I Apologize. This Is A Stupid Idea, Not Worth Reading, Vibrato to Pitch Shifter?

[Merlin]

I tried it in excel. Original sine and cosine signals in red and blue (with a DC offset just so they are easy to see on the graph).
One cycle of the original frequency is equal to exactly two divisions on the vertical grid lines.
It's hard to see because the change is small, but it's there

[jeff]

The only signal we have is 440hz 0° 440hz 90° and an LFO signal.
The ratio of 0°:90° = phase of the 440hz wave.
Advancing/retarding the phase as the note is oscillating at 440hz is what causes the pitch to shift.


Imagine this:
You have a slider with 440hz@0° at one end 440hz@90° at the far end.
Any point on that slider IS 440hz at some phase angle between 0° and 90°.


Say you took a piece of wood and cut a sine wave groove into it, so that the slider would fit into the groove.
Now say you slid that piece of wood left to right across the slider, so the slider follows the groove.


That's mechanical vibrato!
That's what the mix tube/LFO does.
So the LFO is the Ratio of the mix(phase) over time.


Notice the sine wave. Nice and curvy. It's not a straight line.
If we focus on one half cycle of the wave, the part where it is at max positive sloping down to max negative. That's the speed of the ratio(phase) over time. Again, it's NOT a straight line. The mix does not change lineraly over time.


What does this mean?
If you start with the slider at the 0° end, and slide your grooved wood across, the slider would: Accelerate down as the wave starts getting faster n faster, crossing the middle(0v where the sine wave is it's straightest) would be the fastest speed, as the slider pushes to 90°, the waved groove starts to curve, decelerating.


This results in 440hz, getting lower and lower at the top of the curve, hitting its lowest pitch in the middle(where the groove is straightest), then rising and rising back to 440hz at 90°.


The next half cycle, from 90° back to 0° is the reverse. That's the half of the LFO that goes most negative to most positive. Follow that part of the curve. Acceleration, max speed, deceleration as the curve approaches max.
BUT....
Where as the LFO crossed the midpoint in the 0° to 90° half of the wave the note was lower than 440hz, when it crossed the midpoint in the 90° to 0° half of the LFO the note was HIGHer that 440hz.


This smooth rising and falling is caused by the sine wave accelerating, decelerating, accelerating, decelerating.


So what would happen if we took our wooden sine wave grooved guide and used a straight zig-zag groove instead?
What if we used a 'magic' groove?
What if our groove were a series of downward diagonal groves, where, when the slider hit bottom, it instantly teleported the slider to the top of the next diagonal, downward groove again?


Does that make sense?

[tubeswell]

The different-sized fonts are hurting my eyes. And the long-winded analogies are difficult to follow.



Can you please condense your explanations and provide a schematic?

[jeff]

TL:DR
Mix a 0° with 90° signal = phase change = pitch change
Use a sine wave LFO = nice vibrato
Use a triangle wave = jump in pitch/ up, down, up, down
Use a triangle wave AND alternate 0° with 180° as that mixer is faded out, alternate 90° with 270° as it's mixer is faded out = pitch shift.
Cycle 0° 90° 180° 270° 0° =pitch down
Flip switch to swap 90° and 270°, to cycle 0° 270° 180° 90° 0° = pitch up


Need parts, working on build/schematic.

[jeff]

The idea.
Probably not the best way
Maybe won’t even work
But, that’s the idea


EDIT: Oops the LFOs should be 90° out of phase for each tube, not 180°

[HotBluePlates]

... With the Vox/Wurlitzer circuit, as you wiggle between the two signals 0° 90° you are either bringing up the leading or lagging phase. ...

... The Wurlitzer (and Vox) vibrato circuits use 2 signals 180º out of phase, but then they're bridged through R-C networks.  The resultant is something in-between, which is about 90º phase-shift. ...

... When the LFO is off, You have a 50/50 mix of 440hz at 0° and 90° and you hear 440hz.
As the mix goes to 60/40 ...

My last input on the topic:

A snippet about the Wurlitzer Vibrato from the book Electronic Musical Instruments by Richard Dorf.  Vox lifted this circuit nearly 100% to make their Vibravox circuit (and Vib/Trem channel of AC15 and AC30 amps).

   - The "phase modulator" V3 mixes the 0º and 180º signals, who themselves have been applied to R-C networks.
   - The result of phase-split, R-C shifts, and V3-mix is a constant 90º phase shift, but only as long as the LFO is off.
   - This phase-change is not perceived by the listener.

   - The LFO is also applied to a phase-splitter.
   - When the LFO is activated, one side then the other side of V3 is the dominant signal.
   - The changing-dominance shifts the absolute-phase from 90º towards 135º, then back to 90º, then towards 45º, then back to 90º.
   - Changing-phase is the only time of pitch-shift; this circuit cannot yield a constant pitch-shift in the manner of a POG, Harmonizer, Whammy pedal, etc.

