LFO Vertical Symmetry and Clipping

[LaloFP]

Hi guys!

I built a chorus based on the Walrus Julia (extended Boss CE2). I used PedalPCB schematic but used my own board and components.

Works great except the Shape Switch (Triangle to Sinewave with two 3mm Red LEDs):


VREF_B is just a 10k voltage divider:


In the oscilloscope I confirm that only one side is being clipped:


The blue line is VREF and is 4.49v and the rails for the Opamp are 9.05 and GND
So the LFO is exactly half biased, but the LFO output is not centered on 4.5v (that's the root reason)
Based on a tremolo I did some time ago, I think that's normal and expected because of the real properties of the opamp (compared to an theoretical ideal Opamp)
(photo without the clipping)


Questions:

• Is that correct? Or could I vertically center the LFO output to 4.5v? (without changing VREF)

If I'm correct, I have these hypotheses, what do you think?:

• The schematic is wrong and the Voltage Divider is not symmetrical in the Original Walrus (couldn't find guts photos)
• The schematic is wrong and there is another difference
• The schematic is right but Walrus use two different LEDs with a different threshold each (guts photos seems to dismiss this)
• The schematic is right and Walrus use identical LEDs but with a smaller threshold, clipping more each side, but still asymmetrically (like my 2 White LEDs example below)

What do you think?   

Other test's I did with different LEDs:
Red & White LEDs


2 White LEDs

[Locrian99]

The change is really really subtle built with the red led's.  There is a mod for using I believe 2u2 mlcc's in place and it gives a slightly more noticeable change in shape.   But still it's subtle at best. 

[ElectricDruid]

It's quite likely that the LFO is a bit off-centre because the op-amp can go closer to one rail than the other. That behaviour depends on exactly which op-amp you use for the LFO. Some of them are more symmetrical than others, some get closer to the rails than others, etc etc. In this circuit, that will matter.

The amount of clipping and the final waveshape is going to depend heavily on the forward voltage of the LEDs too. I'd be inclined to experiment with that, or with alternative clipping diodes (series si diodes, maybe?) to get the best shape.

I know you said you didn't want to change Vref_B, but the obvious and simplest way to get the clipping more even would be to use a trimmer for it, like they've done on the other Vref supply. In fact, given that the things that Vref supplies are a lot less fussy than that LFO, I do wonder if the Vref and Vref_B shouldn't be the other way around. It would make more sense.

[Locrian99]

Here's a couple threads on the subject. 

https://forum.pedalpcb.com/threads/caesar-chorus-question-shape-switch.16938/

https://forum.pedalpcb.com/threads/caesar-shape-switch.14426/

[LaloFP]

I know you said you didn't want to change Vref_B, but the obvious and simplest way to get the clipping more even would be to use a trimmer for it, like they've done on the other Vref supply.

Yeah, but my intention here is not to practically fix the effect (i've already done it with the white+red LEDs). Is to clear the theoric doubt about if the rest of the components can contribute to that offset and to confirm that the only players are (and i'm not missing other possible contributors):

• Opamp real Specs
• VRef

It's quite likely that the LFO is a bit off-centre because the op-amp can go closer to one rail than the other. That behaviour depends on exactly which op-amp you use for the LFO. Some of them are more symmetrical than others, some get closer to the rails than others, etc etc. In this circuit, that will matter.

Can one totally infer that looking at the datasheet?
This graph (TL072, what I'm using now) tells me the upper limit but I didn't find the lower limit:

I also couldn't find that graph for the TL022 (the original walrus spec)

The change is really really subtle built with the red led's.  There is a mod for using I believe 2u2 mlcc's in place and it gives a slightly more noticeable change in shape.   But still it's subtle at best.

Thanks mate for the links! I'm going to test those caps. Its subtle but for me there is a clear difference in the feelling between a sharp and a smooth lfo (depending on the other settings)

[LaloFP]

The change is really really subtle built with the red led's.  There is a mod for using I believe 2u2 mlcc's in place and it gives a slightly more noticeable change in shape.   But still it's subtle at best.

Hey I just tested that and the problem with caps (i remember it too from the tremolo i did some time ago) is that the clipping amount and final amplitude is dependant of the cap. When the LFOs freq is higher, the amplitude gets reduced a lot so you lose effect

[ElectricDruid]

Yeah, but my intention here is not to practically fix the effect (i've already done it with the white+red LEDs). Is to clear the theoric doubt about if the rest of the components can contribute to that offset and to confirm that the only players are (and i'm not missing other possible contributors):

• Opamp real Specs
• VRef

Yeah, seems like those two cover it. There are various opamp specs that might be important though - although I do think that the output clipping level is probably the most important.

It's quite likely that the LFO is a bit off-centre because the op-amp can go closer to one rail than the other. That behaviour depends on exactly which op-amp you use for the LFO. Some of them are more symmetrical than others, some get closer to the rails than others, etc etc. In this circuit, that will matter.

Can one totally infer that looking at the datasheet?
This graph (TL072, what I'm using now) tells me the upper limit but I didn't find the lower limit:

I also couldn't find that graph for the TL022 (the original walrus spec)

The left-hand Y-axis says "+/-10V" or whatever, so it's at least *implying* that the maximum positive and negative output voltages are the same. Your practical experience says that's not entirely true though, and I think mine does too.

Hey I just tested that and the problem with caps (i remember it too from the tremolo i did some time ago) is that the clipping amount and final amplitude is dependant of the cap. When the LFOs freq is higher, the amplitude gets reduced a lot so you lose effect

Yeah, that's the difference between filtering and clipping. For some applications, reducing the depth with increasing frequency is what you want, so it's not always bad.

[LaloFP]

Awesome! Thanks buddy