The comprehensive phase shift oscillator thread

[midwayfair]

I feel like we could use one of these. Or at least I could!

I know some people hate this type of LFO, but I like it.

What:
"A phase-shift oscillator is a linear electronic oscillator circuit that produces a sine wave output."

Basic reading. http://en.wikipedia.org/wiki/Phase_shift_oscillator
Includes formulas.

Why:
Sine wave oscillators are good for some musical effects because we seem to like how they sound. Especially tremolos like the EA Tremolo and phasers like the univibe.

How:
I'm interested in discussing various ways of dealing with the oscillator that looks like this in its most basic form that I see in stompboxes, using a transistor:



I'll start with super basic stuff and we can go from there. Please please please correct anything I get wrong. Math is fine, but I think we should also use as much plain language as possible, for people like me who consistently got beat up in alleyways by numbers when they were kids.

Most of the time, C1 = C2 = C3 = typically 1uF. Smaller values will produce faster speeds to a point. Larger values will produce slower speeds to a point. The original EA tremolo article suggested that these be tantalum for best performance; I don't know how accurate that still is.

R2 and R3 form high pass filters with C1 and C2. Varying the resistors will change the frequency, which ... changes how frequently it oscillates (rate). Making them smaller increases the speed, but if they get too small the oscillation stops. Making them larger decreases the speed, again, to a point. In many designs, only one of these is variable. There is always a smallish resistor that limits how low the speed pot can go.

R1 affects the "depth" of the oscillation, but not necessarily the depth of the effect, which can be controlled in other ways (and often is). I've found that values much below 2K are less stable. Values above 15K are also unstable. Can someone explain this? Is it simply setting the highest voltage in the swing?

You can make an LED blink in time with the LFO by putting it in series with R1, but to change how strongly the LED blinks you'll need to vary R1 to make the LED darker.

Some circuits that use PSOs
-EA Tremolo. Original article linked here: http://www.diystompboxes.com/smfforum/index.php?topic=68571.0
Interesting features of this implementation: Decouples the oscillating signal from the collector of the phase shift oscillator, and then uses that to feed a control voltage to a FET's gate and vary the resistance between the FET's drain and source.

-Tim Escobedo's Wobbletron uses the same setup with a smaller speed range.

-Univibe (http://www.geofex.com/Article_Folders/univibe/uvfrindx.htm)
This is barely recognizable because it is driving the lamp, but it's in there. I highly recommend reading the Technology Of article if you haven't already. As usual, it's excellent reading at any level. Interesting features of this implementation:
Dual gang speed pot varies R2 and R3 simultaneously.
LOTs of current gain.
Diodes clip the waveform!

-Magnivibe. http://i605.photobucket.com/albums/tt137/pinkjimiphoton/MagnavibeSchemsmall.jpg
Interesting features of this implementation:
The depth varies the intensity of the LED, resulting in a dimmer LED at low depth settings. It never goes completely dark, however, so there's technically no minimum depth before the LFO craps out. I call this "interesting" but it's actually a problem and a big limitation, but it's the easiest way to vary the intensity of an LED, so you can use the LFO to drive a vactrol. In a bit, I'll post another way to drive an LED (or two or three) with this LFO that doesn't vary R1 and makes the LED brighter as the depth goes down, with a complete depth range from all the way on to square on/off.

In my experience, this is just about the least number of components required to get a functioning LFO, especially if you want to drive an LED. It's useful for tight corners despite its limitations.

-See the wikipedia article for a PSO that uses an op amp. You can vary the negative feedback resistor to change the oscillation. I can't remember ever seeing this used, but it seems like it would be a sine wave from an op amp, so I'm not sure why it isn't.

Some observations

I've noticed that R2 and R3 can also be referenced to Va instead of ground and the effect works the same. That's interesting; I'm not sure what use it has.

