Fuzz experience

[Tek465b]

Hi ppl, good read on this section of the forum .

Anyway, i did a quick test using a npn si fuzzface (dunlop blue jh-f1 clone with 2sc945 hfe 200-210) a signal generator and an oscilloscope. using a 0.1v-p-p 2khz sinewave With fuzz at max and a 30k input resistor.
Just want to share my finding.

With a trim pot on Collector of Q1 and also on Q2.
First, no distortion/fuzz happen on Q1c no clipping,(well some of it show on the positive side, but its Q2 base that hit cut-off)
If we increase the resistance of Q1C it increase the gain (and reduce Q1C voltage)
(say from 1.36v to 1.28 q1 collector voltage or Q1 collector load from 18k to 33k) "more signal out of Q1c but still no clipping, checked on oscilloscope"
That drive Q2 harder,
Q2 clip on the negative side of the waveform. making the fuzz sound
When Q2 is driven harder, it will start distorting(clipping on both side of the waveform).(q2c at 4.5v)
If we make the collector voltage of Q2C lower(3.75v), this remove the distortion and increase the fuzz .
Or you can make it with lots of headroom by increasing Q1c voltage and keeping Q2c voltage around 5v or make it a squarewave distortion bi decreasing Q1C voltage and keepinhg Q2C at around 5v
Also, if we make Q2C voltage high like 5.75 volt, it will fuzz/clip on the positive side of the waveform instead. but it will start clipping the waveform on both side much sonner when you drive it hard.


Hopefully that make some sense. and it can help others understand the circuit and tweak for more or less fuzz and/or distortion

EDIT:
please note the voltage are for reference only, other transistor with different hfe will slighly offset the value, but the principle of operation remain the same.

As for my, i like 1.35 on Q1 and 4.5 on Q2. with 30k on the input.
This way i barely get any fuzz with my strat and low output pickup(more like a clean fuzz, with fuzz set to max). and the 30k make it wonderfull with the wah :O. Add the univibe and it do hendrix very well.

[yladrd61]

I like Q1 at 1.30V, Q2 at 5.20V

[daveweyer]

Tek,
It seems like you should be seeing quite a bit of distortion on Q1, especially if it is a high gain device.
If there is 1.3 Volts on its collector, that means there is only .7 volts left to pull down before the transistor is in saturation, i.e. the base will rest at .6 volts above the emitter, and that's all the more it can pull down the collector no matter how high you raise the base above that.
So if your first transistor has a voltage gain, say, of 10, then .1 volt of signal on the base will try to create 1 volt of movement on the collector, but the collector can only move down .7 volts till it's done, so .3 volts of the wave will be cut off.
With a grounded emitter stage, you have the most voltage gain possible in that stage, and if your transistor has a high hfe, you will have an almost continually saturated stage unless there is feedback to lower its gain.
The wave should be clipped on the negative half very similar to a half wave rectifier, until the positive peak also gets clipped.
If you are using a resistor in series with the base of Q1, then that will make the circuit into a current source, and it will be the current in the resistor which determines the gain, especially if there is a negative feedback signal also at the base. That appears to be the best way to change distortion characteristics in that kind of a stage, but with sufficient pickup output, that resistor begins to drive the stage more like a voltage source, and the negative half of the waveform clips again. It doesn't really matter how you drive the stage, once there is sufficient current at the base to cause a .7 volt movement in the collector, you have asymmetrical clipping.
That current source input was part of the West Coast fuzz mod.
I hope this makes sense based on what you said........

[Tazin]

Please excuse my ignorance when it comes to Fuzz Faces, but do they normally provide a rather large increase in volume when they are engaged?

[daveweyer]

Tazin,
The original idea was to use the voltage divider on the output (collector) of Q2 to try to make the output level of the unit equal to the input level, so you could switch the thing in and out of the circuit with some degree of loudness neutrality, only getting fuzz instead of a huge volume increase.
But the gain pot was added to allow you to get a preamp boost as well, at least somewhat.

I found out when I first got my hands on one of the originals that you could get a lot of gain boost if you wanted, and that sounded quite a bit different than just the fuzz.
It seems like extra gain was always part of the various mods which showed up in Jimi's gear, most folks leaving the voltage divider in the circuit to control the absolute amount of output boost. I put a pot in place of the voltage divider so you could get all the output boost you wanted, at the price of some noise. Low noise transistors were advisable along with low noise collector resistors and emitter resistors. Wirewounds were the quietest.

So only a moderate output gain boost on the original units. The real gain boost was in the transistor stages, but since they clipped almost continuously, it acted like a huge compressor--you know the sound, at low volumes a huge gain, but as soon as you strike a string the circuit reaches a volume which it can never go beyond, because it is absolutely clipped at that level. That's why you get such feedback when you're just holding the strings, HUGE internal gain, gain that stops at a certain level of output, an absolute brick wall limiter.

