I've spent a great deal more time with the Fz-1 circuit since my last round of posts and have arrived at a place I'm pretty happy with sound wise, so I thought I'd share my conclusions here. Before I do, I'd like to state that my knowledge has not gotten to the point where I can mathematically explain most of this, or do a proper engineering circuit analysis (like a loop analysis?), so if anyone can shed some light from that aspect, please do as I am eager to understand as fully as I can.
There are a lot of places and layouts out there that advocate using a trimmer on the collector of Q2 with various arguments as to why. In my opinion, after much experimenting, this is a rabbit hole to stay away from. In short, it is possible to get decent sound using this method, but you more often than not end up with a pedal that is not as loud as it can be and a Attack knob that has extremely limited functionality.
I eventually came to the conclusion the best place to instal the trimmer is between the -3 volt rail and the base of Q2 or, put another way, in place of the 470K resistor. I have yet to find a instance where a 500K trimmer is not sufficiently large for the job.
A little about Q1 and Q3…all pretty much garnered from SmallBear's article on the Fz-1a and my own experiments.
Q1 is a emitter follower and therefor a input buffer and impedance matching device. Great little website that explains some of this:
http://hyperphysics.phy-astr.gsu.edu/hb ... cc.html#c2
The upshot of all the technical info is this…you want as close to 1 volt on the emitter of Q1 as you can get. You might be able to achieve over 1 volt, I have had some transistor selections that hit almost 1.4 volts, but the range you're really looking for is between 0.8 volts and 1.2 volt. Less than this and the current flowing into the base of Q2 is too small, more than this and you start to introduce a lot of highs that you really don't want in the sound. I have found, given the range of around 0.8 volts to 1.2 volt on the emitter, it's pretty easy to select a transistor to hit the marks and just stick with the stock resistor values in this stage. Someplace in the 60-80 gain range with around 150µA leakage (that's microamps in case in my little mu symbol doesn't translate on your machine). That's more a minimum leakage guideline…but too much over that and you might start introducing hiss and other white noise, so watch out. There is no real benefit to going over 60 on the gain in my experiments, BTW.
Q3 is, mostly, there to amplify. It can and does add to the sound and there is some distortion introduced there depending how you set it up, but it is not the major contributor to the fuzz effect. To get this stage to work you have to have leakage. It's way more important than gain. My preference is to find a transistor that has a good amount of both, however. I like to try to use something with at or over 100 Hfe and about 250µA of leakage. It's not necessary to mess with the original resistor values in this part of the circuit either, stick with the stock values and just try different transistors till you get about 2.5 to 2.7 volts on the collector (my personal preference and measured with attack on maximum). You can also go down to around 80Hfe, but I think they sound best with someplace around 100. It's the leakage that turns this on and starts to drop the collector voltage where you want it. Remember…as the collector comes down in voltage from the power rail, the amount of current flowing increases….you want to have a good amount of current flowing, but not necessarily the maximum amount that could flow. The reason is, if you keep it near saturation (which would be a collector voltage of 3 volts as I understand it, or equal to your power source), you get some cool clipping in the signal that adds nicely to the overall fuzz sound. Or, so my technically challenged brain seems to think. Am I wrong in this understanding? It sure does seem like what my ears hear.
OK, so that leaves Q2 and this is where the magic happens. Once I got away from trying to set this stage up by adjusting the 1K5 collector resistor, I started to get results I really liked. As stated, I prefer to use a 500K trimmer in place of the 470K from the power rail to the base. All other values in this stage stay stock. I have found that if I use a transistor that has right about 200µA of leakage and a gain around 90 or so (think 85-90 Hfe on up) I can get some kick ass sounds and sounds that match many of the known Fz-1 recordings out there. How I bis it is pretty simple too. I set the Attack pot on MAXIMUM and adjust the trimmer till I'm getting slightly over 0.2 Volts...try for less than 0.4 Volts on the collector, but it will be touchy and kind of difficult. When you do this, you should see in the area of 2.7 (+/- 0.2) Volts on the collector at the MINIMUM Attack setting.
It is my experience that if you're sporting around 200µA of leakage and set it up this way that the sweep of the Attack pot becomes a very useful tool in this pedal. At minimum the sound is akin to medium/strong overdrive kind of mild distortion…as you start dialing it up you begin to hear more and more fuzz, than more and more compression. Eventually, as you near maximum the signal becomes so compressed and fuzzy you start to hear a volume drop. There is definitely a sweet "range" (not spot) with this method and it's usually between (roughly) 8 and 3 on the Attack knob. Unity gain, no matter the Attack setting, is almost always around 1PM on the volume knob.
Last, a couple notes. I have no idea why Gibson chose to go with the 470K. It makes the pedal almost impossible to bias to where it will work at all, let alone well. I went through a process where I held everything constant and only changed out the Q2 transistor (using all the same make and model) with leakages ranging from below 100µA to over 1000µA (that's 1 milliamp!). It wasn't until you get over 500µA leakage that you start getting good strong sound and you need to get near 1000µA to start to gain some functionality from the Attack pot. They must have had some seriously shitty RCA 2N270's in mind when they picked that 470K. I've found, using my method and RCA 2N408 transistors (the replacement for the 2N270) that my trimmer usually ends up around 100K. I have one old black can RCA 2N270 that has leakage so high, and it's so unstable, that I have never bothered to get a reading on it using RG's method. It starts off around 800µA leakage and just shoots up from there like a rocket. I put that in for Q2 and it produces some awesome fuzz sounds and gives the attack pot almost as much functionality as my method…but it's so damn hissy and noisy when you're not playing it's unusable.
Here is a clip of what mine sounds like. I basically play the same lick from 7PM (minimum) through 5PM (maximum). It was recorded with a portable Edirol digital recorder set in front of my Fender Champ 12 and using a Gibson SG on the bridge pickup. The compression and limiters on the Edirol are off and the amp volume is set very low.
" params="color=ff6600&auto_play=false&show_artwork=true&show_playcount=true&show_comments=true" width=" 100%" height="81" iframe="false" /]
Here is a picture of one of my builds inside and out.