Thoughts about Tri Chorus, Dimension D/C and Stereo Chorus Setup?

[Mr. Lime]

Hi guys,

it's been a while since my last activity here.

Recently I've been wondering about those great sounding analog studio chorus effects like the Roland Dimension D or Tri Chorus rack gear.
These effects are basically two or three chorus circuits in parallel. The Dimension D (and C pedal) seem to use a single LFO which inverts it's phase 180 degree so when the left side ramps up, the right side goes down.
The Tri Chorus has 3 parallel circuits and also 3 LFOs which seem to offer controls for depth of each. I guess the LFOs have different oscillation frequencies in the "preset mode" which causes a unpredictable liquid sound. In manual mode only one LFO is active and there seem to be a 120 degree phase shift between the 3 chorus circuits, so it's more similar to a Dimension D unit.

As this stuff is quite expensive, I wonder how different a simple stereo rig with two chorus pedals with slightly different settings would sound like compared to that rack units?
The Dimension Ce-2w costs roughly as much as 2 classic analog pedals which offer way more control - so what's the point of it if part of the Tri Chorus Stereo magic lays in the different LFO settings to get that spatial sound?


Is there anything to consider in regards of phase cancellation?
I would guess having two parallel chorus effects with random slope waveshapes may result in something similar like the Tri Chorus in "preset mode" as there should be some cancellation on each L/R channel, which contributes to a liquid sound, isn't it?

Looking forward to your thoughts!

[Mark Hammer]

I have not been able to compare them directly, but there are two major categories of multiple voice choruses.  One is the use of counterswept delays, synced to a common LFO.    The other is multiple delays with independent LFOs.  The key feature of the first category is that one of the delays is a bit sharp, and the other is a bit flat, reducing the perceived pitch wobble found in simpler single-delay choruses. When the LFOs are unsynced, there is the remote possibility that at some point they might all be a little sharp or a little flat.  But it's remote.

The Juno chorus also uses two counterswept delays, and the ensemble switches affect the sweep of both delays equally.  It uses a pair of MN3009 bbd chips with 1/4 the number of stages of the MN3207s used in the DC-2, and copies (e.g., the Behringer CC300).  In theory, the 3009 could be clocked to achieve  the same delay time as the 3007/3207, but my guess is that Roland used the potentially shorter delay time of the 3009 because of the much wider bandwidth of a polyphonic synth, making the comb filtering of content well above 6khz more relevant.

[Ripthorn]

The LFO synchronization and the use of a single dry signal and recombination stage are the biggest reasons for the dedicated effects. If you are feeling DIY, I have released projects for the community for both dimension and tri-chorus. The tri-chorus is a PT2399-based chorus, so it's got the inherent delay. The dimension chorus uses the ES56028 (same as in the modern Boss CE-5) and has controls to make it much more flexible than the standard 4-preset dimension chorus. You can find them here: https://scientificguitarist.wixsite.com/home/projects. The Alternate Dimension is the dimension pedal while the Diamond Church Street Chorus is the tri-chorus.

[Mr. Lime]

I just purchased a Boss DC-2w to hear how I like a single LFO countersweeping.
For sure I like it way better than the other analog chorus effects I had before. In the FX loop I don't find it subtle at all but lovely sounding for cleans.

Curious how I would like a Tri Stereo Chorus and I would love to have a look at the schematic of one.

@Ripthorn, cool projects you got there!
Love the idea of the ATTiny841 LFO but I prefer BDD chips over PT2399.

I was thinking about putting two Electric Druid Flangelicious effects together, remove the feedback loops and alter some components to chorus operations. With Tom's Multiflange delay clock + LFO chips, parts count would be kept low.
Maybe adding a daughterboard with a Stomplfo to take use of CV of the MULTIFLANGE chips manual mode to synchronize the LFOs but an external CV might do the trick as well. With the polarity switch counteraweeping would be possible.


How the Tri Chorus LFOs work describes a german website:

The LFO mode is really weird: If »Preset« is active, the modulation is done by two sine LFOs, each with three outputs that are phase-shifted by 120°. One of the LFOs oscillates at approx. 5.5 Hz, the other at approx. 0.25 Hz. Both are superimposed/added in a ratio of approx. 1:6, so that one only achieves 15% of the modulation depth of the other. Both run completely independently of each other and are not mutually modulated. Both the Intensity and the Rate controls are then disabled. The result is an extremely subtle and unobtrusive, yet lively, floating chorus effect. You miss it as soon as you turn it off. This is probably the typical TSC sound.