[jeff]

I tried it in excel. Original sine and cosine signals in red and blue (with a DC offset just so they are easy to see on the graph).
One cycle of the original frequency is equal to exactly two divisions on the vertical grid lines.
It's hard to see because the change is small, but it's there

Just saw this, thank you.
I think it went from page 1 to page 2 and I didn't see it.

[jeff]

   - The "phase modulator" V3 mixes the 0º and 180º signals, who themselves have been applied to R-C networks.
   - The result of phase-split, R-C shifts, and V3-mix is a constant 90º phase shift, but only as long as the LFO is off.
   - This phase-change is not perceived by the listener.

 

After the phase inverter you HAVE two signals that are 180° out of phase with each other.
Send those two signals through the R-C you have two signals that are 90° out of phase.


The purpose of the filter is to input two signals 180° out of phase with each other and output two signals 90° out of phase with each other.


Right? What am I missing?


.       180°
.  /-----------\      /--------
--PI.            Filter.  ^90°v out of phase with each other
   \-----------/      \--------
.         0°


I guess I don't understand what you're trying to teach me.
Are you saying the 2 signals going to the mix tube are 180° out of phase with each other?
If so, we wouldn't need a filter, just hook the PIs outputs straight to the mixer.
The filter inputs two 180° out of phase signal, and outputs two 90° out of phase signals.


"The result of phase-split, R-C shifts, and V3-mix is a constant 90º phase shift,"

[jeff]

   - The changing-dominance shifts the absolute-phase from 90º towards 135º, then back to 90º, then towards 45º, then back to 90º.
   

So when one turns on, and the other off, we're at 135°
When that turns off, and the first is on we're at 45°
135°-45°=90°
The signals are 90° apart.
No?

[jeff]

I tried it in excel. Original sine and cosine signals in red and blue (with a DC offset just so they are easy to see on the graph).
One cycle of the original frequency is equal to exactly two divisions on the vertical grid lines.
It's hard to see because the change is small, but it's there

Merlin, did you make that graph?
What frequencies did you use?
I'd like to see that with a 440hz input and 8hz LFO.
Betcha you'd see the stretch, compress even better.

[jeff]

I think I got mixed up with what phase goes where.
This is what I’m going to build.
I’ll let you know how it goes


EDIT: Bad Idea
I shouldn't've tried to whip up a block diagram with a half-baked idea.
This is not the solution, but I'm getting there.

[jeff]

OK, I think I've got it this time. But I've learned my lesson, so I'm going to build it first and test it before I release another block diagram.

[HotBluePlates]

... I'm going to build it first and test it ...

This is perhaps the best option.  I suspect you will get something, though not exactly what you anticipated.

Right? What am I missing?

.       180°
.  /-----------\      /--------
--PI.            Filter.  ^90°v out of phase with each other
   \-----------/      \--------
.         0°

I guess I don't understand what you're trying to teach me.

The part you drew is sorta-right, but only after the 2 filter-outputs are summed in a differential amplifier (that responds to the difference of its inputs).

The LFO signal is also phase-split, and also applied to the same Diff-Amp:

                                       LFO +
.       180°                          |
.  /-----------\      /--------\  |         
--PI.            Filter.  ^90°.  Diff Amp ----  Absolute-phase 90º
   \-----------/      \--------/  |       
.         0°                            |
                                        LFO -


Read the article & check out its schematic-fragments.

   - Only when the inputs to the Diff-Amp are equal-amplitude will the Output Phase be 90º.
   - The 2 opposite-polarity LFO signals cause one signal, then the other to be stronger (not like a switch, but smoothly-varying).
   - When a side becomes the stronger-signal, the phase of the Diff-Amp's output moves closer to the stronger-side.
   - When the Diff-Amp's phase is changing to move towards 45º absolute-phase, the pitch will rise.
   - When the Diff-Amp's phase is changing to move towards 135º absolute-phase, the pitch will fall.

   - The changing-dominance shifts the absolute-phase from 90º towards 135º, then back to 90º, then towards 45º, then back to 90º.

So when one turns on, and the other off, we're at 135°
When that turns off, and the first is on we're at 45°
135°-45°=90°
The signals are 90° apart.
No?

This Confirmation Bias is killing me: you're stuck on the idea of "90º" so everything looks like it reveals "90º."
  - Read the article, and Dorf says the R-C networks are such that the phase-change due to the LFO signal may only reach about 45º in either direction, even though the 2 audio signals applied through the R-C networks are 180º apart.
  - That "90º change from 45º either-direction" is not certain:  it could be "30º change" but it could also be "70º change."


There is no vibrato until the output of the Diff Amp (you pause before its input).  There is 1 Output Signal from the Diff-Amp, and it's absolute-phase goes:
90º -> 81º -> 72º -> 63º -> 54º -> 45º -> 54º -> 63º -> 72º -> 81º -> 90º -> 99º -> 108º -> 117º -> 126º -> 135º -> 126º -> 117º -> 108º -> 99º -> 90º

[jeff]

First off,
I respect the heck out of you. Back in the day, You, PRR, Sluckey, Kagliostro, etc. taught me everything I know.


