Sili-Face II: thoughts, questions, discoveries

[mongoonlypawn]

Finished last night.  I strayed from the schematic just a bit:
used 2.2nf input cap instead of .01uf
used 1.5K coming off Q2's emitter instead of 1K
I almost eliminated the 100uf cap coming off +9V because I couldn't hear any difference. 

I made the bias trimmer player-operable.  It yields a Q2 collector voltage of 4.4V one one end, 7V at the other.  Fairly subtle distinction, in my opinion.  Higher voltage = more aggressive sound.

Because I misread the layout, I initially failed to connect lugs 2 & 3 of the trimpot to each other.  Sounded great, actually.  Very Rat-like, but the bias was way off.

I'll be building another (I'm giving this one to a friend), but I'd like to be able to dial in the velcro rasp and gated Morse-code-Atari-fart mis-biased tones I could get with my Multi-Face.  Is there a way to do that? 

Finally, I didn't find the Sili-Face II especially conducive to volume knob clean-up / jangle.  Maybe it's my guitar and amp.  but are there any details I should watch out for?

Cheers.

http://www.home-wrecker.com/sili-face2.html

[brett]

Hi

I didn't find the Sili-Face II especially conducive to volume knob clean-up / jangle.  Maybe it's my guitar and amp.  but are there any details I should watch out for?

Easy. 
The volume roll-off happens because the output impedance of a pickup is similar to the input impedance of a fuzzface (about 20k each).  Rolling the volume down increases the impedance seen by the pickup.  The Sili-face has a permanent 18k input resistor in series with the impedance of Q1 (50k to 100k).  That's 3 or 4 times too much.  Replace the 18k resistor with a link.  Then reduce the impedance of Q1 by using a transistor with lower hFE.  A vintage 2N3904 might be ok (hFE=200, Z=50k), but modern ones are too high (hFE=350, Z=100k ).  Use an old BC107/108/109 or a 2N2222A with hFE around 200, or a BD139 (around 150 - this is probably your best bet for this circuit). 

Check out the schematic of the Axis Face for a standard Si FF with good clean up via the volume control.  It has no input resistor and a transistor with hFE=70.
cheers

[B Tremblay]

Try putting in the stock input cap and see if that helps with the guitar volume cleanup.

[petemoore]

  YAFF / AMZ...leetle 100ohmer under Q1 there.

[mongoonlypawn]

Thanks for the input, gang.  I'll certainly look into the input cap, the 100 ohm resistor coming off Q1's emitter, and the YAFF designer's contention that 2N5088s have way too much gain for a Fuzz Face.

Further, I found a bwanasonic schematic (circa 2004?) for a modded SF2 that looks interesting, if maybe a bit large for a small RadShack board.

Until then, how difficult might it be to make the SF2 bias-adjustable into sputtering ick range?  (See "Multiface" comment in original post.)  It looks like the two circuits bias from different points.  What say you?

[Gus]

Petemoore

A 100 ohm emitter resistor with a 100K collector what does it do? anything?

Tell me what a 100 ohm emitter resistor will do with a 100K collector resistor as in a sili-face.. What is the Ic and then what is the transistor re at that current what is the open loop gain?

[mongoonlypawn]

While I'm asking questions:

What does the grounded cap coming off +9V do?  I considered leaving it out when I didn't hear a difference on the breadboard.

I've read a post (I think on this forum) that seems to say that the 10K trimpot's wiper should connect to the output cap but NOT to the 4K7 resistor.  The schematic sure looks like those two trimpot lugs are jumpered, but I might be reading it wrong.  As noted in the original post, both configurations work and sound pretty good, but the jumpered version yields better bias readings and a "more-FF-than-Rat" sound, so I went with it.  Please advise.

[tiges_ tendres]

While I'm asking questions:

What does the grounded cap coming off +9V do?  I considered leaving it out when I didn't hear a difference on the breadboard.

Power supply filtering.

[PRR]

> eliminated the 100uf cap coming off +9V because I couldn't hear any difference.