Using more gain at the output changes the whole equation.

[Tek465b]

Exactly. also the jh-f1 has that 10ohm on the fuzz pot/capacitor and the capacitor across the FB resistor this way a little more negative feedback are still in the circuit + the 30k input resistor... exactly like you said

A unloaded guitar output is about 0.1vpp for single note(or about 0.2-0.3vpp when playing chords), but it also has a high output impedance."vs the fuzz input"
I tryed to replicate this with the 30k (the signal generator at 0.1vpp am using it's output ompedance is much lower that a guitar at 5ohm"yes 5ohm", the 30k is effectivly turning it into a high output impedance current generator) and also because i like a 30k at the input when am playing.
I beleave most guitar have an output impedance(and not resistance) of 6k to 20k+ ohm.
I could re-do the test and set the input resistor at 6k and 0.3vpp to simulate a guitar with strong picking chords without any fuzz input resistor, but i think its gonna be a waste of time. Strat dont have a really strong signal after all and if you add the 10k-30k fuzz input resistor mod as it seem popular latelly , your never going to meet the criteria(input drive/bias) to get any nagetive clipping on Q1.

I can take a camera shot with the oscilloscope and post it here if you don't beleave me.
To me, with the blue jh-f1 fuzz circuit and Q2C at ~4.5v, all the clipping happen on Q2. i was surprized because everyone say the clipping happen in Q1 and that's what i was expecting.. but well.. Analog tektronix 465b 100mhz oscilloscope can't lie

It show very clearly on the scope. if Q1 was clipping on the negative side.. it will show on Q2(or simply the fuzz output) as positive side clipping. and for me, it do not. i also checked Q1c directly and no negative clipping there but i get positive clipping at Q1( but happen only when Q2 is in saturation fully turned on cause Q1C is not hitting the VCC supply, far from that). And yes my voltage polarity is correct, not inverted.(i know it can look like it is but its not)
except if i drive it really hard until i get clipping on Q1 with Q1c really low and then set Q2c voltage reall high around 6v to avoid clipping on q2 so its fuzzing and not squarewave distortion. then the circuit work like you describe it but it take some strong signal no input resistor and its not the normal state of operation i observed for the fuzzace at least for this circuit the dunlop blue fuzz. never seen a fuzzfaze biased at 1v at Q1 and 6V at Q2.

i was really bored as you can see i spent quite a bit of time on this loll.

[daveweyer]

Tek,
No the 465 won't lie. But what happens is the negative feedback through the resistor creates the opposite clipped waveform at the input, and that tends to make the negative clipped waveform look okay, but not totally okay, and if you analyze the harmonics you see still a lot of distortion--even though it looks like a good sine wave.
In other words it does not totally fix it, but lowers the gain enough that it keeps the transistor from pulling the collector all the way down to saturation.
Anyway, that was kind of the point of the West Coast mod, to get the second stage doing more of the distortion--so it would be a cleaner kind of distortion, and also to have more signal at the output to overdrive the Marshall too.
But you have to keep the signal on the base of Q1 from driving it hard enough to reach total negative cycle clipping.
The input resistor takes care of most of that as long as there is also some negative feedback.
30 K is pretty high for getting thick fuzz; what I found is that maybe 5 or 10K allows the guitar to have more of an effect on the first stage--the space of the last two or three numbers on the guitar volume control drastically alters the gain of the first stage and takes it from just kind of clean distortion, to totally nasty--by shorting out most all the negative feedback.
In my experimental units I used a pot as the input resistor because it gave you a lot of fuzz variations. I also biased Q1 directly from the supply so I could have a separate negative feedback path with only AC on it--that gives you network options.
On Jimi's equipment, everything had to be preset, there were only two knobs to use.

[Tek465b]

Also want to say thanks you for everything you share with us in the forum, i really appreciate your presence here. it has never been so exciting, thats why i started posting.

i tryed 5k-10k for the input(well using a 50kohm pot). but i can't get a good wah sound like this. 30k is the lowest i can go and still make the wah sound good.
Am not a fan of heavy fuzz sounds so it work good for me. i can just preset the bias for more or less fuzz from Q2.

EDÌT:
Yep, squarewave create harmonic, i dont have a spectrum analyzer at home but i know that.
The fast risetime of the clipped wave make the high frequency component of the signal.
Wich is why i love to use squarewave when doing troubleshooting work, can test the amplification stage or see the response of the filters etc to find faults. one frequency to test a wide spectrum of frequency.. am more into transceiver repairs. i love stuff in the 100+MHZ range. it get touchy

[daveweyer]

That's a problem alright, one way to have the best of both worlds is to put the resistor in the Wah. Then you can get that incredible change in the fuzz by just using your guitar volume, and still have a great Wah tone which won't make the fuzz squeal.
Another little trade secret.........

