MultiPhase Diary


MultiPhase Technical

This page began in February 2006 as a journal to document the progress made on the MultiPhase project.

I was diverted from the project to pursue other projects, but it has always been my intention to complete the MultiPhase. 

Though at this writing (late May of 2010) the project remains unfinished, it remains one of my favorite endeavors.

In short, I am an effects freak, and am particularly enamored with modulation effects.  Of course, one of the premier modulation effects is phase modulation, and this page contains 34 samples of the MultiPhase doing its phase modulation thing.

I've recorded zillions of samples from all of my projects, but these samples rank as among my all-time favorites.  The concept of stacking many stages of phase modulation renders an effect I like to think of as "Flasing", which is a combination of "flanging" and "phasing".  With each phase stage added, the number of peaks and notches increases.  After a certain point, the phase shifting begins to sound less like classic phase shifting (such as the EH Small Stone) and becomes more of a flange-like effect.  However, it would take far more stages than 32 stages to produce something that sounds like delay based flanging.  Even then, I don't think a many-staged phase shifter would sound like a flanger, because the peaks and notches are not spaced in the same intervals as a delay based flanger produces.  Add to that the rather slovenly response of optocouplers (as compared to transistors or OTAs), and you get a rather "gooey" phase effect as opposed to a sharply defined phase effect one would expect from so many stages.

In any event, take a listen to these samples if you have the interest and the time.  You will see, as time progresses in this diary, more concepts are introduced and the MultiPhase begins to take on the characteristics of, well, the MultiPhase project.


22 February 2006

00:00 - 00:18 - Just the dry signal (DW6000) for comparison to what comes next.

00:20 - 01:24 - This is the Compact A arrangement. 8 stages, with negative
regen around the first two stages, through the tiny little passive LPF.

01:27 - 2:09 - 16 Stages, no resonance.

02:14 - 3:16 - 16 Stages, a moderate bit of resonance.

While testing the outer limits, I had 20 stages operating from the single CV circuit. This many stages abbreviates the sweep a bit, but it still was quite interesting to work with. Below is a sample of noise being put through the 20 stages with some positive regeneration. The modulating signal is a mix of triangle and pulse wave.

The following is a sample of noise through 16 stages of phase shift, with the phase modulated throughout the entire range. Initial high resonance, later resonance reduced somewhat. Modulated by a triangle wave.

05 March 2006

Tapping Different Stages for Stereo

Below are some stereo samples of the phase shifter. On each sample, the phase shift for one channel is taken from the 14th stage, and the other channel is taken from the 16th stage.

Each sample begins with the 'dry' signal, then has a section with the stereo configuration mentioned above, then ends with the same signal through the stereo inputs of the Dim C.

This first sample uses resonance tapped from the 16th stage. The Dim C is in Mode 4.

The next two samples were made with resonance tapped from 20 stages of phase shift. Stages 17 through 20 are modulated from a different control circuit, with less modulation depth, by the same LFO.

The Dim C portion at the end of the first sample is in Mode 3. The Dim C portion at the end of the second sample is in Mode 4.

15 April 2006

A bit of a brief update: I've got both 16 stage sections on the breadboard now. I have yet to put in the regeneration circuitry for the second section. The first section has a soft-knee compressor/limiter (a THAT Corp design using THAT devices) - more on that later. Below is a pic of the current breadboard(s) the MultiPhase is occupying.


MultiPhase Breadboard

Yes, it's a mess. The breadboard on the right contains the original 12 stages I started with. The breadboard on the left contains the compressor/limiter, wet/dry mixer, and four more stages for the first section. The top breadboard contains the second section of 16 stages plus the wet/dry mixer for that section or the chained sections.

32 Stage Noise Samples

The noise samples were done with 32 stages of phase shift with 32 stages of positive regeneration.

This first sample is of the MultiPhase modulated with the hypertriangular waveform.

This second noise sample is the MultiPhase modulated with a triangle waveform.

