The Univox Super Fuzz is indeed super - possibly my most favourite octave fuzz ever...   maybe my desert island fuzz (at least top three).  It's a legendary fuzz sold under many different names with early beginnings in Japan as the Honey Psychedelic Machine.

UNIVOX SUPER FUZZ - VERO LAYOUT

note:  with the exception of the cap from Q4 & Q5 to ground, all 10u caps can be 1u without much effect on the frequency response.

Sixties version (grey box)

Seventies version with trimmer (super fuzz on stomp switch)

ORIGINAL UNIVOX SUPER FUZZ SCHEMATIC

There's two versions around - one is referred to as the sixties version, the other is the seventies.   

The sixties version has a 1n cap across the collector and base of Q2.  

The seventies version has a 2n cap in the same position mentioned above, and the 10k internal trimmer to adjust the octave

SUPER FUZZ LTSPICE ANALYSIS

note:  all responses below use the first schematic, without the trimmer / 2n cap unless marked as such.

Octave effect

The octave effect is probably clearer on the actual scope screen grabs at the end of the post, but here it is anyway.

Frequency response - pre-output stage

Output tone 1 and tone 2

Caps

First off - with the exception of the 10u cap in the emitter of Q4 & Q5, all 10u caps can be 1u with almost no impact to the frequency response.

As a reference, this is the output of the boost stage with a few different cap values.  At 40hz there's less than half a db difference between using 470n caps and 10u caps.  

1u is a safe bet and you can easily avoid electrolytic caps with this value - why avoid electros?  Well at least try and avoid them on the input and output caps as electros leak, contributing to any potential switch pop.  The bigger the value, the higher the leakage.

If you do want to play with the frequency response, here's two caps that may not come to mind immediately.  The 10u from the emitter of Q4 and Q5 (C9) have a pretty large effect on the bass / low end response.  The 1n cap on Q2 (C4) has a direct effect on high frequencies.

Trimmer

I'd never looked at the 10k trimmer properly until recently.  It's a common mod to adjust the amount of octave to account for using unmatched transistors for Q4 and Q5.   I'd built pedals using it, and could hear the differences, just never ran it on spice before.

As you can see, the trimmer effects the frequency response and output in certain positions.  No doubt a result of less octave being produced.

The trimmer does change the total resistance from the base of Q4 and Q5 from 22k to anywhere in the range of 22k - 32k depending on where you like to set it.

Alternatives to the stock trimmer set-up:

Reduce the 22k resistors to 18k - this way when the 10k trimmer is set to the middle position (5k), the total resistance is 23k, which is closer to the original value of 22k.

Another option is to only have the trimmer on the base of a single transistor, to adjust it to match the other.  Same values as above work just fine.   

SUPER FUZZ OSCILLISOPE BENCH TEST

These are screen grabs from my oscilloscope. The input signal is a 440hz sine wave, approx 130mv TRMS

This is the output of the first gain stage - which is just a boosted signal that has some clipping already.

The fuzz was sitting on around 50% for all below.  I didn't include the output of the long-tailed pair, as it just looks almost the same as the output without the notch filter (I may have even gotten these two images mixed up that are so similar).

This is a reasonably hot signal, I have other Super Fuzz builds where it's still a fairly clean sine wave at this stage.

This is the collector of Q3, which is where the signal is split and inverted from both the collector and emitter, which feeds into either side of the long-tailed pair.  

Now here's the emitter of Q3, which always seems a bit different to the collector

This is the output with the notch filter on

and again with the notch filter off - but who does that, right?

This is fuzz on zero, with the notch on - quite the strange looking wave