Modularized phaser at GEO

[R.G.]

Incidentally, we have yet to see a single design/layout here that tkaes the delightful properties of SIP op-amps into account.  These little beasties, I hasten to remind you, are now part of Small Bear's regular stock (thanks Steve, time for me to justify that recommendation), and can make layout of just about anything a cakewalk.

Good point, and easy enough to do. I haven't messed with them because they are a little harder to find and because the outputs "point" different ways out of the chip. If the outputs pointed the same way they'd be flat out great for phasers.

But that does bring up point 7: component height.

Committing to a fixed-height installation where you're trying to get boards to stack makes component height critical. The slots in the box are 0.098"/2.5mm wide and the same apart, so board spacing is fixed in increments of 0.196"/5mm. So boards can be placed in increments of 5mm/0.196", giving spacings of 5, 10, 15, 20, etc. and
0.196, 0.392, 0.588, 0.784, 0.98, etc.

Board stock is usually 0.062"/1.6mm thick, and the wires between boards have to be folded in there as well. The wires may be able to weave through components if you don't use ribbon cable, though. Through-hole pins stick through the board by about 1mm, perhaps. So you lose about 2.6mm in thickness for the boards and leads.

That leaves component heights available of 2.4mm (probably not enough except for a bare board), 7.4mm/0.291", 12.4mm/0.488", 17.4mm/0.685", etc. That has implications:
(a) You can't use stand-up 5x11mm electros on boards which will be on every-other-slot spacing. To get boards on every other slot, you have to use lay-down electros or the ultra-miniature shorties.
(b) You probably can't use greenies film capacitors without laying them down, likewise eating up board layout space.
(c) A 9mm 90 degree mount pot is about 12mm/0.472" high, so boards with pots have to go 0.588"/15mm spacings or more.

Has anyone noticed that this is an exercise in the water-balloon approach to design - any time you fix one part of the design, the other parts have to be squished around to fit that constraint! In this case, the choice of capacitors is limited because you picked a particular box to start with. And we likely can't use TO-220 regulators without laying them flat. And it's really, really hard to use standard sized pots. This principle is one that does NOT get taught in engineering schools. You learn it by having it quietly circle around behind you in designs and biting you in the butt. 

[KMS]

The P box will make more room for sure.

The width is still used up at the board location but now we have a little over 4" for controls.  2" on the lower part of the box is plenty for the stomp switch and in/out jacks.....and 2" room on top for LED, DC jack, a couple of pots, and the battery and maybe a micro toggle and some other mod control.

The wiring going across and between the two boards is how would have tried to wire it no matter what......longer leads....more working room.

Sounds like it will work just fine.

[mdh]

R.G., any news on where this project stands at the moment? I'm still uncertain about the feedback arrangment (point M in the schematic?).

I really like the idea of using pcb mount controls, particularly with the boards being mounted vertically in the box. For a modular Phase 90++ (sensu Stroustrup) the rate pot is easy -- just stick it on the control board. Whew, that's done, two off-board wires eliminated. Other controls will be a little more tricky, though. If each phase stage board has its own bias trimmer, then the typical depth mod (a pot in place of the 1M resistor from the gates of the FETs to the wiper of the bias pot) wouldn't be feasible beyond the 8-stage scenario. If we want to use pcb mount pots, we're kind of screwed for even 8 stages, because a dual-ganged pcb mount pot is going to occupy serious real estate, and we need two more wires between the phase boards. However, I'm not really certain that this would be necessary, because I don't think that it's really that unreasonable to obtain matching octets and dodecets. Hexadecets may be troublesome, but if you're that serious, just buy stock in Fairchild and be done with it  IIRC, my batch of 50 Fairchild 2N5485s would have yielded a tolerable dodecet, or a couple of octets. Maybe I got lucky, but I doubt it. With creative jumpering of the bias trimmer and R113 on phase boards 2+, we could connect the gates of all FETs to one point if desired, making the depth mod feasible without a multi-gang pot (I think that's right). Perhaps you had that in mind already.

I'm a little more confused about feedback, as I alluded to above. I can imagine adding some pads to the phase boards such that we could mount a feedback pot on the first phase board with jumpers on subsequent boards. I'm assuming that the feedback resistor should be between the first and last stages.