Each section can be separately offset, with a master offset to control both sections in tandem. In the following sample, the MultiPhase is again modulated with a triangle wave, but the second section of 16 stages has a negative offset, creating a uniquely different shift. Varying offsets between the sections provide a multitude of timbral variation.

Here's a sample of the DW6000 run through 32 stages of phase shift with 31 stages of positive regeneration. The sample starts with a short bit of the dry signal, then goes into the phase shifted signal. Modulated by triangle wave.

Here's a stereo sample of the DW6000 processed by the MultiPhase. One channel is 32 stages of phase shift, the other channel is 24 stages of phase shift. Five stages of regeneration are used. This is a good example of the Vactrols being modulated up to low audio rate!

30 April 2006

This weekend was the first real chance I've had to put in the regeneration and soft-knee compressor/limiter on the second section of 16 all pass filters. Having done that, it became possible to make some stereo samples with each channel having full regen capability. So, today I did just that =0).

Stereo Phase Shift Samples

All of these samples are processing the DW6000, using a common modulation scheme - each channel was modulated anti-phase. In the samples where more than one LFO was used, I mixed the two LFO's together sending the inverted copy to one section and the non-inverted copy to the other. In the final design I intend to have two LFO's, each capable of producing triangle, sine and pulse waves. Just as in my modular, I'm going to use Ray Wilson LFO's (that's what I'm using now). One LFO will produce an additional hypertriangular wave, while the other LFO will be variable between triangle/sawtooth, sine/'sinified' sawtooth', and variable pulse width. In addition to the LFO's, I plan to put in an envelope follower and external CV inputs.

All of these samples use only the MultiPhase for effects, no other effects were used - the path was DW6000->MultiPhase-> recorder.

Mixed LFO Percussive Sample

This sample embodies my philosophy concerning phase shifters: I love the traditional LFO swept phase shifter, but phase shifters are capable of so much more. In this sample, I mix a pulse LFO with a triangle LFO to produce a percussive type of effect. Resonant phase shifters make excellent percussive sounds! I have yet to pulse the front end of the phase shifter, which I think will be great for rendering exotic Tom type sounds, but in the meantime this sample gives a great example of the effect, I think.

Each channel is 16 highly resonant stages - I'm using 15 stages of resonance. One channel is using inverted resonance while the other is non-inverted.

16 Stage Swept Sample

This sample contains two different sections. Each channel is made up of 16 stages of phase shift and 15 stages of resonance. The difference between the two is the amount of resonance and the center and width of the sweep, as well as different voices from the DW6000.

Playing With 8 Stages

This sample contains three sections. The first short section is 8 stages of phase shift per channel, with 7 stages of resonance. Note that the lower number of stages renders something more along the lines of a 'standard' phase shift sound. My samples have always concentrated on the higher number of stages the MultiPhase is capable of, but it can do a pretty good 'stompbox' phase shifter sound, too (usually in the range of 2 to 8 stages).

The second section uses the same configuration, but with a wider sweep.

The third section uses 7 stages of regeneration, but 16 stages of phase shift, moving the timbre somewhat to outside of the range of the 'stompbox' style of phase shifting, but still retaining a bit of the essence of it.

More Mixing of Phase and Resonance Taps

This final sample of the day again embodies, at least to me, some of the unusual effects that phase shifters can impart. It uses 7 stages of regeneration, and 4 stages of phase shift. The really unusual part is derived from the modulation - it uses triangle, hypertriangular, and pulse modulation. Note that the inverted copy sent to one channel contains the inverted hypertriangle wave. This is one case where inversion will be quite different - for stereo anti-phase modulation using the hypertriangular wave, the original triangle wave must be inverted and then converted to a hypertriangular wave - other wise, it's 'upside down' to its intended purpose. But, the effect was pretty cool here, I think.

02 May 2006

Korg DW6000 Eaten By Phase Shifter

Tonight, as I was driving home from work, it occurred to me that I'd never tried the cross-regeneration idea after putting in the regeneration circuit on the second section. The idea sprang from the fact that a switch will be necessary to feed section B's regen tap back to the regen input on section A for cascade operation. Being a huge fan of the DPDT On-On-On toggle switch (one of man's greatest inventions, BTW) I figured why not provide another position on the switch and allow section A to feed its regen tap to section B while B is sending its regen to A. The stereo possibilities were something I could not imagine, so tonight I went downstairs to the lab and switched the two wires around. The result was fantastic - it turned the stereo image around in new ways.

