PWM phaser in the works

[Cotoletta]

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

Uh I'm a dumbass haha. I was powering both 40106 and 4066 from the +VR line.
Corrected this I still have no phasing. If I feed the 4066 from pin 8 of the 40106 again I have a volume boost but this time no distortion.

*EDIT x2: everything works. Sounds great!

[mykaitch]

I really want to get my head around what you're doing because I've just built the phase90.
Your scope pic shows what I would expect,sq waves.
The anal sw needs a logic level to control it and turns on or off.
In the sch it seems that a set voltage is switched.
SO...PWM what???

In the orig p90 there is a simple osc and a charge cap.The cap will charge in a crude sort of ramp.
When the gates to the FETS are turned on by this ramp they will exhibit a changing resistance and this
'sweep' the opamps. We can see all this with a scope.

So my Q still is, just what is being controlled by means of PWM because ultimately those cmos switches will only turn on or off.
Please don't misunderstand me, I want to know what I am not seeing here.
You have to have some sort of linear control to sweep the phase. If you just switch it is only going to be in phase or out,albeit changing at the LFO rate, and probaby clicking....

[Freppo]

*EDIT x2: everything works. Sounds great!

Awesome! I'm glad to hear you got it working.

Please don't misunderstand me, I want to know what I am not seeing here.

I think you are a bit confused about how PWM works. The pulse oscillator is not to be confused with an LFO.
The PWM signal turns the switches on/off at a very high rate (about 90 000 times per second in this case)
That is why you don't hear any clicking. It's above the range of our hearing and what our equipment can reproduce.
The video demo I did in my first post shows the PWM signal very "zoomed in", at something like 0.002 second.

The width (duty cycle) of the pulse determines how much time of every cycle the switch stays in either state (on or off).
So with a narrow pulse the switch is mostly closed each cycle, which translates into a high resistance.
And a wide pulse keeps the switch mostly open each cycle, which gives us a very low resistance.
Remember that each cycle is repeated 90 000 times per second, but the effect of changing the pulse width adds up "over time".
This way it's possible to go inbetween fully closed (infinite resistance) to open (only a few ohm's of resistance) using PWM.

Check out this video
https://www.youtube.com/watch?v=mVx02s1fHIY

The width of the pulse however is modulated with a regular LFO.
If I scoped the LFO output insted we would see a triangle wave at only a few hz per second.

I hope that helps. Maybe someone else can chime in a explain this better then I can.  
Cheers / Freppo

[Cotoletta]

Hey Freppo,
Just added two extra phasing stages and sounds all cool.
One question: I think the phasing is a little bit uneven. Looks like the PWM signal goes to a lower duty cycle too fast, as can be seen your videoclip. Any plan for modifying this?

Btw great circuit. Sound very similar to the phase 90 clone I have in my other breadboard without the distortion issue.

[Freppo]

One question: I think the phasing is a little bit uneven. Looks like the PWM signal goes to a lower duty cycle too fast, as can be seen your videoclip. Any plan for modifying this?

Cool. I'm glad you like it
The latest schematic doesn't have the same LFO as my first video, so the sweep should be better now.
But you can try to offset the Vref of the LFO to change how much time it spends at either ends of each sweep.
Try a 20K trimpot between +9v and ground, insted of the 10K and 12K resistors coming for the LFO IC.

Here is the latest schematic again, in case you missed it:
http://parasitstudio.weebly.com/uploads/2/4/4/9/2449159/parasite-phaser-final.pdf
The feedback pot numberings are backwards and I haven't added the LED rate indicator stuff yet..
It's a 22K resistor from the CD40106 pin 9 that goes to the LED anode and to a 100nF cap to ground.
LED cathode to ground.

Cheers / Freppo

[Cotoletta]

Cool. I'm glad you like it
The latest schematic doesn't have the same LFO as my first video, so the sweep should be better now.
But you can try to offset the Vref of the LFO to change how much time it spends at either ends of each sweep.
Try a 20K trimpot between +9v and ground, insted of the 10K and 12K resistors coming for the LFO IC.

Here is the latest schematic again, in case you missed it:
http://parasitstudio.weebly.com/uploads/2/4/4/9/2449159/parasite-phaser-final.pdf
The feedback pot numberings are backwards and I haven't added the LED rate indicator stuff yet..
It's a 22K resistor from the CD40106 pin 9 that goes to the LED anode and to a 100nF cap to ground.
LED cathode to ground.

Cheers / Freppo

Yap, that's the schematic I used.
Just tried the 22K trimmer, it is possible now to adjust how much time the duty cycle stays at higher values, but the transition to low duty cycle looks like the same.

[mykaitch]

The PWM signal turns the switches on/off at a very high rate (about 90 000 times per second in this case)
That is why you don't hear any clicking. It's above the range of our hearing and what our equipment can reproduce.
The video demo I did in my first post shows the PWM signal very "zoomed in", at something like 0.002 second.

