Decoupling capacitors in Fuzz Faces?

[gaussmarkov]

surely this has been discussed, but i cannot find it:  the basic (germanium) fuzz face has no decoupling capacitors for the power supply (e.g., R.G.'s technology of the fuzz face, http://geofex.com/Article_Folders/fuzzface/fffram.htm).  yet they appear in silicon fuzz faces like the JH1 and JH2 in R.G.'s article and in the runoffgroove sili-face ii (http://www.runoffgroove.com/sili-face2.html).

does this matter?  would the dallas arbiter fuzz behave noticeably differently if a 100uF cap were put in parallel with the circuit?  i think i understand what happens without the cap:  the current demands of one transistor appear as voltage changes in the output of the other transistor.  hence, there is a possible effect on the perceived sound of the circuit.

thanks in advance, gm

[John Lyons]

I'm curious about this as well. Sorry, nothing to add...

John

[petemoore]

does this matter?  would the dallas arbiter fuzz behave noticeably differently if a 100uF cap were put in parallel with the circuit?  i think i understand what happens without the cap:  the current demands of one transistor appear as voltage changes in the output of the other transistor.  hence, there is a possible effect on the perceived sound of the circuit.
 If there is a battery clip, and someone doesn't connect a PS to it, for the most part no, batteries put out ripple free DC, like a big capacitor the voltage generally stays quite 'constant' ie no quick fluctuations or rise/fall of voltage.
 If you try connecting a PS, things are probably going to be very different, and the PS ripple will get right into the audio signal with undesirable results.
 A BFC across the rails will even alot of this out, [maybe not enough for FF...depending on the PS and...], a small cap like .001uf in addition to a large
[has to be electrolytic, which don't necessarily handle the HF's completely]] cap across the supply, perhaps even a 100ohm resistor between PS and board V+...
 See the filtering on a Phase 90, IIRC it has the 100ohm>cap ripple filter.
  yet they appear in silicon fuzz faces like the JH1 and JH2 in R.G.'s article and in the runoffgroove sili-face ii (http://www.runoffgroove.com/sili-face2.html).

[JonFrum]

Two posts came in while I was typing.

I'd just add that I wouldn't call that cap a decoupling cap. As stated above, I think it's a filter cap for catching noise from a rectified power source. To decouple the two gain stages, I think you'd need a cap for each stage, with a resistor between them - a pi filter. That way, each stage would pull current from its own reservoir at the top of each cap.

[gaussmarkov]

maybe i should explain in more detail.  my question is not about filtering the power supply.  the issue is decoupling, as it is described in posts like these:

• http://www.diystompboxes.com/smfforum/index.php?topic=51502.msg386650#msg386650 (R. G. Keen)
• http://www.diystompboxes.com/smfforum/index.php?topic=53004.msg402729#msg402729 (Mark Hammer)

AFAIK the BFC does not only act as part of a low pass filter.  it also works as a decoupling capacitor to decouple the two transistors.  without the BFC, because there is significant internal resistance in the power supply and/or because a small resistor has actually been placed in the power supply filter, the current draw of one transistor can appear as a reduction in power supply voltage to the other transistor.  and that reduction in supply voltage will affect the output signal.

[gaussmarkov]

here's an LTSPICE demo.  the schematic:



this schem has constant current to both transistors' bases for the first 4ms.  the power suppy is ideal and there
is no ripple so everything just sits constant.  then i turn on a periodic current to the first transistor only until 12ms
pass and then i turn on a periodic current to the second transistor.  these currents cause the transistors to produce
periodic voltage changes at their collectors.

i put in the possibility to have a ripple current on the power supply and see the low pass filter action as
well, but the amplitude is 0V.  so the power supply voltage will be constant in these simulations.

the outputs with a 1uF capacitor in the power supply filter:



what you see is that the output of the second transistor isn't flat, even though it is getting change to the current
applied to its base.  that's because it's power supply is coupled with the power supply of the first transistor. 
edit: you also see that the second transistor's output is affected when the sine wave current is turned on.

the outputs with a 47uF capacitor:



now the capacitor is big enough to supply the extra needed current to keep the supply voltage approximately
constant for the second transistor while only the first one is drawing current.

[Skreddy]

The fuzzface sounds "browner" with an old-fashioned carbon battery.  It has that that saggy character, that mushy sound, that brownness.  An alkaline gives it a brighter, less interesting tone, as would a power supply no doubt.  If you add a big cap across the power rails, this current reservoir would no doubt mitigate the mojo factor of the imperfect, relatively high impedance power supply.

[gaussmarkov]

one more graph for the 47uF simulation:  this shows the current going through the decoupling/LPF cap C1 along with the output of the second transistor as above.  the cap can be seen buffering all the current variation needed to stabilize the voltage.

[gaussmarkov]

what you see is that the output of the second transistor isn't flat, even though it is getting change to the current
applied to its base.  that's because it's power supply is coupled with the power supply of the first transistor. 
edit: you also see that the second transistor's output is affected when the sine wave current is turned on.
for the second transistor while only the first one is drawing current.

sorry ... the first sentence should say "... even though it is gettting no change to the current ..."

[gaussmarkov]

fwiw, i tried putting a decoupling capacitor into a basic (germanium) fuzz face and i could not detect any difference in output.  i am going to guess that the current demands of the first transistor are small enough not to matter for the second.  a simulation might confirm this.

all the best, gm

[gaussmarkov]

well, an LTSPICE simulation shows that there can be a difference.  here's a simulation of
a 70mV 400Hz sine input into a full-on fuzz, with a 1uF (green) and a 51uF (blue) decoupling cap.  the
1uF value is too low to make much difference from no decoupling cap.  i put an internal resistance for
the power supply of 100 ohms, what i think a 9V battery might have.  the Q2 collector resistor is set
to get about 4.8V at that collector when everything is quiet.  for some reason the SPICE simulations seem
to require low resistor values relative to the recommended values.  others have found this also.



i think this difference could be audible. 

here's the LTSPICE schem that i used:



this is a modification of yeeshkul's schem posted here.  you can get the AC128 model in that thread.

so now i am thinking that a decoupling cap might smooth out the output of some fuzz face builds.

comments welcome, gm

[Skreddy]

I would much prefer the green output to the blue.

[gaussmarkov]

I would much prefer the green output to the blue.

that might be the better one.  in that case, one will want to experiment with lowering the value of the decoupling cap in a runoffgroove sili-face ii from 100uF.