Gus's Octave Up SIck Box / octave tuning and theory

[John Lyons]

Nice fuzz octave here! sorry about the size, it's not my schematic...
Right click > "view image" will help view it full page.

With the gain maxed the circuit compresses a LOT.
A pumping/breathing slowish compression.
By backing off the guitar volume or turning down the gain the pumping clears up and the fuzz octave sounds nice.
By lifting one of the 33K resistors from ground you can get just the fuzz which is nice as well.

I'd like to shift the threshold of the octave so it's not quite as squashed with the gain turned up.
I raised the 33Ks to 100K but this didn't do that much that I can tell.
Are the germaniums/33k.100K  responsible for the threshold of the squish?
I like the feel of the germaniums but I'd like to raise the threshold...
Where should start?

thanks

john

[John Lyons]

Bumpsky

[DougH]

The over the top compression of the OUSB is intentional (hence "Sick"). IIRC it has something to do with how the op amp is set up but Gus will have to comment on that.

The 100k and the 2 33k's sets the bias voltage/current for each of the germanium diodes. The two 33k's are essentially in parallel so you have a voltage divider with 100k and 33k||33k. That will determine the bias voltage. The diodes are biased on and once they are on they are on and moving that voltage higher won't make much difference. http://en.wikipedia.org/wiki/Diode

[John Lyons]

Ok, thanks for the clarification on that Doug.
With fuzz only (lifted 33K) it's well behaved so I was thinking that the diodes were part of the "Sick" 
I'm just looking to shift the usable range of fuzz with the octave on. (I'd like more fuzz gain with the octave on before it totally caves in)

Calling Mr Smalley....

john

[DougH]

Well, it could have something to do with it, I don't remember how it works. Gus will have to report.

[Mark Hammer]

That first op-amp stage is set for some moderately serious gain (max = x371), so you may want to consider toggling between the 270R ground-leg resistor and a 470R value, for high/low gain.  Another thing you might want to consider is replacing the 33k fixed resistors with 27k and a 10k trimpot that has the wiper tied to ground.  Between the tolerances of the ground resistors and the 10k components used for the unity-gain section, plus variation in the diodes, you may be able to use this to nail your ideal octave generation.

With one of the ground resistances lifted, so as to eliminate the doubling, you will get some crossover distortion because the remaining diode is in series.  I was going to suggest using a switch to disable one of the diode connections, as in the Foxx Tone Machine, but I gather the +9v bias is part of the "sound" so maybe lifting that second diode is not called for.

A person may wish to insert a variable resistance between the signal path and the back-to-back diodes, to soften up the additonal clipping.  Of course not having heard the thing, I'm just shooting in the dark with respect to what might conceivably "improve" it.

[John Lyons]

Thanks for the tips Mark
The trimmer to balance the octave is a good idea, like the superfuzz...

With one 33k lifted I like the sound and amount of gain so that's ok with me.
Mainly I would just like to know why the signal compresses so hard with the octave diodes in circuit.
The second stage op amp is set up like a buffer/unity gain I believe so nothing odd going on there.
Maybe I ought to try SI diodes and see how that sounds, it should raise the threshold above the meltdown point.

john

[DougH]

If both diodes are biased "on" there should be no balancing required given that the op amp gain stg and inverter functions as a nearly perfect phase-splitter.

You might be right about the high gain of the gain stage, Mark. I thought about that but 51db didn't seem that excessive to me. I've played with op amp circuits around 60-70db without the ridiculous compression this has (but then I had other problems like oscillation). Then again, slamming 51db's down to subterranean frequencies (the 100uf cap in the inverting circuit) into a rectifier may have something to do with it...

[Mark Hammer]

Well maybe what this wants is a smaller ground cap than 100uf (3.3uf to 4.7uf) to taper those lows, and a cap in the feedback path (680pf to 1000pf) of the 1st stage to rein in the highs.  Those who like the splat can simply toggle in a 22uf-100uf cap in parallel with the suggested smaller value to regain it. 

It strikes me that mid-humps can be very useful things in the octave-up context, if they can provide a bit of an advantage to the most likely fundamentals.  I don't pretend to fully understand it, but the network between the emitter of the phase-splitter in the Foxx Tone Machine, and the base of the transistor before it strikes me as largely responsible for the robust nature of its octave.  And I'm guessing that this network provides a gain advantage for the mids.

