How to improve almost any phaser

[Mark Hammer]

I tried a quick and easy experiment with a Ropez/Ross phaser last night and was pleased with the outcome.  The boards that Francisco Pena laid out and sells for that phaser make it easy to implement, but it can be applied to practically any phaser you can think of.

When it comes to phasing, the result is a product of the sum total of all phase-shift applied.  So, 4 stages of phase shift provide up to a maximum of 90-degrees of phase shift per stage at any given frequency, which totals a maximum of 360 degrees of phase shift.  Essentially, you get one notch produced per 180 degrees of total phase shift, so 360 gets you two swept notches.

Since the sum total phase shift is what matters, not all the stages have to sweep.  certainly most of them need to, or else you wouldn't hear any swirl, but they don't ALL need to.  A great many commercial phasers make use of this principle, and include some fixed stages on top of the swept stages.  Probably the most commonly encountered example is the venerable MXR Phase 100, which uses 6 LDR-based swept phase-shift stages, but adds 4 more fixed stages on top of that for a total of 10 stages, producing up to 5 notches ("up to" because not all notches are audible or within the passband at all points in the sweep).  Boss also uses the addition of some fixed stages to spruce up some of their phasers.

So, last night I decided to throw in 2 stages of fixed allpass/phase-shift to this Ropez, via a small daughter board that could be easily assembled in a postage-stamp, sized piece of perfboard, and hot-glued to the side of the chassis.  I used what has become almost standard fare for such things: a 4558, 10k feedback resistors in each stage, 10k to the inverting input, .01uf to the noninverting input, and 10k to Vb.  That's a total of one 8-pin DIP, 6 resistors, and 2 capacitors.  You've seen this a million times.

The Ropez (at least the last iteration of it) has a number of pads aimed at integrating another quartet of swept allpass stages to make an über-phaser with 8 swept stages.  On the diagram, two of these pads are labelled X and Y and are for inserting the additional 4 stages that Francisco made a daughterboard layout for.  Cut the path between these two pads, run X to the input of one of the fixed stages, run the output of the second back to Y, connect your V+, gnd, and Vb, and away you go.  You now have 6 stages, 4 swept, and 2 fixed.  The same logic can be applied to a Small Stone,  Phase 90, or whatever.  The regeneration works exactly as before.

How does it sound?  A little richer, and to my ears a little more pleasing at the top of the sweep.  Not surprising since the added phase-shift really only starts to kick in above 1500hz or thereabouts.  I tried it out with the piezo bridge pickup on my Parker this morning, and it sounds great for that.  I like the vibrato produced when the dry is cancelled a little better too.  On this particular phaser, I have a switch for going from phaser to phasefilter (two stages are converted from allpass to lowpass), and am pleased to report that the mod does not interfere with the pleasing sound of a phasefilter.

While it might be a little tricky doing this to a P90 installed in a 1590B chassis, it may well be possible to build it onto a small, low-profile board it you are handy with surface-mount op-amps and 1/8w resistors.  It's one of those things that you can build on a small daughter-board and try out on whatever phaser you have handy.  If you like it, keep it, and if you don't, it's no big deal to put things back the way they were.

But on the whole, definitely worth trying.

[aziltz]

I don't want to detract from such a detailed and focused thread, but I looked around quickly for "phasefilter" as it seems really interesting.

Is it possible that this is how the EHX Worm gets all its multi-effects?  At first glance, it looks like a multi-stage phase shifter, but by redirecting the feedback and a few outputs (?) you can get Tremolo, "Wah" and Vibrato all from the same LFO and "shifting network".

Are there any concrete or schematic examples of how to wire phasefilter from an existing phaser?

Thanks for all the info Mark.  I'll surely try this and your new suggestions for additional fixed allpass/phase shift stages in the future.

[Mark Hammer]

Ton Barmentloo is a generous and well-respected member of this forum.  I make a point of not inquiring about any of his designs in too much detail because I know that IP is his livelihood.  The Worm is one of his.  That being said, the Worm is optical, and the phase-filter option is something that essentially relies on using OTAs (3080, 13600, 3094, etc) for the allpass stages.

You can learn more by looking up the SSM2040 datasheet ( http://www.synthdiy.com/show/?id=85 ) and by looking at the design of the AMS-100 phase-shifter projects in DEVICE ( http://hammer.ampage.org/files/Device1-6.PDF ).

