LFO skewed sine

[lars-musik]

Hi folks, another question regarding the Ibanez RC99. The LFO produces a somewhat skewed sine wave. I like it, but would still like to be able to understand and modify it. Can you point me to the right direction. Should be some asymmetry around U8, I guess.

[ElectricDruid]

Does it?! It looks like a standard schmitt trigger integrator LFO to me, which makes it a triangle output. Of course, if you add a bit of distortion to the triangle, you can get plenty of varieties of sine-ish and skewed-sine waveforms. Another thing you sometimes see is some post-LFO filtering of the waveform, which smooths it out and limits the depth at higher frequencies (which can help for chorus). I don't see any sign of that here though.

Where are you measuring and seeing the skewed sine? I'd expect a triangle at U8 pin 7. I can imagine that the Q1 bias set by SR3 might alter the effect that has on the clock quite a bit and by then some distortion could bend the triangle into something else. But I'm speculating and I'd have to simulate it to be sure.

[antonis]

+1 to what Tom said..

"Flaw" here is the deformed triangle and not the skewed sine..

[lars-musik]

Thanks Tom. Up until now it's only a "perceived skewed sine wave". I will do some reading in the direction you gave me, followed by some experimentation.

[lars-musik]

hi Tom, Antonis,

still on a learning curve - these are the waveforms on the respective pins of the MN3102. If I'd like to un-skew the signifcant one
(1) which one would be "the relevant one"
(2) where in the LFO generator would you start looking for waveform manipulation (and would a sine wave be possible)?

Thanks very much for your input!

[ElectricDruid]

This is back to that complicated problem of "how does modulation affect the pitch shift of a BBD?"

I've spent a lot of time looking at this, and I don't really claim to have a complete answer.

There are various things at work. The first is how the modulation affects the clock frequency. This could be linear-frequency, e.g 1V of modulation produces X KHz of clock frequency shift. It could be exponential-frequency, e.g. 1V of modulation produces an octave of clock frequency shift. Or it could be linear-period, e.g. 1V of modulation takes X msecs off the clock period. This produces a 1/x curve compared with the first case, and is generally regarded as the best case for chorus.
The second thing that's at work is how the changes in the clock frequency are reflected in pitch shifts in the signal going though the BBD. This is often misunderstood. The naive assumption is that the pitch shift is the derivative of the modulating waveform, but this is not actually true. It's close, but it's not right, at least for analog BBDs. It's ok for digital delay lines.
For a triangle wave, this assumption would mean that there were two pitches produced, becuase there are two slopes on the waveform. For modulated BBDs, the situation is slightly more complicated than that.
The reason is that the *total* delay seen by a given output sample is the sum of the previous X clock periods (where X is the effective number of delay stages). So the delay is an integration under the clock period curve. It's not just the derivative at a point, but a sum over an area. This means that for the triangle wave modulation case, there are *four* situations, not two:
    1    Going up
    2    Going over the top
    3    Going down
    4    Going around the bottom
Now, (1) and (3) are the "square wave" parts of this, where there's a constant pitch shift related to the slope of the triangle. That's the only part that's generally considered. The other two parts act to slew the pitch between these two frequencies, and this happens quicker at the top (where clock frequencies are higher and hence delays shorter) than at the bottom (slower, longer delay, longer slew).
I've been meaning to write a follow-up article to the one that's already on my website with what I've learned since I wrote that (some years ago now already) but it's hard to express this stuff in a way that makes it easy to understand, so I oftn get bogged down in it and it never quite seems to get done in a way I'm happy with. One day...