Building a Roger Mayer Stone Fuzz

[willienillie]

^ I'm seeing basically the same thing.  But the "red" on the "1k" isn't nearly as obviously red as on the 5.6k.

The 5.6k is the input resistor.  47k fixed instead of a trimmer.  BCE pinout.  The output cap looks fairly small for 100n, but not impossible.

I bet it sounds like sh*t anyway.

[Rob Strand]

Having traced a lot of boards from photos I've learn't the hard way on more than one occasion not to guess values, only write down what you know to be true, and if all else fails wait until new info comes available.   

So for the cap:  On Gottfried's schematic he has actually written "C3 could be 10n".   So his style is to put guess values on the schematic and (usually) write notes at the bottom.  Sometimes he uses guess values from other similar pedals.  So maybe be did trace this from photos and could not see the 10n/100n.  So we don't *really* know the cap on the Stone Fuzz.

WTF??? I see a 5K6, 1K, 100K, 47K and the one that could be a 10K or a 1K ??

^ I'm seeing basically the same thing.  But the "red" on the "1k" isn't nearly as obviously red as on the 5.6k.

The 5.6k is the input resistor.  47k fixed instead of a trimmer.  BCE pinout.  The output cap looks fairly small for 100n, but not impossible.

Maybe 


In order to add more info I noticed Gottfried's schematic had voltages marked on it.


If we start with 0.74V on the emitter of Q2  the emitter current is,

IE2 = 0.74  / 2.2k  =  336uA        ; assuming the pot is in tolerance.

Now working from the collector side, the same current must pass through the collector,

IC2 ~ IE2

So we can actually calculate the collector resistance which matches the voltages on the schematic,

RC  =  (Vbat - VC2) / IE2   = (9.3 - 4.3) / 336uA = 14880 ohms

It's pretty much in the middle of 10k+1k and 10k + 10k = 20k.    So surprisingly the measurements haven't helped resolve the issue!

If the battery was flat, say 7.3V, then we get RC = 8.9k.   

Another factor is the tolerance of the drive pot could make it considerably different to 2.2k in either direction.  So if we assume a full battery of 9.3V we can estimate the pot value for the two cases of RC = 11k and 20k
RC = 11k   =>  Rpot = 1.6k
RC = 20k   =>  Rpot = 2.96k

Which are both equally bad estimates for 2.2k!

So given the battery is possibly a little flat, if I chose Vbat = 8.5V then with RC = 11k I estimated a pot value of 1.9k ohms.   That set of values looks quite feasible.  If RC=20k the pot estimate is way too high.

So the next step was to do a simulation.    It becomes pretty clear that in order to get VC2 = 4.3V  the resistors need to be 1K and 10k.

So from all that I suspect ultimately 1k + 10k makes more sense!

[jgenet99]

So is the dress blue?

The close up of the resistors is very strange...sometimes they both look definitely orange to me, and other times one of them looks red! Thank you Rob for finding the alternate gutshots, it's interesting to see the variation. There are some previous threads on here where people go through the circuit in detail, but as I said when I started, some of it doesn't seem to reconcile neatly. I'd love to have someone actually be able to review the circuit in detail.

As far as how it sounds, this is the guy who's the secret behind Jimi's sound after all. Everything I've read has been good, and once I biased mine second transistor to 4.5, I think it sounds great (that's with 100n on the closing cap, 10n just sounds too weird to me.)

Thanks everyone!

[Rob Strand]

There are some previous threads on here where people go through the circuit in detail, but as I said when I started, some of it doesn't seem to reconcile neatly. I'd love to have someone actually be able to review the circuit in detail.

Honestly, it's pretty much a silicon fuzz face, especially if you bias Q2 around half supply.  IMHO the thing that makes that pedal distinct is the 1nF across the base and collector of Q1.   1nF is a hefty value and it's going to darken the sound way beyond any slow old Ge transistors.


FWIW, more pics at.
https://blogs.yahoo.co.jp/tbmmb561/GALLERY/show_image.html?id=14788112

Can't see the "1k" resistor but the 4k7 input resistor is visible under the PCB.
The cap is hard to read.  From what I can work out the cap value is printed on the face shown.

[willienillie]

As far as how it sounds...and once I biased mine second transistor to 4.5, I think it sounds great

I was kinda joking, but the one pictured without a trim pot seems unlikely to be biased well.  Unless they sorted through transistors to find an ideal set, which wouldn't be very efficient practice for commercial production.  Maybe they only did the first few that way, then added the trimmer.

[bool]

...
1nF is a hefty value and it's going to darken the sound way beyond any slow old Ge transistors.
...

But the lower output impedance (12k-ish max) is going to "crispify" that darkened tone.

IIRC the JH tone had to a lot to do with dampening the guitar tone; the stories about coiled cables etc etc ... so the large CB cap sort of makes sense when viewed in that contexst.

[willienillie]

There's this Swedish Hendrix concert that somehow hasn't been taken down from Youtube:

He's only got a wah, a fuzz, and Marshalls.  His main dirt tone is the overdriven Marshalls.  He only kicks the fuzz on for the super fuzzy bits, then back off.  Seems almost any fuzz would've sufficed here.

If his wah was unmodified, not true bypass, it was probably loading the guitar signal down more than his cable would.

He messes with his knobs alot, seems confused for a while by the Gibson.