If you feed a voltage to the junction of R2 or R3 and the capacitors they connect to, you can interrupt and mess with the LFO. Doing this with an envelope signal is the basis of RG's Vibramatic (which I used for my Blue Warbler pedal). The LFO will spin back up in its own time.

The BYOC disconnects Q1's connection to ground in bypass to use an LED in series with R1 as the bypass indicator. Users swear this works, but at slow speeds I've found that this means there's a delay before the LFO spins back up. Instead, I recommend SHUNTING the LED in bypass -- just make your switch connect its anode and cathode. The LFO will continue working and your LED will flash only when the effect is on.

Higher speeds eventually start decreasing the depth. So if the LFO isn't going fast enough, it might be worth looking at using smaller caps for C1-C3 instead of trying to squeeze more out of the speed pot.

I've had some success in changing only the value of C1 to change the range of speeds, but I'm not sure how reliable this is. If it does indeed work, I think this is a better solution for having a slow/fast switch than the usual method of using a parallel or series resistor to change the range of the speed pot, which makes the LFO stop working in some settings. I'd love to hear more thoughts on this.

---

Ideas, thoughts, corrections, etc?

[GGBB]

Awesome topic idea!

Most of the time, C1 = C2 = C3 = typically 1uF. Smaller values will produce faster speeds to a point. Larger values will produce slower speeds to a point. The original EA tremolo article suggested that these be tantalum for best performance; I don't know how accurate that still is.

I suspect (i.e. have probably read somewhere) that the Tantalum requirement stems from the much better leakage characteristics of Tantalum over Aluminum and the lack of (affordable) 1uF film capacitors in that era.

[Mark Hammer]

The tremolo in my Heathkit TA-16 amp uses one....way down in the lower right hand corner.

[tca]

You can do it with mosfets: Fender Tremolo



from geofex.

Not quite the same?!?

[midwayfair]

Not quite the same?!?

It's in there! There's the 3 cap ring and I think it's using negative feedback from the MOSFET farthest on the right.

[tubegeek]

http://web.archive.org/web/20101015022441/http://aikenamps.com/PhaseShiftOscillators.html

[tca]

It's in there! There's the 3 cap ring and I think it's using negative feedback from the MOSFET farthest on the right.

I think that far right mosfet is just a buffer.

I think that far right mosfet is just an voltage amplifier do drive the neon bulb.

[midwayfair]

It's in there! There's the 3 cap ring and I think it's using negative feedback from the MOSFET farthest on the right.

I think that far right mosfet is just a buffer.

I think that far right mosfet is just an voltage amplifier do drive the neon bulb.

Ah, you're right.

I see the feedback now, it's the 1Ms over to the left.

Re: Tubegeek's link:
Is something analogous to C1 useful in a transistor design to prevent distortion?


At one point I had some ticking from a transistor PSO and used a 22pF cap across the collector to base and it made the ticking stop, but it might have just been that I turned off the desk lamp, too.

[Digital Larry]

Just guessing, but C1 reduces gain at higher frequencies so I'd guess it's there for stability.

[Kipper4]

Well what the heck. So if I am asking a stupid question I'm sure someone will let me know.

Seeing as the PSO can drive an LED then does anyone see a good reason it couldn't be implemented to drive an envelope filter via an led/ldr and make an auto wah?

[midwayfair]

Well what the heck. So if I am asking a stupid question I'm sure someone will let me know.

Seeing as the PSO can drive an LED then does anyone see a good reason it couldn't be implemented to drive an envelope filter via an led/ldr and make an auto wah?

Yup, that works. You need an LDR with some pretty specific characteristics -- really low light resistance and not so high dark resistance. NSL-32 might work.

However, I think an op amp LFO is better overall choice for an auto wah, since it would facilitate a saw tooth waveform, which sounds more like using your foot.

