Capacitors in Parallel with a Vref resistor... What The Heck?

[Paul Marossy]

The spikes arise are because the Muff control is boosting the treble.  If you use a low of low-pass filters to clean-up overdrives then you might not see it often.

If you think of the output of the clipper as an approximate square wave, treble boosting a square-wave has peaks on the edges.

The suck-out after the edge is the mid-cut.

Here a cool pic on EQ'ing square waves,

That's interesting, and a very cool .GIF

Yeah, most of my drives smooth out all that stuff I guess, which is why I've never seen this sort of thing on my scope before. I hate using anything that sounds harsh and way trebley, give me "woman tone". ha ha

What would this look like on a spectrum analyzer? How do you determine what harmonics are happening? Seems like you can't really do this with an oscilloscope and a sine wave.

[Paul Marossy]

The other thing to remember is that wherever filtering is happening, you've also got phase shifts. And those phase shifts will affect different frequencies in the signal differently. For a complex waveform like a distorted guitar, that can make it a lot different from what we might "expect" to see.

These waveform pictures are all "time domain" - what the signal does at each moment in time, but for filtering, we'd be as well to look at the "frequency domain" - the amount of each frequency in the signal across the audio range.

Because of the phase shifts, the relationship between the frequency domain and the time domain (the waveform image) is not at all obvious. You can make a thousand waveforms that sound *exactly* like a square wave, and yet don't look anything like one, just by changing the phases a bit.

Aha, phase shifting is a concept I can kinda understand. That would explain some of why these wave forms I was looking at were surprising to me. But I was just feeding a simple sine wave thru my pedal and into the scope, I know that using a guitar is a completely different animal in that respect.

How does one view the frequency domain then? A spectrum analyzer?

[Rob Strand]

What would this look like on a spectrum analyzer? How do you determine what harmonics are happening? Seems like you can't really do this with an oscilloscope and a sine wave.

When you clip a signal like a guitar the spectra has a lot of spikes (narrow peaks due to harmonics).   When you pass that through a filter the spikes just get increased or decreased based on the filter's frequency response.

I don't find looking at spectra very useful for distortion pedals.  The information is there but it's in a messy form.  I find the frequency response more useful.  If you decrease a low-pass filter cut-off from 4kHz to 3kHz you know what it's going to sound like.

If you want to try it, just pass the signal into your PC and display the spectrum with some audio software.  If you look at the spectrum with a 4kHz low-pass filter then a spectrum with a 3kHz filter you can see the highs are decreased but you are extracting simple information from a big mess.

[teemuk]

It did not catch my eye but are those schematics "verified"? Do they depict the circuit of a real commercial design or should one regard them as possibly erroneous? Are we sure the caps (in feedback circuit) aren't supposed to be IN SERIES with the resistor?

Because, IME, arranging the circuit the way it is now (caps in parallel to resistors) ensures all high frequencies above the turnover frequency of the filter will have voltage gain equal to astronomical figures of Opamp's open loop gain. Amplifying highest frequencies in such manner will not only amplify the highest and harshest harmonics of the distortion but I'm practically sure that the circuit will also oscillate in some manner because of doing so. If not constantly then you most likely see at least spurious oscillations at sharpest "corners" of the clipped waveform. Yes, the additional filtering may make it seem less apparent.

Yes, filters effect square wave response distinctly. That's electronics 101 and also one of the reasons why square waves are used as alternative test waveforms in addition to sinusoids. Sinusoid input will reveal distortion, square wave input will reveal response and overall stability of the performance.

Folks not understanding this (how filters work) like to make statements such as "tubes don't clip hard" when in practice the tubes can clip the output signal to perfect "square wave" but any filter post clipping will simply hide the most apparent "square waveness" or flat tops of hard clipping.

Another nice revelation of similar vein is to look at output of a multi-band distortion circuit. It does not clip in conventional manner, clipping top of the waveform. Because the composite output signal is a sum of two or more distorted signals any phase difference will create a hard "transition" section and output signal from, e.g. 3-band distortion, will actually resemble a southern american pyramid with "layers" of clipped sections. It kinda looks like crossover distortion but it isn't.

[teemuk]

Pure sinusoid is a waveform that is constructed from a single frequency only. When you distort it the distortion adds "harmonics", frequency components that were not present in the original signal. The spectral analyzer will basically show what harmonics are created and in what magnitude.

But do note that harmonics always relate to specific input frequency e.g. 2nd harmonic of 100Hz is 200Hz but 2nd harmonic of 1kHz is 2kHz. So, for example, low pass filter with turnover f of 5Khz will attenuate harmonics above the 5th when the input signal is 1kHz, but for input signal of 100 Hz the same filter will only attenuate harmonics above the 50th.

[Paul Marossy]

It did not catch my eye but are those schematics "verified"? Do they depict the circuit of a real commercial design or should one regard them as possibly erroneous? Are we sure the caps (in feedback circuit) aren't supposed to be IN SERIES with the resistor?

