Interesting effect from C-B caps on Si FF...

[ThermionicScott]

Hey gang, I've just breadboarded a country&western radio receiverFuzz Face with 2N3903s from Aron (Q1 hFE=80, Q2=89; I've ordered a variety of other transistors in hopes of getting a ~120), and gone down the rabbit hole of tweaking it for bias and frequency response.  (Inspired by YAFF and others, I added an 82Ω emitter resistor on Q1 to adjust the gain down a tiny bit and widen the ratio.) According to R.G. Keen's article and this thread, putting small caps across the collectors and bases will serve to roll off frequency response and prevent oscillation/RF reception.

Now here's the weird thing:  the cap across Q1 seems to increase gain and high-end to my ears!  I've tried 68pF, 151pF, 220pF, and the effect is unmistakable when I insert/remove the cap with a pair of tweezers while playing.  Does this mean that preventing amplification in the ultrasonic frequencies frees up gain for audible frequencies, or does it indicate that I've screwed up something in the build?  I'm not aware of any obvious oscillation, it just sounds like a fairly low-gain fuzz.  (I've been avoiding higher-gain transistors to chase "clean up" from the guitar.)

Thanks!  Here's a crummy pic, ignore my 4049 project in the background.   

[antonis]

the cap across Q1 seems to increase gain and high-end to my ears! 

Here we have a typical ENT doctor client..

[mozz]

Gains are low for the silicon version. But if it sounds good..............

[Rob Strand]

ow here's the weird thing:  the cap across Q1 seems to increase gain and high-end to my ears!  I've tried 68pF, 151pF, 220pF, and the effect is unmistakable when I insert/remove the cap with a pair of tweezers while playing. 

Collector-base (CB) caps are often used to intentionally roll-off the highs.  IMHO, for the Fuzz-Face people focus on transistor gain too much and don't try CB caps.  It certainly helps smooth the sound out.   You can do it on Q1 and Q2.

Transistors have their own "build in" CB caps.    The old germanium transistors can have quite high CB capacitance.  In order to make a silicon transistor behave like a germanium you need to add some CB caps.    Exactly like what you have done.   Another option is to use silicon power transistors as these tend to have high CB capacitance but it's far more economical to add a CB cap to a small transistor.   The power transistors also have low gain (hFE).

Does this mean that preventing amplification in the ultrasonic frequencies frees up gain for audible frequencies, or does it indicate that I've screwed up something in the build?  I'm not aware of any obvious oscillation, it just sounds like a fairly low-gain fuzz.  (I've been avoiding higher-gain transistors to chase "clean up" from the guitar.)

That reasoning isn't correct.

The main effect, regardless of oscillations or not, is the CB capacitance rolls-off the high frequencies.

If you have oscillations *then* rolling-off the highs can prevent or reduce the risk of oscillations.   Adding the CB cap is *one way* to do that.    There can be many causes for the oscillation and many of those are knocked on the head by rolling-off the highs.   

In feedback amplifiers the addition of the CB cap is actually one of the better ways to prevent oscillation - unfortunately the explanations of why that's true gets very technical.  You might see capacitor C2.   This doesn't roll-off the highs that you hear at all but it prevents oscillation of the amplifier.    So quite different to deliberately rolling of the highs for sound like you are doing.



EDIT:
If adding the cap increases the gain then maybe you have a build issue, or, maybe you have oscillation and adding the cap is fixing it.

[ThermionicScott]

Gains are low for the silicon version. But if it sounds good..............

Chalk it up to reading too many articles and threads about germanium Fuzz Faces and what could be done to get silicon ones in the ballpark! 

[ThermionicScott]

EDIT:
If adding the cap increases the gain then maybe you have a build issue, or, maybe you have oscillation and adding the cap is fixing it.

Yeah, that's what I was wondering.  I'm on board with the reasoning for adding the caps, and what they *should* do.  I fully expected the caps to darken the sound and perhaps remove a little apparent gain, that's why it was perplexing that the opposite happened!  I used some of the lowest gain transistors I have, and tried to keep leads short and away from each other, so I just wasn't expecting oscillation... 

I'll double-check my wiring and get back to you guys, thanks!

[Rob Strand]

Yeah, that's what I was wondering.  I'm on board with the reasoning for adding the caps, and what they *should* do.  I fully expected the caps to darken the sound and perhaps remove a little apparent gain, that's why it was perplexing that the opposite happened!  I used some of the lowest gain transistors I have, and tried to keep leads short and away from each other, so I just wasn't expecting oscillation... 

I'll double-check my wiring and get back to you guys, thanks!

When you have RF oscillation you can often get away with quite small caps, small enough that it doesn't roll-off audio.   Obviously using an oscillating circuit as a base-line for tone is going to mess-up any comparisons as the circuit isn't working normally.   If you suspect oscillations you can also try adding 100pF to 470pF to ground across the input terminals.   Even if you don't end-up keeping a cap there it might help confirm oscillation is present.    Adding bypass caps (say 100uF in parallel with 100nF) to the power rails on a breadboard is also a good way to avoid weird behaviours.  Best to keep the caps close to the circuit.

If you have an oscilloscope then you can actually see the oscillation.   Without an oscilloscope you have feel-out the problem by trying a few remedies then decide on a fix.

[anotherjim]

If you put the C-E cap on Q1, there will be less treble feeding Q2. Depending on the Fuzz control, Q2 has negative feedback either voltage feedback to Q1 base or degenerative current feedback in Q2 emitter resistor.
So, I think adding the cap on Q1 can lead to an overall treble boost ('cause less treble in negative feedback) and this could be percieved as being louder.

What happens with the C-E cap on Q2? Should still reduce treble that can get fed back but you listen to Q2s E so ought to hear some treble cut but not the best place for RFI reduction.
A standard RF limiter for the circuit appears to be 1nF (I think it could be smaller) to ground on Q1s base where it cuts both the input and feedback highs. In addition to this, C-E caps to make Si more like Ge can be fitted to taste.