In »Manual« mode only one of the LFOs works. Here the modulation depth (»Intensity«) for all three delay lines can be determined separately and the speed can be determined with the rate control (together). Depending on the settings of the controls in relation to each other, very subtle effects are also obtained here, but much more striking sounds are also possible. The most intense chorus effect is achieved at Intensity settings of around 1 o'clock. After that, the device delivers a clear vibrato or Leslie-like effect, especially at high rate settings. This operating mode also delivers very pleasant, lively sounds with a touch of metallic shine, which is particularly beneficial for guitars and electric pianos. However, the sound is much more static than in preset mode. Incidentally, the speed of the (manual) LFOs is controlled via vactrols. This is probably the reason for the somewhat sluggish behavior of the rate control and the modulation that is perceived as particularly »soft«.

If you press both buttons at the same time, the three modulation signals are added. This makes the effect slightly more static, but a little more intense. If you switch off both buttons, the modulation is disabled. The audio signals then pass through the bucket brigade with a fixed delay time. Since the delays are not 1

Not sure if a modded stereo Flangelicious would sound anything close to that double LFOs on three delay lines..

[Mark Hammer]

The LFO synchronization and the use of a single dry signal and recombination stage are the biggest reasons for the dedicated effects. If you are feeling DIY, I have released projects for the community for both dimension and tri-chorus. The tri-chorus is a PT2399-based chorus, so it's got the inherent delay. The dimension chorus uses the ES56028 (same as in the modern Boss CE-5) and has controls to make it much more flexible than the standard 4-preset dimension chorus. You can find them here: https://scientificguitarist.wixsite.com/home/projects. The Alternate Dimension is the dimension pedal while the Diamond Church Street Chorus is the tri-chorus.

Chorus tha employs a little more delay time than usual can sound pretty good.  Though it is presently in disrepair, I have one of those old blue MXR Digital Delay rackmount units.  It allows the user to apply modulation to any of the delay ranges  Modulating in the 20msec and 40msec ranges (which can sweep to half and double the stated delay) gets you that Pat Metheny kind of chorus tone which, with the right kind of tone shaping, is rather pleasant.

[Mr. Lime]

The LFO synchronization and the use of a single dry signal and recombination stage are the biggest reasons for the dedicated effects. If you are feeling DIY, I have released projects for the community for both dimension and tri-chorus. The tri-chorus is a PT2399-based chorus, so it's got the inherent delay. The dimension chorus uses the ES56028 (same as in the modern Boss CE-5) and has controls to make it much more flexible than the standard 4-preset dimension chorus. You can find them here: https://scientificguitarist.wixsite.com/home/projects. The Alternate Dimension is the dimension pedal while the Diamond Church Street Chorus is the tri-chorus.

As the Tri Stereo Chorus seems to have two LFOs with different rate/depth settings, someone would have to use two ATTiny841 LFO to emulate the modulation of the rack unit.
The delay circuits of your Diamond Church Street Chorus could be substituted with analog BBD and clock chips.

I'm kind of interested to try the Arion SCH-1 chorus circuit too and think about building 3 of them and replacing their LFOs with two ATTiny841 chips.

[Ripthorn]

The LFO synchronization and the use of a single dry signal and recombination stage are the biggest reasons for the dedicated effects. If you are feeling DIY, I have released projects for the community for both dimension and tri-chorus. The tri-chorus is a PT2399-based chorus, so it's got the inherent delay. The dimension chorus uses the ES56028 (same as in the modern Boss CE-5) and has controls to make it much more flexible than the standard 4-preset dimension chorus. You can find them here: https://scientificguitarist.wixsite.com/home/projects. The Alternate Dimension is the dimension pedal while the Diamond Church Street Chorus is the tri-chorus.

As the Tri Stereo Chorus seems to have two LFOs with different rate/depth settings, someone would have to use two ATTiny841 LFO to emulate the modulation of the rack unit.
The delay circuits of your Diamond Church Street Chorus could be substituted with analog BBD and clock chips.

I'm kind of interested to try the Arion SCH-1 chorus circuit too and think about building 3 of them and replacing their LFOs with two ATTiny841 chips.

The ATTiny841 has 4 PWM output channels, meaning you could send out 4 different LFO signals. There are other chips that have more PWM out that would be more suited to a circuit needing more, rather than running two 841's.

[Mr. Lime]

The TSC would need a microcontroller with 6 PWM outputs for a correct emulation then.

I only have an Arduino Uno board so far and have never programmed with it. Unfortunately, I'm also quite limited in time at the moment, but I find the topic super exciting.

Which suitable microcontroller with sufficient PWM outputs did you think of instead of using two 841?

The wiring would also have to be changed, since depth would have to be separately adjustable for all 3 delay lines. Harald Sabro has a corresponding depth control in his CE-2 version.
What is the advantage of your variant, if I may ask?




Another consideration is to reduce the circuit by one delay line and add a second LFO to a Dimension C-like circuit. It is questionable how much this would differ from the tri chorus in terms of sound.

