How to improve almost any phaser

[Mark Hammer]

Here's a picture to illustrate just how small and compact you can make the daughterboard.  The Ropez board is shown for reference.  Again, note that the height of the add-on board is basically the height of the chip itself.

[merlinb]

So, 4 stages of phase shift provide up to a maximum of 90-degrees of phase shift per stage at any given frequency,

Shouldn't that  be 180 degrees per stage? Well ok, not quite 180, more like 170 in practice. (This is my first foray into phasers!)

[Mark Hammer]

If you connected the non-inverting (+) pin directly to ground, and didn't have the JFET or the cap tied to the + input, yes, you would be right.  It would be a 180-degree inversion of the signal.  It is the differential nature of the signal applied to the inverting and non-inverting pins that limit the phase-shift of some content to 90 degrees max, per stage.

Although let me take a few steps back from that and ask for the experts to step in here.  I know that, at one level, each stage IS essentially inverting.  We can easily demonstrate that by the fact that resonant feedback needs to pass through an odd number of stages to have the appropriate phase relationship, regardless of how many stages there are altogether.

So, the phase-shift that is applied, is essentially phase shift within each stage, ( ) though we think of it cumulatively across the stages.

The use of "maximum" implies that some frequencies enjoy less phase shift by being closer to the boundary of the passband.  So, while there may be 90 degrees of phase-shift in stage X for anything 1khz and above, there may be only 45 degrees phase shift for content at 500hz, and less for 250hz, 125hz, etc.

[frequencycentral]

Regen passing back to an even number of stages sounds really good. You should have a listen to my Sonic Death Ray soundclips.

[merlinb]

If you connected the non-inverting (+) pin directly to ground, and didn't have the JFET or the cap tied to the + input, yes, you would be right.  It would be a 180-degree inversion of the signal.  

But when the FET is driven hard on then the + input basically is shorted to ground?? I got this:
http://s81.photobucket.com/albums/j207/merlinblencowe/?action=view&current=CIMG5703.mp4
I'm not seeing things right, that is swinging from zero to 180 degrees?

[Mark Hammer]

Is that an emulator, or an actual functioning circuit?

[merlinb]

Is that an emulator, or an actual functioning circuit?

Real circuit, hooked straight up to the scope. A single opamp band-pass stage (no resistor in parallel with the transistor though, like you sometimes see). Signal 1kHz, 200mVp-p.

[Gurner]

Is that an emulator, or an actual functioning circuit?

Real circuit, hooked straight up to the scope. A single opamp band-pass stage (no resistor in parallel with the transistor though, like you sometimes see). Signal 1kHz, 200mVp-p.

I reckon it's your 1khz test signal that's the issue.

Take your signal's frequency down to 82.4Hz (this being the frequency for low open 'E' on a guitar) & rerun the test - for a one stage allpass, you should  see an initial 180 degree inverted signal (ie vs the original - & provided you're at the right end of the 'sweep') which will lag as you 'sweep'   ...for typical components used in a phaser stage,  certainly at 82.4Hz, it won't shift a full 180 degrees (the max 'shift is likely to be nearer 30-60 degrees)

here's a video I've just made like yours, but at 82.4Hz, a real circuit (not a sim) using a 4.7K & 0.1uf (it was all I had to hand), blue trace is  the circuit's input signal, pinkish trace is output from a one stage all pass...

http://www.youtube.com/watch?v=jbMc0uPN_pg

[frequencycentral]

[frequencycentral]

Verified:

[frequencycentral]

I've got a Phase 90 PCB hooked up to my breadboard ATM, just testing a few ideas using the fixed stages board. First result that I REALLY REALLY like is to use 0.01uF for C1, C2 and C3; remove C4 and stick the PCB in the regen loop - really nice, almost 'vocal' sounding.

[frequencycentral]

So............even lower values for C1, C2 and C3 are even better, such as 3n3 or 2n2. It works well with 3 stages but dissapears if I add any cap in the C4 position. I've added a switch to select between normal regen and Borg Implant regen. I've got a whole new range of regen timbres now at the flick of a switch. This phaser also has the width and bias pots. As a side note I had no 4.7v zeners, so I used a 3mm blue LED + 1n4148 / 1k resistor to get a 3.5v vref.