With a fresh battery it may have zero audible effect.

A weak battery has high internal resistance, signals from output may sneak-back through the power rail to input stages, and change the action. A many-stage amp may be fine when batt is new and putt-putt-putt motorboat when battery is old. That may be hard to predict, you have to use it a while. So it is "good practice" to throw a fat cap on a power rail.

It is possible that the Sili-Face will function with a really-bad battery. Lower output, different flavor, but maybe no show-stopper problem. Or maybe not. I'd want an hour to ponder the question on paper. Pondering costs $10-$100 per hour. The cap costs a buck. In DIY you don't think too hard when a sure-cure is small cost.

And if you don't use a battery, you are probably using Wall Power, and need all the filtering you can find.

[trjones1]

A PN2369A transistor is a good bet for a sili fuzz.  Gain from the datasheet is 40-120, but in practice they're usually 110-120.  I just made a fuzz face with Q1=109 and Q2=126.  It's the only silicon fuzz face I've ever really grooved on.

[brett]

Hi
if I may answer Gus' questions concerning

the 100 ohm resistor coming off Q1's emitter

You have kindly alerted us to the siliness of some fuzzface designs.  It is clear that the 100 ohm resistor does at least 2 things: It INCREASES the input impedance (approximately by hFE x 100 hms = 20 k), and it LIFTS the base voltage of Q1 (by Ic x 100 = not much).

While not obvious to all*, these are bad things to do to a FF.

So Gus has asked "what does it do?" and I've suggested bad things.  So why did anyone put it there?  The answer is probably that they saw emitters on other designs and thought that the FF should have one as well.  IF the collector resistor on Q1 was 10k, then a 100 ohm resistor would at least limit the gain (10000/100 = 100), but with a 33k or 100k resistor, the gain is maxxed out anyway.

Scorecard for the 100 ohm resistor = 1 very bad effect (input impedance), 1 minor bad effect (higher base voltage) and 1 non-existant effect (gain control) = don't do it.
As suggested above, a 2N2369A and BD139s are good choices for Q1.
cheers

*especially those who haven't yet read Technology of the FF, Gus' posts or my comments above

[Gus]

I still use 100 ohms with a 10K collector for a few reasons at the first transistor.
http://www.diystompboxes.com/pedals/gusFuzzFace.gif

My point was when the collector resistor value is greater and/or the collector current is lower what does a 100 ohms do(with 100K and lower Ic).  I suggest reading and doing math or measurements.

[petemoore]

  Excellent suggestion.
 The cost of knowing what a resistor does greatly exceeds the price of 1 resistor, the same equation goes for the Fuzzface [if time spent studying = anything].
 The reason I put it up there and didn't even think it through.
 Math nomenclatures tend to confuse me, I write my own math sometimes, pretty ppoor math by conventional math standards only does ballparks when it's shining. I'm studying math courses...working on it.
 Now that I mentioned math, being 6'2'' tall is no help for striding over math hurdles, I've tried the math route and become repeatedly confused with it enough to choose to let math come to me instead [not to suggest this is any guarantee to work for anyone else]...I still strive for math answers and try it occasionally, make guesstimates all the time, understand which way with this R will tend to make that V go up and..wait to get my nose trimmed...something I can register, like...
 Mention YAFF in a 'can't get FF to cleanup' thread...and get called on it !
 ..Forcing myself to think up an answer...resistors are cheap, resistor length jumper wire is even less !
 Ok..here goes ! [caveat: do the math on your own] bumps the DC bias of Q1B up and tends to make it less susceptible to loading, drops gain slightly.
  Pinches the swing-room, with a 100k collector resistor bringing the collector bias to lower voltage...and the emitter voltage will go up...the operating points will be a smaller potential for swing.
  How'd I do ?
etc.  

[PRR]

> the 100 ohm resistor .... INCREASES the input impedance (approximately by hFE x 100 hms = 20 k)

Well, the transistor runs at about 0.080mA, so the intrinsic emitter resistance is about 30/0.08= 300 ohms, and the naked input resistance is 300*200= 60K.

