Help with negative ground Fuzz Face problem

[soggybag]

I just built a negative ground Fuzz Face, it works fine when it's on. In bypass I hear a high pitched whine. It's not loud, and can be controlled with the gain pot. What can I do about this?

Turning the gain down a little nixes the sound. But I'd like to have full range of the gain knob without having to remember that it shouldn't be turned all the way up.

The sound only appears when the effect is bypassed. I'm using a true bypass with a 3PDT switch with an LED on one pole. I'm using the Tonepad wiring scheme 5 here: http://tonepad.com/getFileInfo.asp?id=76 I'm sure this is similar to the typical 3PDT wiring that is commonly used. The output is open when the effect is bypassed, so the sound must be sneaking in from somewhere other than there.

The input is connected to ground when the effect is bypassed. I have a two caps on the input with pot to "blend" one the larger cap. I notice the pitch of the whining sound changes when I fiddle with the cap blend pot.

All suggestions welcome, thanks for reading!

[Paul Marossy]

Huh, that's weird. I've only had those symptoms once when I built my 2nd Shaka Tube project. It was oscillating while bypassed and somehow finding its way into the bypassed signal. I fixed it by using a grounded circuit input true bypass scheme. But you are using a grounded circuit input true bypass scheme, so I'm not sure what the deal is. Maybe it's due to a bypass wire being too close to one of your circuit wires. Do you have some wires running parellel to eachother for a few inches? That's all I can think of.

[R.G.]

If it's a PNP fuzz face wired for negative ground, then it will probably work properly if you change it back to positive ground.

If it's an NPN fuzz face, it's a wire routing problem.

I've typed in over and over and over again - using the negative ground connection for naturally positive ground pedals can cause intractable oscillation. Sometimes this can be cured by changes/additions to power supply bypassing and wire routing. But sometimes not.

[soggybag]

Thanks for the reply. I have to admit I read that these have problems. Of course you never really see the light until you do it for yourself. In my case I breadboarded the circuit first without trouble. Of course I didn't test the bypass switch.I was very surprised that there would be noise when the circuit was bypassed!

[R.G.]

Thanks for the reply. I have to admit I read that these have problems. Of course you never really see the light until you do it for yourself. In my case I breadboarded the circuit first without trouble. Of course I didn't test the bypass switch.I was very surprised that there would be noise when the circuit was bypassed!

In my experience, it's sensitive to nearly everything. I theorize that the impedance of the con/per-verted ground upsets things, but after putting nearly a week of evenings into solving this "once and for all" I decided to take my own advice: Mother Nature was trying to tell me something.

The sneaky thing is that sometimes, maybe even most times it works. That makes everyone who hasn't hit the bad one yet think that reversing ground always works. And it always does - until it doesn't. 

[johngreene]

I think I know why negative ground PNP fuzzfaces oscillate. Let's see if I can explain it clearly enough......

I'll treat the transistors as simple switches in this scenario.

If we start with Q1 switched 'off' the following happens:
Q1's collector is low bringing Q2's base low. This turns on Q2 bringing its collector "high" (as high as it can get). This means the voltage at Q2's emitter drops. The base of Q1 is connected to the emitter of Q2 through a 100K resistor. So the 2.2uF input cap starts getting discharged through the 100K. It has a little trouble with this as the current coming out of the base keeps adding some current. Once the base of Q1 is low enough to switch it 'on', Q1s collector goes high. This switches off Q2 thus raising the voltage on its emitter and now the 2.2uF capacitor starts charging, this is a little easier because now the base current is helping you. Because of this the gain of the input transistor plays a role as well because it contributes current to the base. The lower the gain of the input transistor, the more current that flows out its base for a fixed amount of emitter current. Under normal conditions, this process converges and you have stable bias.

Now introduce an imperfect battery that has internal resistance. When Q2 switches 'on' it is dumping current through a (roughly) 8.5K load. This causes a slight drop in battery voltage. The battery voltage appears across the path of Q1 emitter -> Q1 base -> 2.2uF cap. The problem with that is when Q2 switches on, it reduces the battery voltage a small amount. The 2.2uF cap is going to do its best to not let the voltage across it change so it brings the base of Q1 even more positive which switches it completely off. Now it has to wait for the 100K resistor to discharge the cap before it can switch back on again. This causes Q2 to suddenly switch 'off' which increases the battery voltage slightly and the 2.2uF on the input of Q1 will pull the base down a little more negative keeping it saturated until the 100K can charge it back up again. It's this reaction of the input 2.2uF cap to power sag that causes the oscillation. You can short the input at the 2.2uF cap and if the short is connected to battery negative, it will oscillate, if you short it to battery positive, it won't.