Because, IME, arranging the circuit the way it is now (caps in parallel to resistors) ensures all high frequencies above the turnover frequency of the filter will have voltage gain equal to astronomical figures of Opamp's open loop gain. Amplifying highest frequencies in such manner will not only amplify the highest and harshest harmonics of the distortion but I'm practically sure that the circuit will also oscillate in some manner because of doing so. If not constantly then you most likely see at least spurious oscillations at sharpest "corners" of the clipped waveform. Yes, the additional filtering may make it seem less apparent.

Yeah, it was a limited edition commercial design. I reversed this thing nearly ten years ago and bread boarded it at the time to verify that it works (and it's still on my big bread board all these years later). I simply wanted to build it and get this off my bread board and close a chapter in my life that I don't want to look back on anymore. Anyway, I've been collaborating with someone recently who has also verified it and was able to determine what these six components actually do. But the three things that I mentioned at the beginning of this thread were what didn't make sense to me. I initially saw the RC combo on the non-inverting inputs as a biasing thing, which I have since learned is actually a filter. Those filters being connected to vref is what threw me off. So the ones on the non-inverting inputs, it's resistor connected to vref, then cap to non-inverting input. If I do it any other way the circuit does not function like it should - badly or not at all.

The cap and resistor on 2nd stage input ARE in parallel. That makes sense to me, one is biasing the input and the cap is I think supposed to be bleeding some range of frequencies to vref (or I guess in this case like an elevated ground).

There is some weirdness in this circuit which I don't quite understand, but when the full schematic is finally revealed maybe something will pop out to the real brainiacs here.

Yes, filters effect square wave response distinctly. That's electronics 101 and also one of the reasons why square waves are used as alternative test waveforms in addition to sinusoids. Sinusoid input will reveal distortion, square wave input will reveal response and overall stability of the performance.

Well, I must confess my ignorance as a self-taught person about the square wave testing. I thought running a square wave into a speaker is bad for it... that's why I use sine waves. But I can see how things would be a little less confusing using square waves, or maybe more obvious as to what is happening.

[Paul Marossy]

Pure sinusoid is a waveform that is constructed from a single frequency only. When you distort it the distortion adds "harmonics", frequency components that were not present in the original signal. The spectral analyzer will basically show what harmonics are created and in what magnitude.

But do note that harmonics always relate to specific input frequency e.g. 2nd harmonic of 100Hz is 200Hz but 2nd harmonic of 1kHz is 2kHz. So, for example, low pass filter with turnover f of 5Khz will attenuate harmonics above the 5th when the input signal is 1kHz, but for input signal of 100 Hz the same filter will only attenuate harmonics above the 50th.

Bingo! That is what I wanted to know about a spectrum analyzer. Sometimes I wish I had one. 

[teemuk]

http://www.r-type.org/articles/art-125.htm

[Paul Marossy]

http://www.r-type.org/articles/art-125.htm

Oh thanks! I'm on it. 

[j_flanders]

Here's another example of that 'shape': ( from when I was comparing some Big Muffs a couple of years ago.)
Top Left: Black Russian
Top right: Big Box NY ri
Bottom left: Tone wicker version
Bottom right: Opamp version

What would this look like on a spectrum analyzer? How do you determine what harmonics are happening? Seems like you can't really do this with an oscilloscope and a sine wave.

This is from the opamp version, feeding it 110Hz apparently.
The harmonics are the 'spikes' to the right, multiples of the fundamental, 220hz, 330hz and so on.
Their presence (because in some clippers there are very few to no even harmonics) and amplitude (compared to the fundamental and to eachother) are a big part of what determines a 'tone' or 'timbre'. There are other factors as well though, like envelope, attack, decay, sustain, release and some more: https://en.wikipedia.org/wiki/Timbre

a spectrum analyzer. Sometimes I wish I had one. 

You most probably have one, if you have a computer.

As for tone generator, you can use this:
Any downloaded or DAW generated sine wave, played by a computer, recorded into a looper pedal (TC-electronic Ditto looper).

As for scope you can use this:
USB audio interface: any digital modeler will do. I'm using a Zoom G3.
+
Computer with a DAW. I'm using Audacity or Reaper, both are free.
Reaper has some free plugins (SPAN voxengo for example) that provide real time spectrum analysing. I like to see (besides hearing) what happens when I turn a knob.
In audacity you have to record first. Afterwards you can hit 'generate frequency spectrum'. But no real time investigating. This has the advantage though that you can export the data to Excel for example and combine the frequency plots of different pedals or versions.

So:

Looper pedal + (+ audio probe)  + usb interface + DAW will get you a long way.

But I've already seen you have some very nice gadgets in that youtube video.

If you wish to analyse guitar pickups and circuits, you can use the same setup but you need a different 'tone generator', namely an exciter probe:

http://kenwillmott.com/blog/archives/152
It's easy to build and lots of fun to play around with and to see what different caps and pots and other stuff (check this: http://kenwillmott.com/blog/archives/246 ) does to your pickup's frequency response.