LFO for stereo modulation effects

[ChanchoPancho]

Hi guys, I've been messing around with stereo effects lately and I'm attempting to build a stereo phaser. The idea is to have a full stereo effect (2 inputs & 2 outputs) so I decided I'll have 2 separate effects paths, identical as each other, but controlled by a LFO that has its outputs phase shifted. I haven't decided if its output will be 90º or 180º out of phase.

Somewhere (in this forum I think...) I read that when it comes to phasers it's best to use 90º out of phase LFO. So I started looking for quadrature LFO's.
I ended testing Elliot's state variable filter as an oscillator: http://sound.westhost.com/articles/sinewave.htm#s61

I used it for controlling the path of a couple of phase 90's. To make it work I had to adjust the LFO's outputs to match the original Phase90 amplitude and offset. I got it to work but, the main problem I had is that when not set in the max rate, it took too much time to get the LFO to oscillate, sometime it stopped in the min rate position.

I'm still looking for other options, but I'll ask in case that someone has a tip:
Do any of you know a simple triangle LFO, with normal (0º) and out of phase (180º) outputs, that will not require a negative supply? Most schems I found used V+, V- and gnd. Any ideas will be appreciated, thanks in advance...

[nocentelli]

The catalinbread pareidolia has a XR2206 for LFO and uses opamps to invert the phase. Try the other place for the schematic.

[drolo]

you could use a standard LFO and hook up an inverter like RG shows here:

http://www.geofex.com/FX_images/Panning%20LEDs.pdf

You don't need the full circuit he has necessarily. In his example he is driving LED's in the feedback of the opamps, but you can drive other stuff, like FETs etc.
I recently hooked it up with just 2 opamps following the LFO. The first one has the offset trimpot which comes in very handy, depending on your LFO signal. If your LFO has it's signal centered around the middle of the supply voltage ( ex from 2 to 7 V instead of from 0 to 5 v) you could probably omit the offset and just use one inverter opamp after your LFO

Works like a charm (for me at least ;-) )

[SISKO]

180° is easy, its just an inverting opamp.

For the LFO you can use this http://home1.gte.net/res0658s/fatman/4069_lfo.gif or the typical LFO as in the Echobase modulation section

[armdnrdy]

Hi guys, I've been messing around with stereo effects lately and I'm attempting to build a stereo phaser. The idea is to have a full stereo effect (2 inputs & 2 outputs) so I decided I'll have 2 separate effects paths, identical as each other, but controlled by a LFO that has its outputs phase shifted. I haven't decided if its output will be 90º or 180º out of phase.

Take a look at the Mutron Bi-Phase.

I believe that it does exactly what you want to accomplish.
You might be able to pick up some pointers!

[ChanchoPancho]

Hey thanks for the replies!

180° is easy, its just an inverting opamp.

For the LFO you can use this http://home1.gte.net/res0658s/fatman/4069_lfo.gif or the typical LFO as in the Echobase modulation section

you could use a standard LFO and hook up an inverter like RG shows here:

http://www.geofex.com/FX_images/Panning%20LEDs.pdf

You don't need the full circuit he has necessarily. In his example he is driving LED's in the feedback of the opamps, but you can drive other stuff, like FETs etc.
I recently hooked it up with just 2 opamps following the LFO. The first one has the offset trimpot which comes in very handy, depending on your LFO signal. If your LFO has it's signal centered around the middle of the supply voltage ( ex from 2 to 7 V instead of from 0 to 5 v) you could probably omit the offset and just use one inverter opamp after your LFO

Works like a charm (for me at least ;-) )

On the suggestion to just invert the lfo signal it's not that easy, remember that in the original P90, polarization is done between 9v and GND so the LFO output has some DC level. To invert this you have to filter DC with a cap, but sice we're talking about frequencies that goes down to 0.1 Hz app, you have to put there a really big cap. The way to do this is to change the supply to +/-9, but that's going to be my last resort.

The catalinbread pareidolia has a XR2206 for LFO and uses opamps to invert the phase. Try the other place for the schematic.

I saw that IC a couple of years ago and it discouraged me that it was so expensive (~USD 16), and it just gives one signal, so i have to invert the output to get the 180° shifted one.

Hi guys, I've been messing around with stereo effects lately and I'm attempting to build a stereo phaser. The idea is to have a full stereo effect (2 inputs & 2 outputs) so I decided I'll have 2 separate effects paths, identical as each other, but controlled by a LFO that has its outputs phase shifted. I haven't decided if its output will be 90º or 180º out of phase.

Take a look at the Mutron Bi-Phase.

I believe that it does exactly what you want to accomplish.
You might be able to pick up some pointers!

I didn't know about the Bi-phase, I'm gonna be studying the circuit, thanks.

In other forum (http://electro-music.com/forum/topic-17749.html) i saw a Tom Henry's article about some quadrature function generator. It looks very good but still uses negative supply, I'll see if I can modify it to only use positive power supply.

[nocentelli]

I saw that IC a couple of years ago and it discouraged me that it was so expensive (~USD 16), and it just gives one signal, so i have to invert the output to get the 180° shifted one.

I've got one you can have for less than that if you want to try it. I've got a couple actually, might try this idea out again. Keep us updated on whatever you end up going with.

[ashcat_lt]

On the suggestion to just invert the lfo signal it's not that easy, remember that in the original P90, polarization is done between 9v and GND so the LFO output has some DC level. To invert this you have to filter DC with a cap, but sice we're talking about frequencies that goes down to 0.1 Hz app, you have to put there a really big cap. The way to do this is to change the supply to +/-9, but that's going to be my last resort.

Why?  Isn't it just as easy to connect the + input of the opamp to the same Vref which biases the oscillator?

[puretube]

any solution here???

[ChanchoPancho]

On the suggestion to just invert the lfo signal it's not that easy, remember that in the original P90, polarization is done between 9v and GND so the LFO output has some DC level. To invert this you have to filter DC with a cap, but sice we're talking about frequencies that goes down to 0.1 Hz app, you have to put there a really big cap. The way to do this is to change the supply to +/-9, but that's going to be my last resort.

Why?  Isn't it just as easy to connect the + input of the opamp to the same Vref which biases the oscillator?

You were right, man. In previous attempts I couldn't d this. Well here is a simulation done in Circuitmaker 2000. UB and its associated components ar added, the rest of the parts are the original.



I haven't been able to breadboard this, but tomorrow I will. I'll keep you up to date! thanks

[drolo]

For some reason I was never able to simulate this in LTSpice ... always gave me absurd results. Maybe my op amp models are all wrong ... I have not tinkered with simulators much so maybe I did something wrong ... at least it worked in real life ;-)

[ChanchoPancho]

I've been using CircuitMaker for like 10 years but couldn't get circuits to oscillate. Rencently I figured out that you have to put an initial condition module for that to happen. It is in the node that connects R4, R1, C2, etc. I don't know If that's your problem in LTSpice but sure you might try it.

[Mark Hammer]

Sometimes, you can have too much stereo.

For example, the Ibanez Flying Pan combines phase-shifting and autopan.  It's a neat effect, but after a very short period becomes unlistenable, and makes you want to shout "Sit STILL, already!!".

My own suggestion has generally been to pan the wet signal such that the overall amplitude of the two outputs remains constant enough not to give you headaches, but the phase-shift effect moves around.  The other advantage is that one doesn't get cancellation "in-air" when a sum and difference signal collide.

But here's another approach to try that is actually a lot simpler.  Simply add a pair of fixed allpass stages to one output.  So, let's say you have 4 swept allpass (phase-shift) stages.  Those get sent to each of two outputs to mix with dry signal.  But on one of those outputs, the wet signal goes through 2 additional fixed stages before blending.  Since there is more cumulative phase shift on that "side", the notches would be located differently.  Heck, if you want to you could make one of those stages lead and the other lag, such that the dissimilarity of the notch locations occurred towards the extremes of the sweep.

Finally, if one is aimng for an always-on sort of background swirl, aperiodicity is a lot more interesting to the ear than predictably fixed offset in sweep (e.g., 90 or 180 degree).  In which case two independent LFOs that result in unpredictable "movement" between the two outputs becomes more interesting than use of something like a quadrature LFO.  Alternatively, a quadrature LFO mixed with another slow LFO added to each side, could produce interesting aperiodic or semiperiodic patterns.

[PRR]

> couldn't get circuits to oscillate

SPICE first removes all the capacitors, finds the DC voltages everywhere, pre-charges the capacitors to those voltages and puts them back, then starts simulating time.

Most oscillators, if you could do that, with PERFECT parts, will indeed sit there forever, never moving.