Unfortunately, I never made a sample of a simple sweep. One of the first things I did was invert the regen on seciton B, which was receiving the regen signal from section A. I still had the patch set up from the last sample above. I tuned the center shift of both sections to the low end where all the rubbery peaks live. I then just hit 'record' and twiddled the modulation knobs for a bit with one hand, while hitting notes on the DW6000 with the other hand. Below is the recording, just as it was done.

IMO, It exemplifies the following things:

A. A phase shifter, especially a multi-stage phase shifter, can really tear things up.
B. Cross regeneration is pretty cool.
C. My wife really puts up with a lot of crap.

It's as above, DW6000->MultiPhase->recorder, no other effects other than the MultiPhase. This is no lilting Jarre Small Stone effect =-D.

06 May 2006

More Cross-Regeneration

I spent some time today documenting where I'm at and bringing both sections to the same configuration (I rewired the regen path on Section A, the first section I breadboarded, to the improved path of section B). I added a high cut control to the regen line - this works fabulously for helping to create some very nice phase shifts at the lower number of stages. One particular configuration it works very well in is using three stages for phase tap and regeneration - it's a wonderful 'terse' little phase shift. The high cut control sprang from the use of the passive low pass filter on the Schulte Compact A. I used a dual 50K pot and an 1800 pF cap, plus two 510 Ohm resistors. The fixed resistors are there to keep the regen line from breaking into ultrasonic oscillation at high regeneration settings.

I've recorded a couple of more expository samples of cross regeneration. In each sample, it's the DW6000 going straight into the MultiPhase and then straight into the recorder - no effects. Each sample uses anti-phase modulation between sections. I recorded first the 'normal' regeneration path, then switched to cross-regeneration and recorded again. No other settings were touched between the two regen examples in each sample.

This first sample uses 16 stages of phase shift and 16 stages of inverted regeneration on each section. The two sections are modulated by a single triangle wave (anti-phase). The sample starts out with normal regeneration, then switches to cross-regeneration. The difference in timbre is quite dramatic with this one; the cross-regeneration creates an effect I haven't quite come across yet. The source was a single held note of the DW6000 - I change the pitch on each oscillator of the voice as the sample progresses.

The second sample has each section using 16 stages of phase and only 2 stages of regeneration. The regeneration is inverted, and the high cut control was used to get a nice blend across the sweep. Each section is modulated anti-phase by a mix of triangle, hyper-triangle and pulse LFO's. The sample starts with normal regneration, then switches to cross-regeneration. Here there is some difference in Timbre, and the stereo modulation is obviously affected as well.

07 May 2006

The High Cut Control

As yet, I have done no more work on the circuit, but have recorded a couple of more samples.

The first sample deals a bit more with the high cut control and its use. The high cut control was inspired by the passive low pass filter of the Schulte Compact A phase shifter. In review, the Schulte Compact A is an 8 stage phase shifter with the unusal touch of using inverted, low pass filtered regeneration around the first two stages. This effectively provides a single peak sweeping along with the phase shifter.

The MultiPhase high cut filter is a passive adjustable low pass filter situated after the regeneration level/polarity adjustment on the MultiPhase. The regeneration first passes through the soft knee compressor/limiter, then through the regen level/polarity control, then through the high cut circuit, which is sandwiched between two buffers. After that, the regeneration is inserted into either the first stage or the second stage of the phase shift section, depending on if even or odd number of stages is desired.