[DougH]

For me the simple answer is "build a tycho" or "build Gus's other octave box". It's an interesting discussion to figure out what causes the squash in this circuit. But since that sound is the intention of the design, trying to "fix" it seems like a waste of time IMO, especially when there are plenty of nice octave circuits that are more tame, if that's what you are looking for.

[slacker]

I've played about with the circuit in a simulator and I think I can see why you get the massive compression and volume reduction on the attack when you've got the gain set high.

I'm not 100% sure I've got the explanation correct and this is using a sine wave as an input but it seems to fit with what happens playing the actual pedal. On maximum gain and a loud input signal you get almost a square wave out of the first opamp, this then gets inverted by the second and the 2 are then mixed at the germanium diodes. As the sides of the waves are very steep they almost overlap and cancel each other out so you just get a very narrow spike at the output. The clipping diodes then clip this into a very narrow near square wave.
As the input signal decays the sides of the waves become less steep and you get a wider signal on the output, as the signal decays even more you eventually get your classic rectified sine wave at the output.
So what you've basically got is like pulse width modulation, on the initial attack the output is mostly off so you get low volume and as the input decays the on time increases and so does the volume.

Here's a screen shot showing the output with different input levels, the green and blue traces are the signals at the junction of the 10uf caps and the 33k resistors and the red is the output.



If anyone's got a scope it would be interesting to try this out for real and see what the results are.

[DougH]

Thanks for posting the graphs. I've seen similar results when I've simmed this in the past. You can clearly see the compression.

[John Lyons]

Thanks SLacker!
Your similation and description follow what I hear the circuit doing.
When you play soft with the gain high the signal narrows and gets thin and low in volume.
As the signal dies the signal widens in frequency and becomes lowder.
Kind of like a slow compressor triggering a high pass filter...sort of.

So to lessen this effect would you offset the mixing at the diodes?
Maybe adding resistance at the diodes input from one of the op amps outputs
that would form a voltage divider?

thanks again for shedding some light here slacker!

john

[slacker]

No problem John, this is one of my favourite pedals and it's interesting trying to figure out why it does what it does.

I tried adding some resistance between the opamps and the diodes but it doesn't seem to do much. I also tried playing with the value of the 10k resistor in the feedback loop of the second opamp to see if changing the relative volume of the 2 signals would help but it didn't seem to.
The only thing that seemed to improve things was adding a capacitor in parallel with the gain pot to lower the gain of higher frequencies. In the simulator a 10n cap gives you a more trapezoid waveform rather than a square one, and because the sides have a shallower slope you get a wider output.  I haven't tried this for real though so it might not achieve much in reality.

[John Lyons]

Huh, I though for sure that putting resistance before the diodes making a 10 pot/33K voltage divider would offset the octave.
I have it all wired up like that but need to test it in a few minutes...got to feed the baby here..crying....

I'll try the 10nf cap too and report back.

john

[slacker]

Like I said I've only played about in the simulator using perfect components so in the real world YMMV as they say

[John Lyons]

Tried a 50K pot after the 10uf off pin 7 to the diode/33K junction. Voila! octave blend.
The voltage divided offsets the level of the two op amp sides to the mixer, enough to cancel the octave.

The cap across the gain pot worked as well in a different way. The level and frequency come into play here.
More fooling with this....


john

[DougH]

That's essentially the same thing as lifting one of the 33k's, John.

[John Lyons]

Well, sort of.
My whole point in posting was to get the octave to not pump so hard with the octave "on".
With the new pot I can turn the gain up and still have a good amount of gain and octave,
flip the switch and no octave, or twiddle the pot until I have the amount I want.
I wanted the full gain but somewhat of a threshold for the octave meltdown compression/pulse width.

The circuit as you know is sensitive to guitar volume and tone controls as well as picking intensity.
Between these parameters there are some nice options. Just wanted a little more variety..

john
.

[slacker]

Yeah you're right John that works, I must have been looking at it wrong when I tried it before. With a 50k in there output looks like a square wave with horns on so there's still some octave there but with a lot of fundamental mixed in.

At the minute I've got a stomp on mine that lifts one of the 33ks but I might try your idea and make that switchable instead, you can never have too many knobs in my book.
If you fancy trying some more mods I put a Superfuzz style filter on mine with a scoop pot. It's pretty cool especially with the octave cancelled you can get some pretty heavy sounds.