Thanks for the reply, yeah I figured you might be doing that, obviously you need to switch at a much higher rate than your bandwidth -- and you do. I have two ICs on my board, I think I might use a PIC out of interest and hook it on to my board to see how it goes. You have more board to cover (unless its SMD) for the sake of 4 fets but then, you have a lot more scope. With a PIC I should be able to do all kinds of stuff 

I use PWM in some of my lighting effects but to avoid FETs etc I use digital pots and let the software sort it out.
Good luck with the project.

[Freppo]

I finally boxed up this one!  



At about 3:25 I switch "mode", which makes it sound very different. Sorry about my crappy playing.
I think it turned out good, and I probably won't tinker with this one much more
But if I am building this one again I'm going to reduce the maximum speed (raising the 330K and 1.5 resistors),
or at least use a log pot to have more control over the slow rates.
Other then that I'm pretty happy. I hope you like it!

You can find the latest PCB layout and schematic at my blog:
http://parasitstudio.weebly.com/building-blog/parasite-phaser

Cheers / Freppo

[anotherjim]

Bravo! Sounds just like a Phaser

My opinion is any LFO faster than 8Hz sounds silly with most modulation effects. My own personal taste is for about 7Hz tops.

I've never experimented with phasers - is there a limit on how slow the LFO can go before the effect disappears?

[Kipper4]

Just started mine this morning. Should have it done today
Thanks for the clip Fredrik. Sound excellant.
Excited or what!!!!!!!!!!

[Freppo]

Bravo! Sounds just like a Phaser

Thanks, it does.. All those hours and just regular phaser sounds. feels alittle underwhelming lol.
But it's nice to make something really useful for a change and it does sound nice.  
Now i'm back experimenting with synthy noisy stuff, it's much more fun!

is there a limit on how slow the LFO can go before the effect disappears?

Not really. It can get extremely slow, but the sweep gets wierd. I think a sinewave or hypertriangular LFO should suit better for a slow sweep phaser.

Btw.. If anyone wants to slow down the overall range of the LFO try changing the 330K resistor to a 470K.

Just started mine this morning. Should have it done today
Thanks for the clip Fredrik. Sound excellant.

I'm glad you like it! Good luck with your build.

[bluebunny]

Cool demo, Fredrik.  I may have said this was on my build list.  I think it moved up a notch or two!

[Kipper4]

Mine's done. Thanks for all the hours Fredrik. It was worth it because now we can make phaser without having to match fets.
I used a 50k trimmer with a 47k in parallel because that's all I had. There's a very narrow margin for biasing as I expected.
I redrew Frederiks layout for perfboard. Now verified. So if you want the perf version pm me.
Thanks again good work
Rich

[anotherjim]

"All those hours and just regular phaser sounds. feels a little underwhelming lol."
That may be, but you've helped to publicise an alternative method, and that's all to the good.

[Kipper4]

I just did the LFO range mod by putting a 150k in series with the 330k on a spdt switch so I can have either the 330k alone or the 330+150K.
lots of room on the board for the mod and
No popping issues with the switch. sounds good.
Thanks Freppo

[Jem]

Very cool and innovative use of PWM - thanks for sharing!

It would be tempting to use the PWM in a small PIC to generate the sweep and add a bunch of waveforms and tap tempo.

Just for fun I was also wondering if you could swap the op amps for CMOS inverters biased for linear operation and have an all CMOS phaser - is this feasible?

[anotherjim]

Not really possible with inverters such as they are. A phase shift network uses 2 paths and they are either fed into a differential amp (the opamps here) or produced from a transistor phase splitter (like Uni-vibe). The CMOS inverter can't do either trick.

If you really wanted (for the hell of it!) to use CMOS, the MOSFETs in a CD4007 might be able to make a uni-vibe type phase shifter, possibly  getting you at least 2 stages out of one 14pin IC!

[snap]

Not really possible with inverters such as they are. A phase shift network uses 2 paths and they are either fed into a differential amp (the opamps here) or produced from a transistor phase splitter (like Uni-vibe). The CMOS inverter can't do either trick.

If you really wanted (for the hell of it!) to use CMOS, the MOSFETs in a CD4007 might be able to make a uni-vibe type phase shifter, possibly  getting you at least 2 stages out of one 14pin IC!

you might as well take a look at this topic: http://www.diystompboxes.com/smfforum/index.php?topic=87949.0

[anotherjim]

Doh! Silly me. Of course you can split phases with an inverter - input and output.
For that matter, you can make a diff amp with 2 inverters and some resistors (invert one input and mix).

You still get the usual hit's with inverters in linear mode. High current consumption and noise.
I go back to mention the CD4007. There may only be 3 inverters, but you can insert some resistance in series with each of the complementary pairs to limit the current draw.

[Kipper4]

Hey Freppo
Where should I hook up the oscope to see what you had in the video please.