If the input 2.2uF cap is returned to the same voltage as its emitter, it no longer has this differential voltage across it because of battery sag. Thus no oscillation. If the power source is a low enough impedance, then the sag is eliminated which also stops the oscillation. But it has to be a really low impedance voltage source. It only takes a few ohms of internal resistance to provide enough sag to cause it to oscillate due to the low Base-Emitter voltage drop of Germanium transistors.

So what can be done? Big, low ESR, capacitors would lower in the source impedance AC wise but you would need a lot of capacitance to overcome the 100K/2.2uF time constant of 220 mSec. So you may only succeed in lowering the frequency of the oscillation, but not eliminating it. Or you could run it from a good stiff power supply. Unfortunately this forces the signal into the base of Q1 to return -through- the power supply and this can introduce a lot of hiss even with a really good (quiet) power supply. Also, A big caveat of this is the well known fact that Duane Allman preferred the sound of a Fuzz Face with batteries on the verge of dying. Implying that this sensitivity to the internal resistance of the power source provides pleasing results in a positive ground configuration. So fix the power source problem, lose mojo. The best solution is probably just don't mess with it. Put up with the fact you can't run it on the same power supply as your negative ground pedals. And even then, it probably isn't going to sound as good as it will running from a battery. It's just the price you have to pay for good tone.

--john

[petemoore]

  I don't know if there's a quick fix for FF.
 Therefore, recommending the perfbread FF as excellent solution to all FF's only leaves 1 more thing:
 The other Fuzzface on perfbread so the 2 can be 'raced', compared until a 'winner' is determined, even though it is all compromise as with audio.
 Any good platform will do as well as designed, start with Axis Face.
 Then alter the design.
 In this way it can be known to a greater degree what paramaters the FF must be within, how far they can be stretched, and where ''the FF center'' you would choose is.
 It takes a bit more time to install the sockets where you want them, otherwise it is the same FF board. I recommend splices and 'large starting values' for the resistors instead of sockets and trimpots [where possible] this way you can always rework it and possible electromechanical failure points don't need to be installed.
 It's the non circuit board items for a 'box' that make the board/population seem almost trivial expense, switch/box/jacks/drillbits etc. is where the cost adds faster.
 I guess I'm saying you just need to build yourself a standard fuzzface [ie convert the one you have] and use it as a known to work comparator.
  Then experiment if desired, from there to see what else if other can be made desirable.

[ricothetroll]

Hi,

That's a great explanation, John Greene ! Thanx a lot.

Could another point of view be : Q2's collector puts current into the (-) of the battery, that causes a sag in its voltage. If the input is referred to this voltage, the sag appears to the base of Q1. That is actually positive feedback from the output : such a circuit reacts to vbe, and Ve = (+) of the battery. Then if the input is referred to the positive ground, the sag has no effect on vbe and thus no positive feedback.

Is this explanation right ? I'm actually still in some learning process.

Best regards.

Eric

[MikeH]

...the con/per-verted ground

I had to read that like 10 times before I got it, and when I did I literally lol'd.


Don't know if it was mentioned above, but every now and then you can get away with the negative ground (aka "positive source") PNP fuzz face by using the "BFC" method discussed in another thread.  Basically you take a Big F'in Capacitor (Like, 1000uF to 2200uF) and strap it across the power supply.  Sometimes you get lucky and this will quell oscillations.  But as RG stated - mother nature is not fond of this arrangement, and usually will let you know.

[R.G.]

I think I know why negative ground PNP fuzzfaces oscillate. Let's see if I can explain it clearly enough......
(long, *correct* analysis follows)

Yep. That is one failure syndrome. It's the first one I went to, and it works a lot of times. I have successfully fixed/pasted over some instances with this. I've done some modelling and a lot of bench work on this issue, and high source impedance is one of the issues. On the other hand, I have also had a very few instances where I applied the bypassing/source impedance fix and it still sang.

There are clearly several issues that reversed ground introduces, which is why there is no *The* fix that I've been able to find. Other than, of course, reversing the grounds back.

Your analysis is right. Now all we gotta do is find the others.

[johngreene]

I think I know why negative ground PNP fuzzfaces oscillate. Let's see if I can explain it clearly enough......
(long, *correct* analysis follows)

Yep. That is one failure syndrome. It's the first one I went to, and it works a lot of times. I have successfully fixed/pasted over some instances with this. I've done some modelling and a lot of bench work on this issue, and high source impedance is one of the issues. On the other hand, I have also had a very few instances where I applied the bypassing/source impedance fix and it still sang.

There are clearly several issues that reversed ground introduces, which is why there is no *The* fix that I've been able to find. Other than, of course, reversing the grounds back.

Your analysis is right. Now all we gotta do is find the others.

I've only seen this one mode of oscillation myself, but I haven't look at very many fuzzface's.