The Shin-ei FY-2 Companion Fuzz is a pedal that has come back into fashion for its grating and unique sound. Even better, it's a simple circuit and the transistors it used are completely unremarkable silicon transistors. My friend at Trevor's Repair Café first called my attention to it with his own version, which I always called Trev-ei, so here's instead the Dyl-ei, my own reinterpretation of the circuit.
But what's there to change, isn't it simple already? As Trevor also observed, the "Fuzz" control isn't very effective, there's one capacitor which seems unnecessary and it would be nice to be able to control that evident notch caused by the bridged-T filter at the end. My solutions are different from his, though.
First, here's the schematic:
And, for reference, since there's not a dedicated article, Trevor's version:
There are some minor differences but I'll focus on the main ones. In general, I've made mine to be closer to the original, while the Trev-ei is more of its own thing.
The input stage isn't changed. It's fine and it's very simple, with its own off-centered bias.
Then we go to two different replacements of the "Fuzz" control. What's in the original can be a bit confusing, but I think it's best explained like this: we have a fixed series 50k resistor from the output of the second stage, and a feedback collector-to-base capacitor tapped across it. The fact that it's wired in series to the input 47nF cap isn't important. What this does is reducing the gain of treble frequencies in a very subtle way, to the point that I struggled to hear much difference. The most important consequence is that of the series 50k, which affects the T-notch frequency response and also drops a lot of volume.
What I decided to do instead is to add series resistance between the two stages: since the second stage has a very low input impedance and the first a somewhat high output one when it's not saturating, it's hard to make a traditional attenuator that isn't heavily tapered by the input impedance and can be set to achieve the original gain, so I opted with this configuration with a linear pot, which not only makes a voltage divider with the input impedance and has a nice sweep, but also smooths out the clipping caused in the second stage.
Trevor's alternative is instead a "gate" control, that attenuates the input signal a bit, but mostly affects the bias of the second stage, to the point where it almost cuts off, which is useful both for interesting sounds and to gate some noise.
The second stage is where it gets interesting. The original features a 100k resistor in the collector load, which is shunted by a 47nF capacitor. I don't have a good explanation for this, other than reducing current draw, but what matters is that frequency response isn't really affected (a 47n cap has 100k of reactance at just 33Hz), just bias. Since the actual AC collector load is then 47k, I changed the biasing resistor to 470k to bring the collector to the same, very low voltage of about 800mV as the original.
Trevor opted to keep the whole collector load instead, which is fine. For grounded emitter amplifiers gain is mostly a matter of bias, not of collector load.
The bridged-T notch at the end is another important point. The main feature of the original is that it's fed from the 50k "fuzz" pot, which causes an attenuation of treble frequencies, while the 50k volume pot is a negligible load. In Trevor's version the filter is driven directly from the second stage, one of the resistor is a much larger 150k, the other is a 100k pot and the 10k volume pot has an important effect together with the 8.2n output capacitor. This results in a notch that can go much lower than the original and is shallower at the original frequency, and bass being cut.
For my version, I tried to stay closer to the original, trying to achieve a similar frequency response without so much attenuation (a common complaint with the original is that it's not loud enough), but also keeping a variable resistor like in Trevor's. This is but one way to control a bridged-T filter: you can vary either resistor, or both together. In the dual version you can vary the resistance to ground, which moves the notch a bit but also makes it flat, or you can do what I did in the Caballero. This version works very well here I think, because by moving the notch up you can use it as pseudo-low-pass filter too, if you want to tame some treble.
Since I didn't have the series resistor taming the treble, I chose instead to use a larger output capacitor to balance that while letting more low frequencies through instead of less treble, to achieve a decent output. This is mostly to taste, but 22n seems a nice balance in general, while 100n was particularly nice for bass. If you decide to go smaller like in the original, to keep the treble at bay as with the 50k resistor, you can instead put a 1n capacitor from Volume 3 to ground. I know there's enough ear piercing treble as it is.
Here's frequency responses for all three versions. Mine and Trevor's are fed from low impedance, to simulate the certainly saturating collector, but as I said that mostly affects treble.
I think I've achieved the goal of making it loud enough: unity was at about 6/10 of the volume pot, but you can always use a linear one if you want to fool yourself into thinking it's louder.
Finally, here's a couple of demos: