PWM phaser in the works

[psychedelicfish]

Sorry, no LED rate indicator  It's because LED needs to be buffered and inverted..

Why not run it off the PWM signal? It would probably benefit from a BJT buffer, but that would still only take 3 extra components.

[mykaitch]

So what am I missing?
PWM - pulse width modulation.
The analogue switches have two states, on or off. They turn on with about 3v on the enable so what is this triangle stuff all about?
In a phaser with FETS the ramp will ramp the FETs and hence the op amps.
Can't see where/why PWM is any use. We need a ramp or triangle, I thought. Won't just banging the ref volts up and down produce a sort of
clicky sound? Sorry, I don't geddit.

[anotherjim]

The analog switches are acting as variable resistors. The resistance they present is proportional to the duty cycle of the switching control. PWM is varying the pulse width of the switch control and hence varying the average resistance through the switch. A long On time compared to a short Off time amounts to a low switch resistance - A short On time compared to a long Off time is a high switch resistance. The switching frequency is well above the audio range so you don't hear them switch. The triangle LFO is modulating the pulse width of the high frequency clock oscillator.

[Freppo]

I suck at explaining tech stuff. But I'll try... It's really simple once you understand it.

It's true that the switches are either on or off. But, take for example PWM controlling the brightness of an LED:
The flow of current to the LED turns on/off rapidly in pulses, giving us the illusion that the LED is constantly on.
Narrow pulses gives us a lower current over time then wide pulses... so the LED gets stronger with wider pulses, since it's more "on"

This is exactly how it works with the CD4066 switch.. it turns on/off very rapidly and gives you the same effect as a resistance
So the pulse width determines how long the switch is open (so narrow pulses=more "resistance" then wide pulses)

Remember that these pulses are very often fast (over the audiable range 20K +)
And for example 20 000 hz is 20 000 swings per second (please correct me if i'm wrong)

In this circuit the CD40106 makes a pulse width oscillator (with very high frequency, about 90K)
It also works as a narrow pulse widener, modulated by a triangle wave LFO.

/ Freppo

[Freppo]

The analog switches are acting as variable resistors. The resistance they present is proportional to the duty cycle of the switching control. PWM is varying the pulse width of the switch control and hence varying the average resistance through the switch. A long On time compared to a short Off time amounts to a low switch resistance - A short On time compared to a long Off time is a high switch resistance. The switching frequency is well above the audio range so you don't hear them switch. The triangle LFO is modulating the pulse width of the high frequency clock oscillator.

Anotherjim beat me to it Much better explanation then I could ever give. Cheers!

[Freppo]

Why not run it off the PWM signal? It would probably benefit from a BJT buffer, but that would still only take 3 extra components.

Yes, but I couldn't find room for there extra 3 components without making the PCB larger.   
But running it directly from the PWM signal is a nice idea and should only require a simple RC filter. I'll try that.

cheers
/ Freppo

[Cotoletta]

Nice idea. I'm having some issues with a JFET based phaser, way too much distortion using humbuckers. Does your design solve this problem?

[psychedelicfish]

Why not run it off the PWM signal? It would probably benefit from a BJT buffer, but that would still only take 3 extra components.

Yes, but I couldn't find room for there extra 3 components without making the PCB larger.   
But running it directly from the PWM signal is a nice idea and should only require a simple RC filter. I'll try that.

cheers
/ Freppo

You don't even need the RC filter, persistence of vision will take care of any need for filtering, especially if your PWM is running at 90kHz.

[Freppo]

Nice idea. I'm having some issues with a JFET based phaser, way too much distortion using humbuckers. Does your design solve this problem?

This design doesn't have any distortion, even when using hot humbuckers.
Sorry to hear you are having problems and I hope you can solve it!
I had the same problem with a Phase 90 that I built on a tonepad PCB.

/ Freppo

[Freppo]

You don't even need the RC filter, persistence of vision will take care of any need for filtering, especially if your PWM is running at 90kHz.

Tried it now. It kinda works directly from the PWM signal, but with a limited range. It never turns off completely.
With a filter in front it goes from off to fully lit, making it much more visable and usable. And it's only two components extra

I made a new PCB with the added LED stuff.
http://parasitstudio.weebly.com/uploads/2/4/4/9/2449159/parasite_phaser_rev1_pcb.pdf

cheers / Freppo

[Cotoletta]

Nice idea. I'm having some issues with a JFET based phaser, way too much distortion using humbuckers. Does your design solve this problem?

This design doesn't have any distortion, even when using hot humbuckers.
Sorry to hear you are having problems and I hope you can solve it!
I had the same problem with a Phase 90 that I built on a tonepad PCB.

/ Freppo

Good to hear! I have a bunch of 4066, tomorrow I'm grabbing some 40106 and give it a try.

[armdnrdy]

Great job!

I have a question.

I know that your goal was to keep this design low parts count and light on controls but...

Is there any reason why you didn't include a depth control?

A depth control will give you the option of dialing back the effect which is especially useful in the upper speed range.

One can always dial back the feedback but...that is a different parameter.

The feedback path reinforces the notches...the depth or width control adjusts the range (amplitude) of modulation from a narrow to wide wave.

If I'm looking at your schematic correctly...it would only take the inclusion of a pot (lug 3) at the output of IC5B, the wiper connected to the 4.7K resistor, and lug 1 connected to ground or Vr.

[mykaitch]

Weird. Analogue switches do what they say on the tin. Look at the data sheet. It may be that they could work as var R in the same way that a 7400 for example can work as an op-amp, but I have to ask why???

[Freppo]

Great job!
Is there any reason why you didn't include a depth control?
If I'm looking at your schematic correctly...it would only take the inclusion of a pot (lug 3) at the output of IC5B, the wiper connected to the 4.7K resistor, and lug 1 connected to ground or Vr.

Thanks! I didn't include a depth control because I don't think that it is really necessary. The effect is subtle enough until you turn up the feedback.
Also, the filter at the output of the LFO smooths out the waveshape, reducing the depth and keeps it from sounding wobbly or unmusical at high rates.

But you could always add a depth pot if you wanted to, just as you discribed.

cheers
/ Freppo

[anotherjim]

I think the point of using CMOS switches as variable resistors is because the multiple stages will track each other very closely and it is cheap.

The alternatives are to use JFET's as variable resistors or Vactrols BUT both types need to be closely matched in order to get decent tracking. Very expensive and tedious to do. If you don't have close tracking of the phase shift stages, the effect is less noticeable than it should be.

This diagram...

Shows a VCA using 2 switches forming a potentiometer. With the clock pulse width at 50% it's dividing the input by 2. The inverter between the 2 switch controls means that as the PWM pulse-width increases (but frequency remains constant), one switch increases resistance as the other decreases. In this case, the switches are directly in the audio path, so an output low pass filter is shown to remove the clock frequency.

If I ever get around to building a Phaser, I will definitely try this scheme.
Although I do have a bunch of 2SK30A JFETS that were long ago pulled from a dead Roland Jet Phaser. Hmmmm.....

[Freppo]

My goal with my phaser was to avoid FETs and Vactrols for the reasons you discribed.
I have pulled my phaser off the breadboard now. I want to try other things for a while.

But soundclip / demo is coming soonish. Maybe I'll do a vero layout aswell..

BTW, That's a neat schem!

[Cotoletta]

Anyone tried to build this? Just built mine for the second time and I'm having a strange distorted sound, no phasing at all. Checked every connection, seems all good. I can hear the LFO ticking if I disconnect from ground the unused triggers inputs so I know it works.

[Freppo]

Sorry you hear you have problems.
I built it and it works  fine. I wouldn't share anything I didn't test first.

Can you post IC voltages?

[Cotoletta]

Sorry you hear you have problems.
I built it and it works  fine. I wouldn't share anything I didn't test first.

Can you post IC voltages?

I know your circuit is tested (well, we have also a video!), maybe I just did something wrong   I also re-did the PWM part of the circuit and now I have a clean signal, no strange noises.
Also if I connect from the pin 8 of the 40106 instead of the 9, I have a slightly volume boost but and a little distortion but still no phasing.
I'll post some voltages.
Note that I have a few differences from your schematic: didn't have a TL022 so I used a LM359 or TL072, but same result. Used two TL072 instead of one TL074.

Voltages:

CD40106 (signal to 4066 from pin 8 )
1  GND
2 4,49V
3 GND
4 4,49V
5 2,76V
6 0,14V
8 0,001V
9-10 4,49V
11 1,13V
12 4,49V
13 0,145V

HCF4066:
1 4,49V
2 4,49V
3 4,49V
4 4,49V
5 4,49V
6 4,49V
7 GND
8 4,49V
9 4,49V
10 4,49V
11 4,49V
12 4,49V
13 4,49V
14 4,49V

(looks suspicious?)

Input/output TL072 buffer:
1 4,49V
2 4,49V
3 4,02V
4 GND
5 4,49V
6 4,49V
7 4,37V

The phasing stages I think are fine, tried connecting some JFETS (with an LFO) instead of the 4066 and I have phasing.

[Freppo]

I'll check my voltages later today, but those 4066 values look a bit strange.
Pin 14 of the cd4066 should be connected to +9v

/ Freppo