

[PRR]

> tremolo running at audio rate is pretty much exactly a ring modulator.

Except we usually run trem with a "DC offset", gain does not pass-through Zero to inverting.

As the Druid says, you can count quadrants.

[ashcat_lt]

> tremolo running at audio rate is pretty much exactly a ring modulator.

Except we usually run trem with a "DC offset", gain does not pass-through Zero to inverting.

As the Druid says, you can count quadrants.

Yep.  Good point from both of y'all.  That is a real difference.

[Astronaurt]

have you tried one of the LMC567 faux-ring modulators? it can achieve a lot of what a "real" ring mod can do with a very low parts count

Awesome! No I haven't, but I actually just dropped by the local Electronics surplus store yesterday and found a couple of those chips. Should be able to breadboard one in no time on the next lazy afternoon. Thanks for the head's up!

I'd love to mess with the full 4-quadrant multiplier idea a little bit down the road, at least because it seems the most true-to-intention and execution Ring mod Route. However, it seems the cost of an AD633 is almost prohibitive if I'm just tinkering... at about ~$10US from Mouser. Any one know of a non-stupid-overpriced source for a 4-quadrant multiplier chip? 

That being said, right now I'm really digging the concept of using the LM13700. With 2 separate OTAs in the same chip, I can totally see using one as a VCO to trigger the other one, arranged as a VCA. Additionally, using one half as a VCO also opens the door for different methods of playing with the Carrier Freq controls- perhaps modulating the CV on the oscillator itself or also using totally different control/carrier inputs.

I've never gotten much into analog sequencing but Ring Mod seems like the perfect place to use it. Anyone checked out the Zvex Ring Tone or Super Ring Tone?

[PRR]

$10 is not "prohibitive". What did you pay for your guitar? Amp? Performance shirt? A good set of strings?

Back when gasoline was $0.25 I paid $3 for a '741 opamp; like 20 bucks today.

I just spent $22 for earthquake straps, and this isn't really earthquake country.

> Any one know of a non-stupid-overpriced source for a 4-quadrant multiplier chip?

Anyone know of a mass-market application for a 4-quadrant multiplier? AFAIK, only specialized and mostly not-cost-sensitive machinery uses 4Q Mults. (Think MRI machines, or Higgs Bison colliders.)

eBay has alleged AD633 for $3/ea or 5 for $13, free shipping. Newark has 91 at $8/each.

[Astronaurt]

$10 is not "prohibitive". What did you pay for your guitar? Amp? Performance shirt? A good set of strings?

Ha! Touché, though I'm not saying it's the most sought after of un-obtainium and commands the riches of a sultan or anything. It's just more than I'd like to pay for a single IC, + at least one extra in event of accidental damage + shipping, when I'm not entirely sure I can sell the end product or even want to use it personally yet.

Really, the only thing prohibitive is my attitude. 

I'll check out eBay though, thanks for the tip!

[Keppy]

Any one know of a non-stupid-overpriced source for a 4-quadrant multiplier chip? 

That being said, right now I'm really digging the concept of using the LM13700.

Theres a 4QM schematic in the LM13700 datasheet. No extra-special chip required.