100 in series with 300, or 20K added to 60K, is a small change, and likely "no change".

> and it LIFTS the base voltage of Q1 (by Ic x 100 = not much).

And the DC shift is about 0.080mA*100= 8 milliVolts, which is less than the spread of Vbe for different transistors and, as you say, "not much".

I think it is a complete waste of 12 cents.

> I suggest ...doing math

but, but... that's cheating!

[brett]

Hi
thanks doing the math better than my "guesstimates". 

At first glance the calculated input resistance seemed high (60k) - most measurements of Zin are about 20 kohms for Ge FFs and somewhat higher (30k) for Si.
But it makes sense (always does in the end).  Consider the Zin for hFE = 70 (Ge or a 2N2369A) - it's 300x70 = 21 kohms  ( ).  For hFE = 100 (other Si), then 300x100 = 30k ( ).

If we look at

20K added to 60K, is a small change, and likely "no change"

and substitute our "classic hFE = 70" Zin of 21k for the 60k in the comment, it becomes...

20k added to 21k, is a small change, and likely "no change".

IMO from 21 k to 41 k is not no change, and not even a small change.  It is a doubling of the input impedance from the tried and true values around 20k.

Again, I find support for my long-standing thesis about keeping the Zin of a fuzzface nice and low.  The Zout of the pickup is not much lower than the Zin of a fuzzface, and this produces important interactions.  Furthermore, the Z of the volume control can be used to interfere with  these interactions (the famous clean up).
cheers

[PRR]

> .... thesis about keeping the Zin of a fuzzface nice and low.

It is essential.

> Zout of the pickup is not much lower than the Zin of a fuzzface, and this produces important interactions.

The pickup is near 5K in bass and 100+K in treble. Naked, the 20K-60K input of super-simple fuzzes rolls-off the highs going in. That's what we want. Broadband or extended-treble fuzz is all nasty irritating fizz. Suck-off the natural string harmonics, then create artifical harmonics of the root pitch(es), that's more musical.

There's probably cases where this top-smash is too much or too little. Playing with series-base or emitter resistors (or transistor current: saves 12 cents!) is part of the fine-tweak.

> substitute our "classic hFE = 70"

Yes, using your values then 100 ohms may be a significant fine-tweak. (However the schem linked in the top of this thread calls for modern high-hFE pars '3905 '5088. As you say, that's maybe too high for some tastes.)

Mentioned once and then overlooked while hashing Gus' question:

> used 2.2nf input cap instead of .01uf

0.01u to 0.002u is a BIG change. 50:1. On the face of it, 9 octaves!! of less-bass. And the original value was already picked to shave the bottom octave+ of guitar. This could be the OP's "...any details I should watch out for?" It isn't "wrong", but I'd want to hear it with 5nF-20nF input cap before asking for more tips.

[mongoonlypawn]

"0.01u to 0.002u is a BIG change. 50:1."

I won't say it's NOT a big change, but the ratio is only 5:1. 

Anyone feel like addressing my "how can I adjust the bias into ugly gated sputtering at will"  query?

[Gus]

Don't forget the feedback resistor from the emitter of the 2nd transistor it is almost the same as one from the first transistor collector to base.

That resistor and the open-loop gain of the first transistor and the source RLC(passive guitar bass) set the gain/EQ of the first stage of the FF type circuit.

The input R is not just the hfe x re but it is reduced by the feedback.

One can think of it as an imperfect inverting opamp circuit with limited open-loop gain.  A circuit with a lot of gain(an opamp) will set the input as a summing node at about 0 ohms

So there is a mix of things going on with the first stage.

[petemoore]

  4k7 + 20k trimpot as Q2C resistor [= 4k7 - 24k7 R sweep].
 6k8 + 10k trimpot covers most transistor bias points though [offers 6k8 - 16k8 as bias R].
 Make the trimpot a panel mount knob.