With REAL parts there's usually some unbalance and alwasy some thermal hiss. Once the operating point moves even a little, it wants to move further, OTOH there's a system to bring it back again. It's unstable and oscillates well.

So as you found, it needs a kick. I often use a time-switch to hold it a volt off-center and then disconnect that 1mS after sim starts. An IC card which is not the exact "stable" voltage may also do it; it starts at IC and then falls toward center, overshoots, etc etc,

> it took too much time to get the LFO to oscillate, sometime it stopped in the min rate position.

All the phase-shift sine oscilators take time to start.

They start from zero, and get going by thermal hiss. Each cycle the hiss is amplified by the excess gain.

Fender Trem LFO needs gain of 27(?) to work at all, and somewhat more to start. Say we have gain of 30, 10% in excess of required. Every cycle (pass around the loop) the level rises 10%.

Say that thermal hiss is 1uV. Say the stable limit-level is 100V (this is a tube job). We need to build-up by factor of 100V/1uV or 100,000,000. We build-up 10% per cycle. 1.1 (10% excess gain) needs 194 passes to get 100,000,000 times bigger.

With a 1MHz radio oscillator, 200 passes is imperceptable, oscillator starts "instantly".

With a 10Hz LFO, 200 passes is 20 seconds.... indeed you CAN'T stop-start the oscillator mid-song, the bridge will be over before the wobble starts.

You can put more excess gain in the loop. For 50% excess gain (1.5) you only need 46 loops. Still that's 4 seconds at 10Hz. And 1.5 times excess gain is a lot to dial-out when level comes up to limiting, you usually have distortion issues, or a wobbly on top of your desired 10Hz.

What Fender did on the better amps was to put about 10V DC (0.2V may be better for 9V work) somewhere and switch it away when LFO had to start. That "throws" it into oscillation. There is a magic kickstart value where the first quarter-cycle is very nearly perfect. Generally 1/4 to 1/2 of the peak LFO voltage at that node in full operation.
___________________________

If you are inordinately fond of BELOW-BASIC oscillator theory, read Niku Yeng's exploration (mentored by Dr. Leif {ghost-written by Barrie Gilbert}).
http://www.analog.com/library/analogdialogue/leif/d_day.html

Warning: lot of words, additionally obfuscated by a future-fiction perspective.

[cloudscapes]

I made a stereo vibrato once that could go pretty deep and slow, and the LFO for one channel was the inverted version of the other (with a bit of offset and reshaping, because of how the vibrato worked).

it made for a pretty seasick experience with headphones

[ChanchoPancho]

Well here is the schematic of the stereo phaser I'm designing.



I wanted this to be as flexible as it can get, so I added a couple of switches. Those switches let you choose between modes of operation wich include stereo/mono, Phaser/Vibrato, LFO sync or inverted.

I haven't build it yet since I'm still developing this in separate pcb's and breadboard. (that's why it version 0.9  )

[drolo]

Looks impressive :-D

When you get to breadboard it, check if your LFO signal is centered around 4.5V. If not you may have to apply some offset voltage to the inverter. For example, if your signal was 0- 5 V, the inverter would put out something like 4.5 - 8v, so you can add the offset to move the signal to the wanted area.

It is explained in RG's panning LEDs document.
I did the same in a tremolo I just finished: http://www.diystompboxes.com/smfforum/index.php?topic=103714.0

But maybe your signal is alright already and you won't need it ;-)

[ChanchoPancho]

When you get to breadboard it, check if your LFO signal is centered around 4.5V. If not you may have to apply some offset voltage to the inverter. For example, if your signal was 0- 5 V, the inverter would put out something like 4.5 - 8v, so you can add the offset to move the signal to the wanted area.

I've already checked that when I did the test. VR2 & VR3 Adjust the bias in each channel through R44/C11 & R45/C12.

I have just realized that I missed something though: This bias adjustments must be before switch 3, because it adjust each LFO output and if implemented as showed in the schematic, may result in wrong bias value when s3 is switched. I'll redraw the schematic after I test a couple of more things.
I'll keep you posted!

[ChanchoPancho]

Hey another question: Do you know a simple way to implement the vibrato mode (Phase 90) with the switch, but not to pop when switched?
By simple I mean a couple of R's C's and maybe a transistor, the idea is to keep the parts count low.

I know that the other way to do this is by adding a blend/mix pot in the base of the output transistor, but I ask you since I would prefer to do it with a switch.