The original phase-filter idea, from John Blacet, assumed the user would be combining clean plus processed signal.  In playing around with it, I found that some excellent and truly distinctive sounds could be obtained by using only the wet signal.  As a result, if I'm going to implement aphase-filter setting/function, I always include a vibrato (dry-cancel) switch as well.

[puretube]

The "WORM" is just a 3-stager... 

[puretube]

Mark: your understanding my positions is well appreciated!

Philtering, Worming, Wahbrating, and Vibwahting can be done with all those more or less known circuitries/technologies/topologies...

the "Worm" bore out of the fact, that the textbook "...pass-transformations" and the Fourier and Hilbert thesis`
didn`t tell me all I wanted to know...

for me, the all-pass was missing in there...


Now when you look at the simplest allpass building block: the phasesplitter with an R and a C
between the inverting and the non-inverting output of a triode or BJT,
1.: you`ll see that from one of the outputs
the R-C series-hookup
looks like a Lo-Pass,
while from the opposite phased output,
it looks like a Hi-Pass...

Combining (adding/summing) these, we get an All-Pass, as is well known...

now: Serially combining Hi- And Lo-pass gives a Bandpass...

Subtracting a Hi-pass from the dry turns into a Lopass,
Subtracting a Lo-pass from the dry turns into a Hipass;

... sorry- I got distracted while typing ...

now: if one combines the lowpass or highpass during the right phase-relationship
with the original ( = dry ) signal or with the 1st All-pass (@ the proper amplitude.relationship)...

[will be continued, one day...]

[aziltz]

i didn't mean to sidetrack the thread, but at the same time, thanks for those comments!

[petemoore]

  On my Tonepad Small Stone PCB, it looks like an insertion point for the extra stages could be made by cutting the trace between R33 and R41 ?
  The two added stages would consist of what is between Q6 emitter and Q10 emitter?

[Mark Hammer]

Yep.  From the looks of it, that's a viable insertion point.  And what you're going to insert there is essentially a replica of what you see built around IC6 in the RPH-10 diagram shown here.  The diagram shows 5k6 resistors to Vb.  I used 10k, but you can use whatever you want, keeping in mind that the corner frequency where they start contributing 90 degrees of shift is given by the ever-popular F = 1 / (2 * pi * C * R).

[Mark Hammer]

Playing around with the pedal last night, I became convinced that I need to figure out how to envelope-control it.  The sound of a gradual upward swell with two notches and lowpass, and some resonance, using only the wet signal, is lush, lush, lush.  Like a dreamy synth pad when applied to chords.  Tack a bit of chorus and decent reverb on top, and you have no need for guitar synths.

The trick is not just getting envelope control, though.  One needs to keep in mind that the sort of sound produced when the on-board LFO is set for slower speeds is NOT equivalent to a straightforward amplitude envelope.  The LFO sweep moves upward exponentially, so there would need to be some sort of envelope follower that would:
a) provide sufficient lag and control over attack/rise time,
b) translate that attack into an exponential sweep,
c) convert that exponential sweep into suitable current to feed the OTAs.

But believe me, the resulting sound is worth it. I gotta make some clips of this.

[space_ryerson]

You've definitely got my attention! I wonder if I could fit this (with the envelope) in a Phase 90 where the battery would reside?

[Mark Hammer]

With an envelope follower, it would need your battery slot to be able to fit in a 1590B, although without an envelope follower, the pair of additional stages actually take up fairly little room, and would even allow you to keep the battery in.  The dual op-amp itself is the biggest part.  1/8W resistors take hardly any space, and you can certainly use small monolithic ceramic caps for the phase shift stages without any space or audio quality concerns.

Note, however, that the rising sweep I'm talking about is not attainable on a P90.  It needs an OTA-based phaser like the Ross or Small Stone, or the DOD FX20.

[petemoore]

  What is the partial schem from ?
  It's stages differ from the SS stages, no Q followers [ ], but it has a feedback path that...goes a different way than the SS's.
   