Second I don't think I disagree with you, I think I'm trying to say the same thing.
When I say 90°, what I mean is NOT 90° out of phase with the output, but that the two signals are at a phase angle difference of 90°


When I say 90° out of phase, I'm referring to fig. 2
https://www.one-electron.com/Archives/Radio/RadioSSB/Dome%201946%20Wideband%20Phase%20Shift%20Networks.pdf


The two lines 0₁ 0₂ show the phase at each frequency after the two filters and the line 0₂-0₁ shows the phase between the two. ≈90° difference between the two signals as compared to each other.


When I say 90° I'm referring to the block diagram on the Dorf paper. Signal 1 is a sine wave. Signal 2 is NOT 180° out of phase with signal 1 it's 90° out of phase with signal 1.


The phase at the output of the filters is all over the place compared to the original signal, but the goal is to feed one side of V3 with a signal and the other side with a signal that is 90° out of phase at every frequency.
That's why I'm stuck on 90°
That's all I was trying to say.


At frequency X after the filters, one may be 20° the other 110° with the original wave but they're 90° out of phase with each other 0₂-0₁.

This Confirmation Bias is killing me: you're stuck on the idea of "90º" so everything looks like it reveals "90º."
  - Read the article, and Dorf says the R-C networks are such that the phase-change due to the LFO signal may only reach about 45º in either direction, even though the 2 audio signals applied through the R-C networks are 180º apart.
  - That "90º change from 45º either-direction" is not certain:  it could be "30º change" but it could also be "70º change."


There is no vibrato until the output of the Diff Amp (you pause before its input).  There is 1 Output Signal from the Diff-Amp, and it's absolute-phase goes:
90º -> 81º -> 72º -> 63º -> 54º -> 45º -> 54º -> 63º -> 72º -> 81º -> 90º -> 99º -> 108º -> 117º -> 126º -> 135º -> 126º -> 117º -> 108º -> 99º -> 90º

Isn't the Dorf article saying that, yes, the input to the filters are 180° apart... But even though that's true, after the filters the difference is only 90°, and therefore, if they're mixed 50/50 you can only ever go up or down 45°?


I get what you're saying

  - That "90º change from 45º either-direction" is not certain:  it could be "30º change" but it could also be "70º change."

Intensity low wiggle 90°-85°-90°-95°-90°
Intensity mid wiggle 90°-115°-90°-75°-90°
But max intensity can only ever be 90°-135°-90°-45°-90° because the two signals we're are 90° out of phase with respect to each other. You Cannot wiggle 90°-140°-90°-40°-90° because the two input signals are ONLY 90° with respect to each other.


Yes, the PI splits the signal and you get two signals 180° apart, but after the filter, the two output signals you send to V3 are 90° apart with respect to each other.
Is that Right?
Are we talking about the same thing here, or am I way off?

[jeff]

Do we see eye to eye on this at least?:


Take a 440hz signal at phase X°
Make a copy of that signal at phase (X±90)°
Sick those in V3, turn on LFO, you get vibrato.


Take a 440hz signal at phase X°
Make a copy of that signal at (X±180)°
Stick those in V3, turn on LFO, you get cancelation(just like the LFO signals, which are also 180° out of phase).


Do we at least agree on that?

[jeff]

Pg 1

[jeff]

Pg 2

[jeff]

Pg 3

[jeff]

Pg 4

[jeff]

Pg 5

[jeff]

Pg  6

[jeff]

Pg 7

[jeff]

This CANNOT be achieved with the Vox circuit alone!


But you can achieve this result, if you duplicate/modify elements of the Vox circuit.
That's what I'm working on.

[jeff]

Prototype schematic

[jeff]

Sorry for wasting everyone's time.
I'm a big dummy and it is only now that I see the error of my ways. I thought if we can bend a note above 440hz and bend a note below 440hz that we could use that to keep the note bent lower.


There is a way, but this ain't it.
I'm sorry.
You tried to tell me, but I couldn't understand it.


I realize this would just make a crappy vibrato, one that ONLY bends the note down and back, down and back.


I feel like a boob and I'm sorry.


I sincerely apologize for wasting your time
Thank you for your help, I was too stubborn because I really wanted this to work.

Sorry
Jeff

[HotBluePlates]

... I thought if we can bend a note above 440hz and bend a note below 440hz that we could use that to keep the note bent lower.

There is a way, but this ain't it. ...

You're okay.  You were just excited by an idea.

I know there are old tube-based circuits that do waveform synthesis.  I strongly suspect the fixed pitch-shifting you want to do is most commonly done with digital waveform synthesis to achieve the pitch-shift (though I confess to not having chased the schematics to know that for sure).

[Willabe]

You're okay.  You were just excited by an idea.

Yep.

And to me true pitch shift vibrato is the stuff! I can see why it would get you excited.