[Kipper4]

Thanks John as soon as I get enough time I'll bread it and report back.
Great news. I have a bunch of differant cheaply ldr I can play with.
And thanks for the saw tooth tip too.
Glad I asked now.
Rich

[amptramp]

For a phase shift oscillator with equal R and C elements, the required gain is 29 or more and the frequency is 1/(2 * pi * SQRT(6) * R * C).  The first point is important - you can't make a phase shift oscillator from a 12AU7 but you can from a 12AX7.

[duck_arse]

I feel like we could use one of these.
.....
Most of the time, C1 = C2 = C3 = typically 1uF. Smaller values will produce faster speeds to a point. Larger values will produce slower speeds to a point. .....
You can make an LED blink in time with the LFO by putting it in series with R1, but to change how strongly the LED blinks you'll need to vary R1 to make the LED darker.
.......
I've noticed that R2 and R3 can also be referenced to Va instead of ground and the effect works the same. That's interesting; I'm not sure what use it has.
......
Higher speeds eventually start decreasing the depth. So if the LFO isn't going fast enough, it might be worth looking at using smaller caps for C1-C3 instead of trying to squeeze more out of the speed pot.

I've had some success in changing only the value of C1 to change the range of speeds, but I'm not sure how reliable this is. If it does indeed work, I think this is a better solution for having a slow/fast switch than the usual method of using a parallel or series resistor to change the range of the speed pot, which makes the LFO stop working in some settings. I'd love to hear more thoughts on this.

my favourite oscillator, probably.

midway, I have found that R2 and R4 can either or both be taken to V+, which is useful to polarise electros correctly. point two of the positives together, and take the resistor there to +. in the image shown here:

http://i210.photobucket.com/albums/bb292/frequencycentral/Vibracaster6111V2.jpg

which I always call rick holt's, there is an extra transistor driving the output. I've found this circuit to greatly reduce the humpy output distortion, but it also makes the "good" oscillation range somewhat less, output falls at both ends of pot travel.

on cap changes, you can use 3 slightly different values to start with and caps for range switching, but again the output levels are not as good.

as for speed, you can increase/decrease the fixed (whichever) resistor to set higher high or lower low speeds. within reason. and I think R1 will change speeds some too.

[Kipper4]

I might have completely got this round my neck but here goes.
Would the PSO be able to be used with an optoisolator in the same fashion as a nsl 32?

[midwayfair]

I might have completely got this round my neck but here goes.
Would the PSO be able to be used with an optoisolator in the same fashion as a nsl 32?

An opto isolator is just an LED and photocell in a hermetically sealed package, so yes. Cf the Magnavibe schem I posted.

[tubegeek]

The only useful thing I can add to this discussion is this:

I built a guitar amp based on this schematic:



This is the Supro 16T/Gretsch 6151, which is the amp that the Supro Deux pedal emulates.

I made the following mods after reading Randall Aiken's phase shift oscillator article that I linked above:

1) In the phase shift section, I upped the cap values somewhat to slow the tremolo down. I can look up what I did, but it was on the order of .02/.02/.03 instead of .01/.01/.03, something like that.

2) Also in the phase shift section, I used a dual 1M/1M pot instead of the fixed 1M/100k+500K pot arrangement to try and keep as much range as possible. It's a log pot working backwards so it's labeled "Sloth" instead of "Speed." Anybody who has a spare dual 1M/reverse log pot on hand is more than welcome to PM me anytime!

The amp works great, I really like it. My plan is to add a pentode input channel to it just for fun, but as-is, it's a nifty little beast.

[Kipper4]

I only have one optoisolator in stock and thats the H11F1
Anyone see a good reason why this wouldnt work?
thanks
Rich

[midwayfair]

I only have one optoisolator in stock and thats the H11F1
Anyone see a good reason why this wouldnt work?
thanks
Rich

Do a forum search, there have been a few threads about it. It has issues when used in the audio path. It might work in an EA tremolo type design, replacing the FET as a variable resistance element.

[Kipper4]

will do mate
ive just been glegging at the data sheet too.