[Ripthorn]

I didn't think the stereo tri chorus was separate tri chorus on each channel. I thought you had three total delay lines and they were split stereo. Is this not correct?

[ElectricDruid]

I didn't think the stereo tri chorus was separate tri chorus on each channel. I thought you had three total delay lines and they were split stereo. Is this not correct?

That's what I've seen. The classic stereo tri chorus circuits come initially from the string synths of the 70s, with the Arp Solina probably the most famous. The sound was so popular that later single-oscillator polysynths like the Roland Juno106 and Korg Polysix used similar circuits to help fatten things up (check the schematics of those synths for details!). The general plan in these circuits is two LFOs with triphase outputs (so 0°, 120°, and 240°) with one LFO set for deep and slow, and the other for shallower and faster. Rate and depth are not variable, typically. Each of the three delay lines gets modulated by a mix of two of the LFO phases, one fast and one slow. Then you've got three complicated wet signals and one dry signal to mix back together however you like to get some crazy-lush stereo image. Maybe put a Dry and a 120° wet in the centre, with  0° and 240° left and right? There are loads of possibilities.

There were other options too, like Roland's "harmonic chorus" which used three LFOs at x0.5, x1, and x2 rates, iirc, but the typical tri chorus is as described above.

I've experimented with various possibilities using a microprocessor to generate the modulation signals for analog clocks and delay lines, since it's easy to make two LFOs in firmware and generate different phases and add them together. Since you've got three delay clocks, you only need three PWM channels, not six - do the adding in the digital domain, not afterwards in analog. In truth, I was somewhat underwhelmed. There's a definite point of diminishing returns, and it soon all becomes very lush and your ears/brain can't really tell one type of lush chorus from another type of lush chorus. I'd hoped to design a chip that could cover all the classic circuits and make them more practical to build (and I have a prototype of such a thing) but it's not actually that fantastic, which I hoped it would be. The different LFO options are not sufficiently different in sound. You're better off choosing one option and going with it, in my view.

HTH

[Mr. Lime]

Hi Tom!

Great info, thanks a lot for sharing. 
I understood the dual LFO approach but wasn't aware of the different waveshapes the TSC is using! I thought it's only dealing with sinewaves but the Korg Polysix takes use of all 3 outputs of the CMOS 4069 LFO as it seems.




I see your point in using 3 PWM outputs instead of six - I was thinking practical with minimal coding skills which i am blessed with.. 
That's something I really have to work on since there are other projects like a Lovepedal Ring Stinger that would benefit from a microcontroller a lot (see HEXE THE CARRIER for example).

I've experimented with various possibilities using a microprocessor to generate the modulation signals for analog clocks and delay lines, since it's easy to make two LFOs in firmware and generate different phases and add them together. Since you've got three delay clocks, you only need three PWM channels, not six - do the adding in the digital domain, not afterwards in analog. In truth, I was somewhat underwhelmed. There's a definite point of diminishing returns, and it soon all becomes very lush and your ears/brain can't really tell one type of lush chorus from another type of lush chorus. I'd hoped to design a chip that could cover all the classic circuits and make them more practical to build (and I have a prototype of such a thing) but it's not actually that fantastic, which I hoped it would be. The different LFO options are not sufficiently different in sound. You're better off choosing one option and going with it, in my view.

That's my train of thought too.
I can imagine that the difference between 2 and 3 delay lines in the band mix would hardly be distinguishable if oscillating with 2 different LFOs anyway.

However, I can imagine that many people would be interested in a simplified (but still advanced) TSC DIY version. I would have gone for the Threeway Stereo Chorus pedal myself, which takes a similar approach. The pedal is unfortunately sold out and alternatives are very rare.
https://www.rtelectronix.com/pThreewaychorus.htm
Maybe a tone control like the SCH-1 is more beneficial (especially in stereo setup) in terms of sound compared to different LFO options?

[ElectricDruid]

I understood the dual LFO approach but wasn't aware of the different waveshapes the TSC is using! I thought it's only dealing with sinewaves but the Korg Polysix takes use of all 3 outputs of the CMOS 4069 LFO as it seems.

May be I'm not understanding what you mean, but that's not the Polysix LFO. It uses two triphase oscillators, and the outputs are probably pretty triangular.

[PRR]

two triphase oscillators, and the outputs are probably pretty triangular.

Is it? I see three integrators, cascaded. That needs a gain of 27 to oscillate, here the LF gain is 3.3^3 or 35 which is margin for self-starting. A hard-clip amplifier might go 10% THD when stablized but the CMOS is softer-clip.

[ElectricDruid]

two triphase oscillators, and the outputs are probably pretty triangular.

Is it? I see three integrators, cascaded. That needs a gain of 27 to oscillate, here the LF gain is 3.3^3 or 35 which is margin for self-starting. A hard-clip amplifier might go 10% THD when stablized but the CMOS is softer-clip.