 

[Mark Hammer]

Well, I made a 2-stage daughter-board last night, as small as I could, using 1/8W 10k resistors and small ceramic .01uf caps.  The op-amp was a normal 8-pin DIP dual.  Total footprint when done was 5 perf holes by 10 (plus the board area between the holes along the perimeter and where the next row of holes would be).  The small caps and resistors assured that the IC (unsocketed) was effectively the "tallest" component on the board.  So, this is pretty darn small, and the sort of thing one could easily fit somewhere in a 1590B, between a pot and the side of the chassis.  In some respects, the wires connecting it to the main board occupy more space than the daughter-board itself.

pete,
The partial schem is the Boss RPH-10, which uses a pair of IR3109 chips for allpass stages, in addition to the op-amp stages shown.  It is basically the Boss PH-2 phaser pedal, plus some extra bells and whistles.  The IR3109 is four OTAs in a package, similar to the old SSM2040, but made specially for Roland/Boss.  It allows for a fairly compact phaser when you want to have a lot of stages.  Consider that it would replace four CA3094s or a pair of LM13600 chips.

[space_ryerson]

With an envelope follower, it would need your battery slot to be able to fit in a 1590B, although without an envelope follower, the pair of additional stages actually take up fairly little room, and would even allow you to keep the battery in.  The dual op-amp itself is the biggest part.  1/8W resistors take hardly any space, and you can certainly use small monolithic ceramic caps for the phase shift stages without any space or audio quality concerns.

Note, however, that the rising sweep I'm talking about is not attainable on a P90.  It needs an OTA-based phaser like the Ross or Small Stone, or the DOD FX20.

Thanks Mark. 5x10 perf is tiny!

If I really feel the need for a rising sweep, I'll take a look to see if my other phaser is OTA based. It's an old Tokai, which I haven't really analyzed circuit of.

[petemoore]

  I have the parts, hot iron, everything laid out...but just couldn't seem to try it without understanding better, or having a schematic for it.
  The RPH-10 opamps [around IC6] are wired quite differently than the arrangement in the Small Stone.
 Looking at the Tonepad Schematic, Cut the connection between R33 and R41.
  Describing the RPH-10's IC6 ..Insert circuit:
  10k/.01uf [10k to an OA Neg input, also the 10k feedback resistor], .01uf to "10k-Vbiased" +oa input.
 Output through 10k to a duplicate wired circuit [the other side of Dual OA] 6k8 output from the second opamp back to the board splice point.
 Also V+ and Gnd. to Dual opamp.
 The reason I ask is because the circuit I wish I could draw up [layout creator giving fits for sometime trying], looks quite unrelated to the phase stages shown in the Small Stone, at the same time, I like that they look simpler.

[StephenGiles]

With an envelope follower, it would need your battery slot to be able to fit in a 1590B, although without an envelope follower, the pair of additional stages actually take up fairly little room, and would even allow you to keep the battery in.  The dual op-amp itself is the biggest part.  1/8W resistors take hardly any space, and you can certainly use small monolithic ceramic caps for the phase shift stages without any space or audio quality concerns.

Note, however, that the rising sweep I'm talking about is not attainable on a P90.  It needs an OTA-based phaser like the Ross or Small Stone, or the DOD FX20.

This is my adaptive peak following envelope generator



Could be useful in this application.

[frequencycentral]

Watching this thread like a hawk!

So Mark, how would you contrast the addition of two static stages versus the addition of two modulated stages?

[Mark Hammer]

It's a slightly different sound.  Remember that when all stages are modulated, the point where they add up to enough phase shift to produce notches keeps moving.  The fixed stages I described add their maximum phase-shift above 1590hz (with amounts lower than 90 degrees below that point), but not before it.  So in a way, the effect becomes a little stronger or more evident during the uppermost parts of the sweep.  It's not huge, I have to say, given that it is after all only 2 stages.  But it's almost like the unit evolves from 2 to 3 notches as you move past the halfway point in the upward sweep.  All those companies that added fixed stages on top of the swept ones knew what they were doing.

[frequencycentral]

I just received my Ropez PCB and the 4 extra stages board. Well I might as well wedge a few more (static) stages in there too! I was thinking of two quad opamps on a daughter board, both configured as 4 static stages, so with 4 static stages either side of the Ropez extra stages board. 

[StephenGiles]

I just received my Ropez PCB and the 4 extra stages board. Well I might as well wedge a few more (static) stages in there too! I was thinking of two quad opamps on a daughter board, both configured as 4 static stages, so with 4 static stages either side of the Ropez extra stages board. 

Mmmm could be easier to just build an Eventide Instant Phaser!