Tube Amps / Music Electronics
|For current discussions, please visit Music Electronics Forum.||The sunn still shines online!|
|3/3/1998 11:26 AM|
||Fuzz Face for the people!|
The DC feedback tends to stabilize the collector current (Ic) of the first transistor (Q1). Independently of Q1 gain, Ic1 will be aproximately 0.25mA.
If Q1 gain is greater, it will need less base current (Ib) for the same Ic. With less base current, the base emitter voltage drop will be smaller. So the first stage wil hit what RG called "mush saturation" earlier.
For a AC128 with a gain of about 150 this is well beyond 0.1 V. Result: waveform will be much more assymetric, "mush saturating" even for low input levels. You back up your volume control and the sound doesn't clean up entirely. As the note dies, it is still dirty until it sudenly disappears as there was a built in noise gate. A "spongy" tone that is very funny, but is not very useful - unless you want to emulate an overdriven sitar or banjo.
For silicon transistors it doesn't matter that much if the gain is too high cause the base will still be polarized at about 0.5V, even with gain>500. The sound will be ugly in spite of the gain being hi or low since the saturation will happen later and will not be "mushy" at all, since the Si junction have a much sharper knee.
Now if the gain is too low, Vbe will be higher, and the overall gain will be lower. So less distortion and since "mush saturation" will arrive later, less assymetry.
It seems that the magic value of 84 for Q1 gain is the one that allows cleaning of the signal as the note dies (or the volume is backed) with an optimal level of asimetric saturation and gain. If you tried it you would hear the note sing, going easily in feedback, with nice dynamics.
The secret here is not only the gain, but mainly the biasing voltage of first stage's base (and the smoothness of the knee in the VbeXIb curve). Otherwise, it would be OK to use medium powered silicon transistors having the right gain (and less noise...)
If you have some Ge transistors at hand but they haven't got the right gain, here are some tricks I tried with good results:
1) Change the value of the 33k resistor (decrease it). This will increase Q1's collector current and so the voltage drop at its base. You will probably need to increase the value of the 1k also. Roger Mayer used 5k6 for the resistor and 2k2 for the pot in his "Classic Fuzz".
2) Put a Ge diode (Si won't work) in parallel with the base-emitter junction of Q1. Some of the base current will be shunted to the diode, so more base current will be needed for the same colector current. This has an effect similar to reducing the gain to about 50 (for an original gain of 150). It's less than the optimal value of 84 but here less definetely sounds better than more. I prefer the sound of this mod over tweaking resistor values, and it's also easier to implement. Attention to the polarity of the diode (must be the same of the base-emitter junction).
Another trick: If you like heavy metal tones , once you get Q1 working right, try using a silicon transistor for Q2 (BC557). This mix gives you the dynamics and tone of Ge (Q1) and the high gain of Si (Q2). Nice for Black Sabbath kind of dirt.
|And now, a word from our sponsors:
|3/3/1998 4:05 PM|
Very interesting work.
I've done quite a bit of simulation work but I have put off the hardware work until I can get my Omni-Fuzz-Face board done. This is a board that has pots for every resistor and dip switches for every cap on the board, intended just as a test bed for resistor values and for selecting transistors for sound, not on just gain. Not enough hours in the day.
>Now if the gain is too low, Vbe will be higher,
>and the overall gain will be lower. So less
>distortion and since "mush saturation" will >arrive later, less assymetry.
to a certain extent this can be compensated by diddling the value of the 100K feedback resistor for both higher and lower gain transistors, but because of the multiple duty this resistor does, you can't completely tune gain variation out.
>It seems that the magic value of 84 for Q1 gain
Wow! I never got that specific. You doing cut and paste or simulation?
>The secret here is not only the gain, but mainly
>the biasing voltage of first stage's base (and >the smoothness of the knee in the VbeXIb curve).
I would have said the doping level and abruptness of the BE junction and the resulting Rbb', but that's about the same thing. I have speculated about using a couple of Ge diodes to pull a silicon emitter below ground and inserting some small resistance into the base to simulate a germanium input with silicon, but ... dammit, hours again!
>If you have some Ge transistors at hand but they
>haven't got the right gain, here are some tricks >I tried with good results:
Good bag of tricks. I might add that for
(1) diddle with the feedback resistor a bit as well
(2) you can use small (10-100 ohms) resistors in series with the shunting diode to make it more smoothly variable in how much signal it sucks away. This can make the "apparent" gain less.
