Tri Vibe LFO

[Kipper4]

What kind of LFO is used in the ROG Tri Vibe
http://runoffgroove.com/tri-vibe.html

Talk me through it please.
What other type of effects use this type of lfo
Any use for an envelope filter?
Who's built this?

[Mark Hammer]

It's actually a pretty standard LFO, with just a couple mods.

First, the 100k resistor to Vref, and the 10nf feedback cap, result in a more trapezoidal wave form being generated, than square.  Normally such LFOs produce a square wave, which is then turned into a triangle by the op-amp with the caps in the feedback loop.  The sudden current draw that producing a square wave results in, is the source of annoying LFO ticking.  By slowing down thew rise and fall of that square wave juuuuuuussssttttt a bit  - so that it is slightly trapezoidal - that current draw is spread out over a couple of milliseconds so that you don't hear a tick.  This is explained in more detail here: http://moosapotamus.net/files/stompboxology-mo-tremlo.pdf

The other twist is the pair of diodes on the output.  These provide a bit of clipping of the triangle output so that it behaves a little more like a sine wave, which fast modulation often likes better.

[DrAlx]

What kind of LFO is used in the ROG Tri Vibe
http://runoffgroove.com/tri-vibe.html

Talk me through it please.
What other type of effects use this type of lfo
Any use for an envelope filter?
Who's built this?

I've built this and also used the same trick for converting a triangle wave to a sine wave in my NZF Flanger...
http://www.diystompboxes.com/smfforum/index.php?topic=107353.msg974877#msg974877.

There are 3 parts to the Tri-Vibe LFO.
 
1)  The first 2 op-amps form a square wave/triangle wave generator (which is very common as Mark said).
Google triangle and square wave oscillator to get an explanation.

2)  The 2 diodes that follow round the tops off the triangle wave so it looks a bit like a sine wave.
In fact in the Tri-Vibe, the waveform at that point has flatter tops and bottoms than a sine wave.
That is deliberate in order to get a better output waveform from the next stage.

3)  The really smart thing about the Tri-Vibe is the stage following those 2 diodes.
If you plot the inverse of the output voltage from that stage (i.e. 1 divided by the output voltage) you get
a graph that is a VERY good fit to the curve     1/(A + Bsin(t) )  where A is bigger than B.
Why is that important?  Well that sets the control current of the OTA all-pass filters.
So the control current varies like   1/(A + Bsin(t) )  too.
You may be familiar with all-pass filters having a capacitor "C" and a resistor "R".
Well for an OTA all-pass filter, the "R" is determined by the control current and is INVERSELY proportional to it.
Therefore, the OTA all-pass filters in the Tri-Vibe have an "R" that varies sinusoidally (like A + Bsin(t) ).
This will mean that the phase-shift at each frequency will vary-sinusoidally too, which is important for nice sounding vibrato.

[Kipper4]

Thanks guys
I will endeavour to devour what you have shared
Rich

[Kipper4]

Being as it's like a sine wave then it ought to perform well driving a twin T filter too?

[deadastronaut]

yep..

[Kipper4]

Cheers Rob.
I feel a breadboard session coming on.

[Kipper4]

Before I bread it can anyone see any reason why it won't wah?
Lfo oppy is a tl074
Filter oppy is tl072
Rob you may see some bits you recognise.
I borrowed your twin t.
I have looked at all sorts of differant filter configurations and decided to go with a twin t after all.
Thanks
Rich

https://imageshack.com/i/nbjg99j

[/URL]

[DrAlx]

I know that in the Tri-Vibe, that varying control voltage ultimately gives you a sinusoidal variable resistance for the OTA all-pass filter.  The reason for having that final LFO section is precisely because it is feeding that OTA filter, and because the behaviour of that filter is affected by voltage (or rather current) in a particular way.
I could imagine using that same final LFO section if it were driving some other sort of variable filter based on an OTA (so long as it has the same sort of behaviour i.e. effective resistance is proportional to 1/current).

I don't see anything like that in your circuit diagram, but then I'm not familiar with that filter.
Wouldn't the LFO voltage be acting as just another input to the filter (via a couple of caps) rather than actually changing any filter parameters?  

Here's my thinking:
If you wrote down the formula for the notch frequency then it would have a load of R and C values in there, with every R value being multiplied by a C value.
That is necessary for the formula to have the correct units (1/(RC) has units of Hz).
If you want the control voltage V (or control current I) to affect the frequency somehow, then the only way that could feature in the equation
is if there is something that gets rid of the V or I unit and converts it to Hz (or an R or C value).
In the case of the OTA all-pass filter, that thing is the OTA transconductance.  
The control current and the transconductance "g" together give you an equivalent resistance to the signal voltage.