One thing I haven't mentioned on this page yet is that, in addition to being able to select even/odd number of stages of regeneration per 16 stage section, the same control will allow selection of even/odd number of phase shift stages as well. The control itself will be a three position toggle switch with the following settings per position:

Position 1: Even Phase, Even Regen
Position 2: Even Phase, Odd Regen
Position 3: Odd Phase, Odd Regen

This development took place a while back through an exchange of correspondence with Juergen Haible, a man who actually knows what he's doing =0). Juergen quite well explained the effects of varying number of stages with a phase shifter, primarily concerning even and odd number of stages. At first I thought he was discussing regeneration - he was, but he was also talking about non-regenerative stages as well. An even or odd number of stages on the phase shift has an impact on the effect as well as even or odd number of stages of regneration. This control will allow selection of anywhere between 1 and 32 stages of phase shift as well as regeneration (though one stage of phase shift or regeneration isn't terribly good for much). Once one gets to two stages, things begin to happen.

Anyway, back to the high cut control: it can be used to tailor the high end response of any sweep, but it primarily comes in handy with a low number of stages of inverted regeneration. I think I understand, for example, one reason why the Schulte Compact A required the fixed filter: If one does not have that low pass filter in place, the inverted regeneration will take on a harsh squeal well before any good, intense regeneration comes into play. This is not so true once one gets to, say, four stages of inverted regeneration.

Which brings us to the first sample of the day. This sample uses odd number of stages (3) for both the phase shift wet signal and the inverted regeneration - the regeneration and the input signal are both inserted into the second stage, and the taps are taken off the fourth stage. The sample uses a hypertriangle LFO for modulation. It begins, briefly, with no compression/limiting, with the high cut set for no attenuation of the higher frequencies. In a word, it sounds like ass. The sweep sounds uneven between the low and high ends, with the treble sounding not all that good. At 00:11 or so, you can hear as I adjust the high cut for attenuation of the treble. This results in not only a more managable high end, but generally a better transistion between the low end of the sweep and the high end of the sweep. A three stage phase shifter can sound quite interesting in its own right, indeed!

The next sample is just an extension of using low number stages and negative regeneration around them, with the high-cut control providing a fine tuning of the effect.

This is a sample using (again) negative regen around the first two stages, with 16 stages of phase shift. Each 16 stage section is set up in this manner and are modulated anti-phase to each other. They're modulated by a combination of pulse and triangle wave. Note as the sample ends, one can hear a bit of self-oscillation. I used the soft-knee limiter compressor to compact things a bit - it in itself can have a drastic effect on the tone of the phase shiter when regeneration is used. It generally starts working on the lower peaks and, as its compression level is increased, begins to 'smear' the higher peaks as well. It provides some really smooth self oscillation. I haven't discussed much about how vital it is to the operation of the MultiPhase (it is), but that will have to happen another day....

14 May 2006

Cross-Fading the Wet Signals

I love it when a plan comes together. Before I'd even attempted it, I'd put in the wet signal crossfaders on my block diagram, just *knowing* it had to sound fairly cool. Last weekend I put one crossfader in and experimented with mono cross-fading. It proved quite nice for mixing together two phase shift responses in various ratios. That resulted in an even wider range of timbre available. Yesterday I put in the second crossfader, and immediately the sound I'd heard in my mind became a reality (this is the first sample listed below). That doesn't happen too often.

The crossfaders (Vactrol based adapted from a design on Mark Verbos' Simple-Answer page) are fed the wet signal from the phase taps on each end. This crossfaded signal is then routed to the wet/dry mixer of each phase bank for mixing with the dry signal. The crossfader controls are set up so that full CCW will mix only the wet signal from the respective phase bank it's assigned to. Full CW will fully mix only the signal from the opposite phase bank. Anywhere in the middle is a proportionate crossfade of the two signals. The real magic starts with voltage control of the crossfade. Applying the same LFO to each crossfader causes a rhythmic shift in the stereo image. Add that to the already shifting stereo field of anti-phase modulated phase banks, and the stereo image can get very interesting in a hurry.

Of course, depending on the speed of the LFO, the shape of the LFO, the modulation intensity, the initial center position of each crossfader, the number of stages of phase and regeneration used, and whether the phase banks are modulated in phase or anti-phase, etc., the effect can range from a very subtle stereo sort of vibrato to a very intense panned phase shift effect.