TBH, I started off by using the fixed stages in the phase path, the effect was quite subtle, much less than I expected. As Causality 4 has 2 fixed stages at it's heart I was surprised how little adding 4 fixed stages in the phase path added. However, 3 fixed stages in the regen path sounds great.

I'm trying to figure out what's actually happening. I guess I'm filtering the regen in some way. As well as adding 270^o of shift in the regen path. Any good explanation would be great - Mark? Stephen?

I've got two of these Borg Implant boards to play with (thanks SC!), so I'll make up the other one and continue to experiment.........more results pending.

[Gurner]

Humour me here.

In the absence of a schematic (of how it 'slots' together)...what role does the above layout play.....4 fixed stagees? (I know what a stage is, I know how a phaser works, but wondering what role a 'fixed' stage has to play?!)

[bside2234]

I thought Mark explained it in the first post. I'll have to go back a read it again but that's where I thought I got my understanding of how the fixed stages worked.

[Gurner]

Indeed he did (I've a bad habit of just picking up the last few posts in a thread...!!!)

Hmmm...still not locking into this one....if you apply four stages of 'static' phase shift, with some frequencies, you could very well be wrapping them around to where you started out? (ie a full 360 = 0 degrees phase shift.....particularly  wrt higher frequencies)...surely all it's gonna supply is a 'phase offset' from where the modulated stages start.....but like I, since these will wrap around & could be back near zero shift....not understanding fully the 'win' here?

[frequencycentral]

^^^

Each stage introduces a small delay inherantly, so even if you did end up with 360^o, the shifted signal will be delayed with respect to the dry signal. You can't ever go back, only forward.

[slacker]

But when the FET is driven hard on then the + input basically is shorted to ground?? I got this:
http://s81.photobucket.com/albums/j207/merlinblencowe/?action=view&current=CIMG5703.mp4
I'm not seeing things right, that is swinging from zero to 180 degrees?

That's what will happen if you drive the Fet from hard on to off, like you said when it's on the + input is shorted to ground so you've just got an inverter. When it's off, so it's some huge resistance, the corner frequency of the RC filter is so low that the you've just got a non inverting buffer. That's what the simulator says anyway
In a phaser the fet or whatever you use would sweep between some limited range of resistances to put the corner frequency somewhere in a useful range.

[Gurner]

^^^

Each stage introduces a small delay inherantly, so even if you did end up with 360^o, the shifted signal will be delayed with respect to the dry signal. You can't ever go back, only forward.

Once you get past 360 degrees wrt phase shift, you're starting again wrt the dry signal ...there's no such thing as 361 degrees. (361 degrees = 1 degree of phase shift or in other words you're right back where you started)....so you definitely do go back. In other words, adding more stages is counter intuitive eg 365 degrees of phase shift is not more delay at all, it's just a mere 5 degrees phase delay wrt the dry signal.  

Additionally, once you get past 180 degrees of phase shift, you are essentially  'leading' the dry signal not delaying at all. (therefore 358 degrees of phase lag shift...is nothing more than a 2 degree phase 'lead')

In a phaser the fet or whatever you use would sweep between some limited range of resistances to put the corner frequency somewhere in a useful range.

No, the sweep isn't to put it in a useful range (that comes from the choice of capacitance & resistance in the all pass). Sweeping is done to move the phase shift & thereby make the overall effect interesting (because a static phase shift just sounds relatively dull by comparison)

[frequencycentral]

360^o of shift is one full cycle delayed with respect to the dry signal. It may look the same on a 'scope when analysing sine waves, but with a harmonically complex guitar signal.........

....and how can opamps possibly perform the time travel back into the past that would be neccesary for a 360^o shifted signal (and therefore delayed by one full cycle) to occupy the same time period as the dry signal?

[slacker]

No, the sweep isn't to put it in a useful range (that comes from the choice of capacitance & resistance in the all pass). Sweeping is done to make the overall effect interesting (because a static phase shift just sounds relatively dull by comparison)

That's not what I said
What I meant was that you need to sweep the resistance over the right range of resistances, and not from some huge resistance to a tiny one like I thought Merlinb was doing. If the corner frequency, and therefore the notches, spends most of it's time outside the frequencies generated by the guitar you aren't going to get much effect however much you sweep it.