I don't see anything like that happening in your circuit.  In other words, I don't see the control voltage acting like a variable resistance in any way.
I'm just looking at this from a maths/physics perspective.  As I said I am not familiar with that circuit, so I could be talking nonsense.
Maybe someone who knows more about electronics than me can comment.

[Kipper4]

maybe i should go down the route of a voltage controlled ota state variable filter instead then.
or is it just a case of adding a resistor and cap network between the lfo and filter?

[DrAlx]

Just to clarify what I said.  The control current varies the OTA transconductance "g", so the formulas for frequency cutoff have terms with "g" in which has units (1/Resistance).
So the control current makes its way into the filter formula via the transconductance.   See the LM13600 data sheet for examples.

I can't see how putting more R and C's between the LFO and filter changes things. 
Without something like the transconductance, your not actually creating a variable resistance.

You can use an OTA as a voltage controlled resistor (examples in the LM13600 data sheet).
I have seen state-variable filters in the same data sheet.  They don't give formulas but I would expect the "g" to effect things in the same way as the high and low pass examples in that sheet.
I've only used OTAs for all-pass filters and for oscillators, so can't offer more advise on whether the OTA route is worth pursuing for what you want to do.

[duck_arse]

hey kipper, what you have is similar to a dr. quack style autowah around IC1b. if you connect a transitor C to the C16//C17, E to ground, B to a 10k(?) to the lfo, you may get a quack. may.

[DrAlx]

hey kipper, what you have is similar to a dr. quack style autowah around IC1b. if you connect a transitor C to the C16//C17, E to ground, B to a 10k(?) to the lfo, you may get a quack. may.

Yes maybe. Control voltage (through a resistor to B) affects conductivity between C to E.
If the effective resistance of the C-E path is inversely proportional to the control current then maybe you'll get something useful.
I did simulate the whole Tri-Vibe LFO in LT-spice quite a while back.  I'll see if I can dig out the min and max voltage values in different bits of the LFO.

[DrAlx]

Here are the waveforms for 9V supply

[Kipper4]

I'm following now. Sorry.
Thanks for posting your tri vibe results. So it swings between 6.4 and 2 volts.
That's quite some swing. I would think.
I missed out that what's needed is a variable resistance between the lfo and ef. Not variable voltage direct.
Maybe it's time to change up and add the modular voltage controlled filter.
I'm thinking of using something like this for the filter now.


http://www.geofex.com/article_folders/fxbus/fxbmod06.htm

If you think all I would need for the previous filter to work is to implement a bjt and resistor then I may just well do that too.
Thanks for all the help.
Rich

[Kipper4]

I've breadboarded the lfo and for some strange reason I'm not getting a voltage swing on the output just 4.5v.
grrrr

Edit.   ooops i had the 68k going from pin1 to 7 instead of pin3 to 7.
Right thats half the battle its working now.
Sometime you just got to post something to force your self to look at things again.

[Kipper4]

feeding a 2n5088 base through a series 22k from the lfo i'm now getting variable resistance.
if i get time after work I'll bread a filter and see what happens.
cheers
guys
further suggestions always welcome

[Kipper4]

im still struggling to convert the lfo voltage to resistance.

[DrAlx]

Use a FET instead of a BJT.  That's the more common way of producing a voltage controlled resistor.  I looked brieflly at the Dr-Q circuit and I'm not convinced the BJT is actually acting as variable resistor there. Maybe someone else who knows how that circuit works can clarify.

If I were you I would LT-spice the rest of the circuit (everything apart from the LFO) and see what the frequency response looks like for various fixed resistors.  That will tell you the range of resistances that the variable resistor needs to cover.

[Kipper4]

Thanks Dr Alx
I'll look into it.
I've seen so many permutations it's a guessing game which will work.
I don't have spice perhaps it's high time I did.
I'm sure there's a fet example on RG s technology page.
I couldn't get the dead Astronaught filter to work on my breadboard so I quickly breaded the op amp one from RG s page again, and although it's perhaps not what I want to end up with at least it's a starting point.
I put an ldr on the filter just to check it functioned as expected and it performed well with a finger proximity test.
It looks like I have a ways to go but every day I'm getting closer.
Thanks for the support
Rich