My concerns are:

1. Maybe I'm advocating adding too many pads, or too much complexity in the way of options.
2. Ticking. As petemoore mentioned, phasers like to tick. I'm a little concerned with the number of wires between boards, and whether we're asking for LFO noise trouble.
3. I'm not sure how good the depth and feedback mods I described above really are. I've read about them, but I've never tried either of these mods. Maybe they aren't worth the trouble.

My non-concern is:

1. Machining boxes for board-mounted controls. Every box I drill is a custom job, and I don't think it would be that hard to get everything lined up. Once you know how many slots you're skipping between boards, just make sure there's room for the pots, draw yourself a little template, center punch the pot shaft centers from the inside, and you're good to go. A milling machine would be nice, but not really necessary.

[R.G.]

Good thoughts - thank you!  I was actually paused, waiting for more of this kind of thinking before I just plunged ahead and did it however I felt like.

First, what I have in mind now is more modular than I originally started with. It's kind of FxBus in a box. The way I'd do a modularized phaser with this setup is that the input/output board is one module. It makes the rest of the world safe for the modules inside their little box. The next module would be an LFO, good for driving one or more phase boards. Each phase board would have a number of phase stages, and would expect as inputs an input signal and an LFO signal. The LFO signal could well have the depth trimmer back on the LFO board, and the phase boards would not need to know about it. The wiring ticking could be solved with careful layout and perhaps using RG174 coax for the LFO wire between boards.

What I have come up with is a bus definition, which wires run between boards, and I have been running test cases in the back of my head for whether there are enough wires of the right kinds between boards to let you do any kind of effect module. With the right setup, I think you could string together a fancy pedal that can't easily be duplicated any other way, and certainly isn't modified as easily any other way.

So the wires I have now are:

• Power -  (+9V) which runs into and through the board to the next board. Any +9V needed is tapped off this line for the board's use.
• Power ground  - "sewer" ground, used for returning power back to the power supply, not to be used for signal ground because this wire contains all of the used electricity from all the boards, and so is highly contaminated
  
• Signal ground, to be used for a 0V reference
• Vref; I go back and forth on this one. Half the 9V supply is such a common need that I thought it was worth the wire. But then all it takes on a board is two resistors and a cap to re-create. I guess that for signals between boards, a master Vref lets you skip a capacitor coupling, but if the Vref is only used on the board, it's not needed. Still pondering this one.
• Signal In; same position in the buss as signal out on the other side of the board. The effect takes signal in, does something to it, and spits it out on Signal Out.
• Signal Out; same position in the buss as signal in on the input side of the board. The effect takes signal in, does something to it, and spits it out on Signal Out. A board may also be a purely sidechain module, in which case it monitors the signal line, but the signal in and signal out are connected with a wire or a buffer.
• Sidechain; The sidechain line is an insert line; if the module received a sidechain signal, like an LFO or an envelope signal, it gets it from the sidechain input side. If it creates a sidechain signal, it creates it and spits it out on the sidechain output side. If it just uses the sidechain, it passes the sidechain signal through to following modules.
• Sidechain-n; may need more than one sidechain wire. Still thinking about this one. Maybe it's worth spending a wire for a dedicated LFO sidechain wire and a separate envelope sidechain wire. Ideas?
• Utility wire; passes through unchanged, unless you need it. Intended for things like phaser feedback or other local communications between modules which need it but don't need to put it on the bus to other boards.

That's nine wires. Too many wires and you eat up the entire module board with wires and pads. But not enough and you can't do flexible insertion of another module.

On controls - I think the controls have to be PCB mount miniatures, with well defined locations so they can be brought out to the top of the box. Otherwise you eat up the whole box top in controls. My sketches show that you could feasibly do a three-control board this way, but only with a specific (Mouser-available) 9mm square pot and a self-knob shaft. The ribs in the box help you locate the pots accurately.

On the bias trimmer issue for phasers - maybe we have to redesign the way the LFO spits out the LFO, or possibly the way the phase board accepts it. The phase board could well take in whatever the LFO line presents it, then add/subtract its own local bias voltage to match its own JFETs. I think that's a lot better than trying to match 8, 12, or 16 JFETs, at least for the casual DIYer.

More thoughts?

[lovekraft0]

On the bias trimmer issue for phasers - maybe we have to redesign the way the LFO spits out the LFO, or possibly the way the phase board accepts it. The phase board could well take in whatever the LFO line presents it, then add/subtract its own local bias voltage to match its own JFETs. I think that's a lot better than trying to match 8, 12, or 16 JFETs, at least for the casual DIYer.