[Xplorer]

Hi guys,

i re did the traces the way i'm used to, close to the schematic visualy. it should fit the components and the fuzz box even if it's big and less efficient. i've included the fuzz revision.
here it is.

http://i.imgur.com/yeSfoJC.jpg

i'll report later as i'll build it.

[Xplorer]

What do you think guys ? we discussed quite intensively about the fuzz, in the Jimi Hendrix west coast mod section, it would perhaps be a good idea to move what's been written there to this fuzz section, to group all ( or most ) of what has been said ) .

for now, a couple pictures of the build, just the pcb made and a few components selection.

http://i.imgur.com/WXljw90.jpg

http://i.imgur.com/9HY1khG.jpg

Dave, i may not have the exact values. hope it'll be fine.

like :

40k instead of 37,4 k
810 ohms instead of 1k
a 200 or 250k pot instead of 150k

i'll use two resistors in serie to get 1,7 M

[Tek465b]

Cool you started building it. Am waiting my Ge from russia. will probably be next week.
I think its a great idea to have the revised version build, a really good idea with Ge tranny.
I made the non-revised version of the board only because i was almost done making the pcb when the revised version was posted.

Good idea to move the fuzz talk here. we should re-post the schematic (unrevised and revised) here for reference.

Now about the value. i know your question are to Dave but i do have some question me too

In the picture of the prototype we can see a 10u cap and a .01(can't read the others).
We dont see these value on the schematic. I guess the 10u is for Q3E, or maby there is one more 10u across it on the back of the board. or i assume the board is one of the very early prototype, and the schematic is a corrected version .
Am just curious.

now, 40k instead of 37.4k is not a big deal(correct me of a wrong).
Q1C will be lower and Q2C will be higher, You will have to play with the base bias to get the correct collector voltage anyway,

810 ohm instead of 1K, this is going to reduce the NFB, you could try and get a 22k that test in the low spec or a higher hfe for Q3. or change the 10k for 8.2k
Dave could probably tell more about it that i can.

The 150k pot with the 50n cap has a corner frequency at 21 hz.
With a 250k pot you should use a 30n cap (or 2x15n in parallel)= 21hz again. (yup right on)
Your probably going to get a slight bit more output on the high fuzz setting.
Or use a resistor across the pot. like 375k, in parallel with 250k it make exactly 150k.

[daveweyer]

For the 2N525s that I have the 37.4K value was better. You can just parallel a resistor across the 40K to get that value.
The .01 on the output was there on the original proto board, but you might want a .05 in case there is a circuit following the FF unit which has a lower input impedance than 100K, between that and the volume pot across the output your 3db down point should be around 40HZ to 80HZ.
The 10uf cap was on the proto input, but a 4.7 appears on the schematic. Adjust for low frequency response when negative feedback is at maximum.
The 100K and the 1.5uf provide a time constant for the DC recovery of the circuit, not too slow, but below the frequency where the AC feedback action occurs.
The 1M7 was appropriate for original 2N525 bias for Q1 voltage of between 1.2 and 1.4 volts, right about where you will want it for long sustain. But it's a taste matter, so you can play with it.

As for tone shaping, any network placed between the 2.2uf and the 22K in the negative feedback path will have the opposite of a normal passive function, since it is in the feedback loop. A Marshall tone stack can be placed there for instance, to control low, mid, and high, tones in the feedback path. An inductor can also be placed there for resonance.
Since the feedback network current will be opposing the input signal current, all manner of interactions can occur, and tones will change as the volume of the guitar is changed, for instance. Many possibilities...............

[Xplorer]

thanks Dave.

yeah, maybe two résistors in serie for the 37,4 k. but mine are different from yours, black, and they maybe have a different hfe for each. maybe a 50k pot there since it looks crucial ?

a whole tone stack for basse, mids, trebles between the 22k and 2,2 uf ? wow, interesting
ye maybe, later.

and a wah inductor for resonance ? i'm not sure of how such system is supposed to sound, i'm curious. do you have some example of fuzz resonance ?

[Tek465b]

thanks Dave.

yeah, maybe two résistors in serie for the 37,4 k. but mine are different from yours, black, and they maybe have a different hfe for each. maybe a 50k pot there since it looks crucial ?

a whole tone stack for basse, mids, trebles between the 22k and 2,2 uf ? wow, interesting
ye maybe, later.

and a wah inductor for resonance ? i'm not sure of how such system is supposed to sound, i'm curious. do you have some example of fuzz resonance ?

Inductor in serie with the cap will reduce high frequency negative feedback. wich in turn will increase high frequency gain/fuzz.
Maby not a wah inductor,you probably need something suited better.
A pot for Q1c seem a good idea.