A few cautions;
Germanium is leaky and drifts with temperature like crazy, and is soldering heat sensitive as well. Always use heat sinks (even just a grip with needle-nose pliers) between the solder and the transistor body, or use a socket.
Simple gain testing on a DMM is a useful sorting trick, but the actual gain of the device to AC signals will guaranteed be less than that shown on the meter, because a DMM does a test of the collector curren with a fixes base current. Leakage adds to this, so the real gain is lower.
Also, germanium drifts with temperature - a lot! While you have a transistor in the DMM socket, just hold your fingers on it. The "gain" will start rising, and will go back down when your fingers are removed. For every transistor, gain is also collector current dependent (in fact, exponentially dependent, that's how they make exponential converters), so drift affects the real gain as well.
What's the real gain? That depends on what the temperature and current are.
About the best you can do is to get low leakage devices as best you can, and sort them into gain buckets. A true enthusiast would temperature-control the germanium devices with an on-board heater/controller, which can be done - but you have to be a fanatic...
|3/3/1998 5:22 PM|
|Marrk||Fuzz Face/transistor rating|
Maybe someone can help me with this question. I have some WEP 630 germanium transistors. The spec. chart on the back of the package says typical hFE is 110. I get readings between 21-25 for my transistors. Is there some way to interpret this reading in relation to a 9 volt circuit to see if they are really putting out near 110 or close to the "magic" number of 84.
I am not gonna go crazy over this, because I really like the sound of these transistors in the fuzz face pedal that I constructed. I just got lucky with these as they were the only ones I tried. Thanks for your input!
|3/4/1998 8:16 AM|
In the semiconductor industry, there are lies, damned lies, and "typical" specs.
"Typical" means that over the whole manufacturing run, if you measured every device, the average, median, mean, whatever center of the distribution will be at that number. This tells you nothing about the spread. One to 100 has the same "typical" value as 49 to 51.
There is probably some variation in the current that a DMM uses to measure the gain, as well; per my previous note, this will itself change the gain reading on any given device. Use gain measurements as a guide, but use what sounds good.
|3/4/1998 5:34 AM|
|anonymous||Re: Fuzz Face for the people!|
2) Put a Ge diode (Si won't work) in parallel with the base-emitter junction of Q1. Some of the base current will be shunted to the diode, so more base current will be needed for the same colector current.
Do I read this as a Ge diode from the base of Q1 to ground? The base-emitter junction is biased into conducting by a 0.65v applied to the base. Why doesn't this diode prevent Q1 from turning on? Or is it reverse biased in relation to the b-e junction?
I can see how this applies with a resistor in series with the diode, but what am I missing for the R=0 version?
|3/4/1998 6:51 AM|
If it is a Ge transistor, it is not biased at 0.6V, but at less then 0.2V. When you put the diode from the base to ground, BOTH the diode and the transistor conduct, so a fraction of the current that would go thru the base now goes thru the diode, and you need more current entering the base node to achieve the same colector current (=less current gain). There is nothing strange about semiconductor junctions in parallel. A lot of current source and current mirror circuits use this.
You could instead put a resistor to ground to shunt the current, but a diode works better because it is a nonlinear resistance very similar to the base resistance of the transistor.
As RG pointed, you can also put a variable resistance in series with this diode to control the amount of current being shunt and so the amount of relative gain reduction.
By the way, the magic 84 number is from a schematic I downloaded (84 for Q1, 115 for Q2) - I forgot from which site. In my own FF I use 40 (using the diode trick) and 180, sounds good to me.
The observations about voltage drop, collector current and assiyetric distortion I made were not based upon simulation. I protoed the circuit and measured with osciloscope, sine generator and DMM.
|3/5/1998 12:39 AM|
I have a question.
Is the FF the first affect the guitar goes into?
What guitar do you use, and with what PU's ?
I'm askin, because I have planned to build one
to use with a guitar. The guitar has DiMarzio
Dual sound PU's, from around 1980.
The output level from the guitar seems to
affect the tone/sound from the FF, as discussed
in the last 3 months on this BBS.
|Page 1 of 4||Next>||Last Page>>|