Yesterday I went through a recording burst and recorded the following samples using the crossfaders. The samples run the gamut of intense to subtle. The DW6000 is the source, the MultiPhase is the only effect used (DW6000->MultiPhase->recorder). I didn't write down any notes of the configuration of settings for each sample. Suffice it to say there is anti-phase modulation and in phase-modulation used, different phase taps and regeneration taps, varying types of regeneration, as well as triangle, pulse and sine modulation of the crossfaders at varying intensity, initial balance, and frequency.

The next experiment will be to see what effect inverting and high pass filtering the crossfaded wet signal will have. This is a technique used on the Roland Dimension choruses, and I'm curious as to what can be derived using phase shift. My curiosity in this was spurred by the following sample, made a few weeks ago, as an attempt to produce a phase shifted based dimensional chorus effect.

The sample starts out with the dry signal, then goes into stereo, unregenerative phase shift with anti-phase triangle modulation. As usual, no other effects were used:

In other news, I've begun work filling out the schematics and have started planning the phase shift boards as well. I've also mapped out the control strategy on a 3U rack panel - looks like nearly everything I really want is going to fit. I'll probably have to drop the idea of having two phase taps per phase bank, but everything else discussed on this page is there.

Relevant Link:


Mark Verbos' Simple-Answer DIY Page

21 May 2006

Envelope Follower

Yesterday I breadboarded the final unexplored section of the planned MultiPhase - the Envelope Follower.

There are a number of envelope follower designs out there. In fact, in keeping with the spirit of the Vactroality (new word) of the MultiPhase, I could have used a Vactrol based envelope follower. I chose not to, mainly because I get enough of the Vactrol slew from the phase shifter itself. I wanted to maintain a fairly sharp response from the EF. I settled on the envelope follower at René Schmitz's site for a couple of reasons - it appeared to be a one control no-fuss envelope follower (I'm at the max for panel space) and you just can't go wrong with anything René has on his site.

The envelope follower is itself an adaptation René did of a design from Elektor magazine. René spiffed it up for operation at +/-15V levels. As envelope followers go, it's a fairly unusual design - it's really just as much a sample and hold as it is an envelope follower.

The envelope follower, as it is posted, has a good amount of ripple on it - I added a simple variable lag circuit after it to smooth it out. I also boosted the level out of it to match the hungry demands of the MultiPhase CV input. The MultiPhase EF section consists of the control in the original design which I label as 'sensitivity' and the variable lag control. In the modulation mixer section, both phase banks have a bipolar attenuator which processes this voltage for each bank. The envelope follower is also one of the four CV sources that are routed through the modulation submixer so that one control can be used to adjust the effect it has on both phase banks if need be.

The envelope follower really brings a lot to MultiPhase. Following are some samples I created yesterday while playing with it. As usual, Korg D W6000 is the source going through the MutliPhase straight into the recorder.

This first sample is a short example of what the EF does with a short attack and decay note. A bit of the dry signal is played, then the effected signal is heard:

Below is a sample of the envelope follower processing a note with a longer decay. The intensity of phase shift and CV effect on the phase shifter is toned down so that the MultiPhase blends in more with the note providing a nice enhancement. A bit of the dry signal is played, and then the effected signal is heard:

The next sample is a good demonstration of a long attack and long decay. By controlling how long a key is pressed down, one can control not only the amplitude, etc, of the note, but also the phase shift. I think this is a good demo of a high number of stages (32) being processed by the envelope follower:

Finally, below is a sample of the envelope follower processing notes that have a long attack and short decay. At the end of the sample is a small bit of the MutliPhase controlled by a small amount of pulse LFO, a great amount of triangle LFO, and the envelope follower.

Relevant Link:


René Schmitz Synths'R'Us Site

28 May 2006

Why I Never Get Things Done

Lately I've been putting the phase stages on protoboard (which I've completed without so much as a single hitch to trigger my debugging state). After I'd completed Phase Board A, I decided to go ahead and match the capacitors to the Vactrols on Phase Board B (which was still on breadboard). After doing that, I started testing the circuit to make sure everything was fine. Oh, it was. Too fine. As usual, I started noodling around with it, and ended up with one of those MultiPhase patches that makes one wonder if it really should be called a phase shifter. I mean, it is a phase shifter, and it will shift phase like any other phase shifter out there, but with all of the modulation possibilities, mixture of stages and resonance, and so on and so forth, it often comes out sounding quite unlike the effect one often associates with phase shifting. It goes so much deeper. I've thought about calling it a 'phase manipulator', but that at once sounds much too grandiose and precise for what it really is.

Anyway, I come up with these really cool patches all the time with this thing. 99% of them go unrecorded, but this time I went ahead and recorded it. The recording is a mixture of two mono tracks. With this recording, I did use the Dim C and the spring reverb from my Sidekick amplifier. Shoulda used the Alesis Wedge, but this was a spur of a moment thing. The sample is a good example of the unphaser-like aspect the MultiPhase quite easily produces. I realize other phase shifters will modulate at audio rate (in fact, non-optical based phase shifters are capable of producing a much more intense effect), but the available modulation sources combined with everything else in this project make it seem to come to life in a different way. I'm really getting excited about it.

Each track is the same voice from the DW6000. The DW6000 runs through the MultiPhase, through the Dim C (Mode 3), then through the sidekick, with the line out of the Sidekick jacked into the recorder. The Multiphase is modulated at an audio rate with the triangle wave, LFO rate by the hypertriangle wave, and also by a fast narrow pulse (all LFO's that will be available on the finished product).

The first recording is a brief bit of the DW6000 voice through the Dim C and the Sidekick reverb so that the effect of the MultiPhase can be discerned. I used this same voice, with longer attack and release to get some of the wailing sounds, along with the short attack/release heard for the percussive sounds. The audio rate modulation combined with the two slower LFO's of the Multiphase produces an unreal change in the timbre. Throughout the recording, I would shift the regeneration from non-inverted to inverted and vice versa from time to time. I used 16 stages of phase shift and 15 stages of regen, IIRC.

01 June 2006

Moving in Quadrature

Yesterday I breadboarded Thomas Henry's Quadrature Function Generator and an additional hypertriangular waveshaper. I must say, it immediately earned the job as LFO3. Modulating the second phase bank with a 90 degree offset wave makes for some great stereo effects. I think even in mono mode it will prove to be quite interesting.

Both the 90 degree hypertriangular wave and the 90 degree triangle wave have something to offer. As detailed on my MultiPhase Project Page, both waveforms will be available, plus the standard antiphase 180 degree configuration. Even with no modulation offset, using banks A and B in parallel creates a nice stereo image - there is always enough of a difference between the two phase banks to elicit a stereo field. Adding 90 degrees mixes it up a bit, and 180 degrees mixes it up a bit more.

Here are a couple of comparative samples of 0, 90, and 180 degree offset of the modulation of Phase Bank B in relation to the modulation of Phase A.

Each sample contains three parts in this order:


1. Phase Bank B modulated by a hypertriangular wave with no phase offset to Phase Bank A.

2. Phase Bank B modulated by a hypertriangular wave with a 90 degree offset in relation to the hypertriangular wave modulating Phase Bank A.

3. Phase Bank B modulated by a hypertriangular wave with a 180 degree offset in relation to the hypertriangular wave modulating Phase Bank A.

The first sample is a slow sweep using fairly high regeneration. The output of each phase bank is tapped at 15 stages, with 3 stages of positive regeneraion (Even/Odd is set for Odd/Odd with the output tap at 16 and the regeneration tap at 4 for each bank).

The second sample is a much faster sweep, with less regeneration. Unfortunately, I forgot to note which taps I was using =0).

The source of the sound is a plain-Jane, unmodulated voice from the DW6000, which is going into the MultiPhase circuit mono and each bank is connected straight to the recorder (no FX used).

TopMutliPhase Technical
This page copyright (c) 2010 Scott Stites. All rights reserved.