Now that's a great idea - using a "universal" LFO signal, like voltage control in a modular synth, and having offset, level (and perhaps even inversion?) available on each board would make for a much more versatile system, and perhaps even allow for an interchangable envelope follower module, for those triggered phase sounds.

[moritz]

On controls - I think the controls have to be PCB mount miniatures, with well defined locations so they can be brought out to the top of the box. Otherwise you eat up the whole box top in controls. My sketches show that you could feasibly do a three-control board this way, but only with a specific (Mouser-available) 9mm square pot and a self-knob shaft. The ribs in the box help you locate the pots accurately.

While I'm out of my depth on the more technical aspects of this project, I have some thoughts regarding the layout...

I've made up a little diagram (below) of roughly how I visualise the layout from RG's description, with the vertically mounted phaser-stage-modules and diagonally running interconnecting wires/ribbon cable (thus allowing the modules to 'unfold' concertina style once outside the enclosure).

Now, if you're going to be using those 9mm PCB mount pots, then aren't the enclosure slots unneccesary for holding the modules in the vertical position, and couldn't they just be held in place by the pots alone? This way you're not restricted to that specific type of enclosure or how you orientate the modules within the enclosure. While it's nice to have compactness, I'm thinking especially for this type of project I'd rather not compromise on features and expandability just for the sake of fitting it in a specific box. Besides, if the end result is phaser that does the job of several other phasers, I'm sure most players won't mind giving up a bit of pedalboard real-estate.

I'm not sure if what I've pictured is totally correct with regard to all the technical aspects discussed here, but the diagram below kinda shows what i'm getting at... A 'roomy' enclosure with the modules horizontally across the top half, leaving plenty of room in the bottom portion for the stomp switch, I/O and other boards, battery, in/output jacks, LED, etc., etc...

[R.G.]

Good comments, m.b.

Your drawing is a good conceptual sketch, although the output line at the end of the chain actually goes back through a wire already put into the wires running between the boards, just to keep from having to run another wire.

The reason I think we still need the ribs on the side of the box is that some boards, most of them probably, will not have pots. The I/O board might have a volume pot. An envelope generator board doesn't need a pot, although it may need a trimmer to set levels. A phase board doesn't need an external pot at all.

It's funny - this concept is exactly the FxBus idea, but with a common, widely available box to put it in. I got about two responses on the FxBus in toto. The inclusion of a box with ribs to hold small boards made this incarnation take off.

A more general way to do this might be to take a piece of 1x8 pine lumber on a radial arm saw and cut thin saw kerfs every 1/2" for about two feet. Then cut off the board at 2' long, and split the board lengthwise, making two opposing board retainers. These holder boards could then be attached to ends and a bottom and make a modular board holder of arbitrary size. All's you gotta do then is make up boards to physically fit and be held in by an aluminum top plate. Some old combo organs were once made this way, in fact. But I bet no more than one or two of these would ever get built. 99% of all effects builders will never venture outside the box-with-two-jacks approach to effects.

As I said in the first iteration of the Guitar Effects FAQ, it's the mechanics that are difficult, not the electronics.

[KMS]

I have a question about FX that are phased by LFO.

I know of three FX like this....but it would be nice to have an exhaustive list of all FX that could be phased by a universal LFO.

Wha Wha (phaser)

Tremolo

Pan

Is there more?

[R.G.]

Just about anything that has a control can be modulated by an automatic knob turner - which is what an LFO is.

Imagine you have a micro-genie that does whatever you tell it to. You tell it to sit and turn the knob on a pedal back and forth at a certain speed while you play. Now list whatever knobs you can have the genie turn.

Any filter, including a wah (bandpass or peaky lowpass) bandpass, lowpass, highpass.
Amplitude/volume give tremolo.
phasing
Delays of all kinds, including flangers, chorus and echo
reverb depth
The list goes on.

One interesting thing is that you can even have an LFO modulate the speed of another LFO. I have a couple of these over on GEO. There's LERA, intended to add a leslie-like speed up/slow down for LFO speeds, and the auto-magical startup of the LFO on the EA tremolo on each note.

So the list is exhaustive - any pedal with a knob.