Ok, so you're saying the waveforms will be more rounded? Yeah, I can go with that. I should have 'scoped it when I had a working Polysix. In theory, a integrator is going to make nice triangle waves up and down, but if you push it hard, then yes, it's probably going to flatten the peaks and get more sine-like. It's a classic way to generate a sine in analog circuits, in fact - soft-clip a triangle. Ok, it's not hi-fi, but this is an LFO not a piece of test equipment.

[PRR]

> Ok, so you're saying the waveforms

Sine-Wave Oscillator -- Ron Mancini and Richard Palmer
https://www.ti.com/lit/an/sloa060/sloa060.pdf
section 8.2 onward describes various numbers of various integrator-type elements making "Sine" waves. The 4-i "Bubba" gives 0.1% sine distortion at its better node. Not triangle at all but filtering.

[ElectricDruid]

There's something going on here that I don't understand well enough. I can see the basic way that those circuits work - they're all "phase shift" type oscillators, where the signal gets shifted a bit and a bit more and then fed back in and amplified to create the oscillation.
What I don't get is the difference between the integrators and proper lowpass filters. I mean, I know that there *is* a difference, and I've seen the equations for it and etcetc. But somehow it doesn't *mean* anything to me - I don't *understand* it in a deeper way.

I suppose the "intuitive" reason for why the polysix LFO is a sinew wave oscillator not a triangle wave oscillator is because there's only a *single* frequency where the phase-shifts add up to reinforce the initial signal. A single-frequency output will clearly always be a sine wave (ignoring any distortion for a moment).

Anyway, thanks for the correction, Paul. I wasn't looking closely enough, but thinking about it more, I see that it *is* sine waves.

[ElectricDruid]

Oh, one other thing:

That TI article says in the part about the Bubba oscillator:

When low-distortion sine waves are required at all outputs, the gain should be distributed among all the op amps.

...which is exactly what's going on in the Polysix LFO design. Having a fixed LFO frequency makes a lot of things simpler!

[Mr. Lime]

May be I'm not understanding what you mean, but that's not the Polysix LFO. It uses two triphase oscillators, and the outputs are probably pretty triangular.

Sorry, it was late in the night when I had a short look at it.

Some more interesting multi chorus circuits:

Boss CE-300


ADA STD-1


To what extent does the use of a microcontroller for the clock create disadvantages or limitations?
I'm actually very satisfied with the Multiflanger and have often thought about a second build with Chorus component values. So why not put multiple Multiflanger in parallel and sync them with a triphase microcontroller?

[ElectricDruid]

To what extent does the use of a microcontroller for the clock create disadvantages or limitations?

There is additional noise caused by the frequency jitter of the NCO. That's inevitable to some extent. Depending on the NCO design, it could range from "awful" to "negligible", but the effect will always exist (and need to be considered) in any design that works this way.

In the Flangelicious (either 4 knob or Multiflange) it's a considered choice - it provides a massive gain in simplicity, and the noise hit isn't that bad compared to a lot of vintage flangers that aren't very quiet either.

I'm actually very satisfied with the Multiflanger and have often thought about a second build with Chorus component values. So why not put multiple Multiflanger in parallel and sync them with a triphase microcontroller?

Yes, that'd be possible. I'd thought to do it "the other way around" by simplifying the triphase LFO into a single processor, and then using three of the simplest analog clocks I could find (either MN3102s or 4047s, probably). That way avoids the limitations of the NCO-as-clock solution, but the *real* reason is because Microchip don't make a chip with three NCOs on it, so I couldn't wrap all three clocks up onto a single chip even if I wanted to! And you know I'd have *had* to have a go if such a thing existed.
Doing it as a "multiprocessor solution" like you suggest is another entirely valid way to approach it.

[Mark Hammer]

I will once again describe an interesting setup I have with a pair of Boss BF-1 flangers.  I lifted the resistor on each that mixes the dry signal in with the wet, such that they each only provide a wet signal.  I made a splitter/mixer unit, so that I could feed and run them in parallel and mix their outputs together.  Setting one to a short fixed delay, and modulating the other, gets me thru-zero flanging with a variable zero point.  Modulating BOTH, however, in unsynced fashion, moves the zero point around randomly.

I also built myself a stereo version of the Magnavibe -a gentle one-stage vibrato modelled on the Magnatone amp.  A dual-ganged pot adjusts the speed of each LFO, and another adjusts their intensity, but they are completely independent circuits.  A very rich and satisfying effect.

While one of the goals of this thread is the achievement of better stereo, a secondary goal is getting past the periodicity of conventional clean plus LFO-modulated delay.  Unsynced LFOs lead to a richer "stereo" than counterswept delays or phase-shifts.