go to Japanese Version
Phase Shifter

Is phase shifter somberness?

I had been had a somberness impression to phase shifter. Repeat the same circuit make me bore and the sound coming from is also smoky. Chorus / Flanger may match to me, I had been thinking like this.
But one day, a friend of mine living US . send me a message.
He said that "Here comes the explosion of the power makin' something!". and he build phase shifter on his bread board and send me some sound clips. It's great. and this blow my mind. My understandings about phase shifter is absolutely wrong.

phase shifter made from single stage phase shifter. blending the original sound and phase shifted sound make notch filter which has some peaks and dips. So we may get the same effects by series of simple notch filter . I will try this later.

When we are making sounds, if we throw away high frequency by Voltage controlled LPF, then it's going to be mild sound. I don't want to throw away high frequency in order to get more hard sound. But HPF will make my sound jellyfish.
So I will use phase shifter as the filter or resonator to add some nuance to the sound. this is the goal of this projects.

Basic Schematics

Basic Methods
This is basic methods for phase shifter. it has non inverting amp with gain 2 and inVerting amp with gain 1. Each one has LPF and HPF. both out put will mix in a op amp so there is no volume change but there is shift of their phase. It shift from 0 to 180 degree by DC to infinity frequency. We can use it as a black box.
For example, if you put the 1k of signal to the shifter which set the center of the frequency to 1k (it's almost same as the cut off point of the RC filter), you can get the signal which 90 degree delayed signal.
More lower frequency will delay less. 550 Hz will delay 60 degree. And the other hand, 1.8k Hz will delay 120 degree and 3.6k Hz will delay 150 degree.
Now, Let's use several shifter at a time in series. 1 stage of it can shift the signals 180 degree. so 4 stage increases 4 times up . 90 degree goes to 360 degree. 45 degree goes to 180 degree. If you add the original signal and shifted signal, then the wave that delay 180 degree will denies original wave. this will get the peak and dip.
I will simulation by the software.

Hand made photo coupler Red #1 - #4

I see a red door and I want it painted black (by Rolling Stones)
Mr. Scott use photo coupler (Vectol) as the dynamic resistor for the shifter. It's so cheap and easy to get device at his place. He send couple of them to me. But it's short to build the phase shifter. I try to get them in my local place. But it not cheap enough to build them. but CdS cell is not so expensive. So I try to make home brewed one.
I go 1 dollar shop to get Christmas Tree light made from red LED, and join them by heat shrink tube.
When I check it them, I saw a red light and I want it painted black.
I made easy V-I converter on the breadboard and make small experiments about the linearity of hand made photo coupler

Raw data of hand made red LED photo coupler
At first I get just data only and plot it so normally. I can see that it so sensitive .But I can't see other than this.
the center frequency of phase shifter can calculate by this formula.

and I notice to R and I change the form of it like this.

It shows me that the frequency and 1/R has the relation in liner. iRene shows me about this by mail. thank you!) So I reform the graph to 1/R vs current for LED.

1/R vs current for LED
This graph shows me the frequency and current for LED has the relation in almost liner. If I use this for sound making, then I need key follow from key boards. but if I put the anti log amp and tune it in V/Oct response then I can use it for sound making. but itS linearity is so poor. it's no need so exact methods here, I think.
CdS cell changes in sensitivity with length of the wavelength of light. Each brand has each character but 500- 600nm is main area (It's Japanese page). Red is over 650nm .(infra red is over 1000nm), Green is around 550nm and blue is under 450nm.
I think that sensitivity of handmade photo coupler can change by their color of LED.

green LED photo coupler 1/R vs current
I think that main issue is the relative difference. this green's graph and red one have different unit in 1/R. Please check it out. the curve looks different but, its difference of each sample is almost alike. but in red one's there is different curve. It not issue about CdS cell but it's caused by second handed LED, I think. Each of LED coming from cheap christmas lights have different brightness. 1 dollar shop can sale this christmas lights in this reason, I think.
I measure it by pocket digital multi meter. you should better think that there is some errors in this data. You should better to get your own result.
In order to find the difference in the sensitivity by the difference in a color, I made another graph. it's average values of each samples.
Red #1 - #4
1mA for 3k and 10mA for 470ohm in red one.
Green #1 - #5
1mA for 10k and 10mA for 1k in green one.

commercial photo coupler also have errors
Green one is more higher resistance but their curve is almost same as each other. When we build this as a effector then high resister is good to reduce the size or value of the capacitor. if it's smaller then we can reduce the PCB size also.

In this experiments, I feel something threshold in the values of resister of CdS cell. Display of digital meter changes its value so many times if it's over 50k. And it seems to be so drifty. but it's stabilized if it goes down under 50k.
I think that CdS Csll has some threshold values.

I don't use it in this project but, the commercial photo coupler that my friend had send me has this characters.

It seems have so big errors in small current. but we can see that the commercial photo coupler also have errors . if its speeds is enough then hand made photo coupler may be not so bad things.

The raw data is fllow..

About Blie #1 - #6 , check here


  • Green LED HomeMade
  • Red LED HomeMade
  • VTL5C2 I put the LFO with the voltage divider made from 4.7k and CdS cell to scope.
    lower signal shows timing only. Upper line just looks like ADSR is the output. But The LFO is too slow to see my scope so I put the VCO with narrow pulse wave. I think that I can see the wave form like round edged pulse wave, but result is like this. attack time is not so late, but release time is too long. We can't see the 0V line. And if I change the pulse with more wider then the lines moves upward. so its not enough to attack.
    This pictures show that CdS cell has so long time release time, but attack time is not so bad.

    Thermal characters

    Ice bath in a grass
    At last, I check their thermal characters. I put thermal meter, TC, Metal film resister and hand made photo coupler to the plastic bag. and check each samples values in room temperature (26.3C). And I put the Iced water and hot coffer in the glass. I got the 3 data.
    White block in the picture is the thermal sensor. I don't tune it exactly . but we can see the relative changing in this experiment.

    Thermal characters
    Metal film register made great results. It looks almost no changes.
    I think that LED may also have small heat when it lights. and I put it into the tube. So the heat is not easy to run. It may keep its heat in it. so its temperature drifting is going to be so big.So this it just as a guide.
    We need anti log circuit at the LED driver. it's thermal sensitive circuit. And it's not easy to make them collect. but there is also big drifting with the temperature. so it's driver's circuit may be enough by easy methods I think.

    The drifting of resistor per 1C is follow..
    HomeMadePhotoCouplermetal Film ResisterTC resister(3300ppm)
    -1.60 ohm0.133 ohm3.18 ohm

  • Next

    Back to home made synth(Japanese only)
    Go to Takeda's top page (Japanese only)

    ", " Copyright 2003, Motohiko Takeda, Crow Hill Laboratory
    Motohiko Takeda Mail:takeda@aleph.co.jp