But this is only one concert, and a rather uninspired one at that.

[Rob Strand]

But the lower output impedance (12k-ish max) is going to "crispify" that darkened tone.

Sure but the darkness will still make it a bit darker.  (Perhaps more so if you back off the guitar's volume control.)

Anyway as far as the Stone Fuzz being different from a Fuzz-Face that feature would be more significant.  The 47k level pot probably lets more highs through.

[Rob Strand]

Well you think telling the difference between Red and Orange was hard.
I tried to enhance the two cap pics for the Stone Fuzz

*** Please click on the image, then click again to see the original images.  Otherwise the browser resizes the image making it even hard to read than it already is. ***

The cap with the steep view angle,

Original size


and the very dark one, which looks like there's nothing there at all,

Original size


From the printed text format and the rounded corners on the cap they may be Vishay caps, like this,



From what I can see,
- I believe I can see a "J" at the right of the top line separated from the text on the left.
- As for the value on the left it does look like a small 'n' at the end.
  If I had to guess I'd say there's not enough characters to make up "100n".
- One may have "100" on the second line which could be 100V.  I don't think it's 100n
  since the tolerance "J" is on the first line.

[Rob Strand]

After some perseverance I finally found a post which says the cap is 10nF.

https://tagboardeffects.blogspot.com/2014/03/roger-mayer-stone-fuzz.html

"gemather27 April 2018 at 21:41

Ey people, today i got a original unit and if you do not tell Roger i say to you it is a 10nF capacitor used. In my country the bc560 are not available and i think they are essential."

[duck_arse]

*** Please click on the image, then click again to see the original images.  Otherwise the browser resizes the image making it even hard to read than it already is. ***

The cap with the steep view angle,

Original size


and the very dark one, which looks like there's nothing there at all,

Original size


From the printed text format and the rounded corners on the cap they may be Vishay caps, like this,



From what I can see

summink went rong wiv yer image links there, Rob.

[Rob Strand]

summink went rong wiv yer image links there, Rob.

Yes the displayed pics and the links are not my original size.   They look a bit crappier than my originals (which aren't great anyway).

Not sure that is happening.    When I attach the pic it doesn't create a thumbnail with a link to the original.
It creates a rescaled image and when you click on the link it goes to the postimages home page.

The original pics are pretty small, less than 100x100, which seems to be screwing up how postimages  normally works.

[Rob Strand]

Despite the post I quoted saying the cap is 10n,  does anyone think the pics are showing 100n and not 10n?
To me, the amount of text there certainly looks like 10n; ie. "10n   J".  I'd struggle to get 100n out of that.  Don't forget the last character is an 'n'.   

[lcv]

Out of curiosity, I just put on the breadboard  a stone fuzz, according to the schematic (10K+10K , but input resistor 5K6 and not 4K7).
Surprisingly,  I read on the Q2 collector 4.48V. BTW the circuit is perfectly working.
So I measured some parameters that turned out to justify  Q2's collector voltage being on the higher range:
Vbatt=9.5v, Rpot=2.5K. Q2E=~0.65V
The transistor where  BC560C ,  from Mouser, measured beta ~ 430  https://www.mouser.it/ProductDetail/512-BC560CTA

I'm not drawing any conclusion from my little trial , but now  it seems to me  possible that  Q2's collector voltage on the schematic was correctly measured .
Maybe someone  can double check building it using  BC560C  transistors?

PS.  I'm going to simulate this in LTspice too.

[Rob Strand]

Surprisingly,  I read on the Q2 collector 4.48V. BTW the circuit is perfectly working.
So I measured some parameters that turned out to justify  Q2's collector voltage being on the higher range:
Vbatt=9.5v, Rpot=2.5K. Q2E=~0.65V

If we take your measurements,

The Q2 emitter current is,
IE2 = 0.65 / 2.5k = 260uA

Which then let's us estimate the Q2 collector voltage as,

VC2 = 9.5 - 260uA * 20k  = 4.3V       ; not far off your 4.48V

Alternatively we just compute the collector current directly,

(9.5 -4.48) /20k  = 251uA

Which is within 4% of the calculated emitter current.

So overall your measurements makes sense.

What was your trim pot resistance?

If we look at the schematic: the Q2 emitter voltage is quite high at 0.74V compared to yours.   The base voltage on Q1 is 0.61V which looks OK.   If we work out the Q1 base current as IB1 (0.74 - 0.61) /100k= 1.3uA.  Lets assume the Q1 gain is 500  then IC1 = 650uA.    To get that we would need the trimpot at RC1 = (7.5 - 1.37)/650uA = 9.4k  to (9.5 - 1.37) / 650uA = 12.5k  which is quite feasible for the 47k trimpot.

A high gain Q1 will bring down the Q2E voltage.  So *all* three factors in your case: the lower Q2E voltage, higher battery voltage and high 2.5K pot value,  help the biasing for the higher Q2 collector resistance of 10k+10k=20k.

I suppose the fuzzy area on the schematic is the battery voltage.    However, the effect of gain, which changes Q2E voltage, and the pot tolerance, all add into the equation.

As far as 1k+10k vs 10k +10k, the biasing is sitting on the fence for tolerance and circuit variations.

[Rob Strand]

I set-up this stimulation:

Transistors:   hFE = 500, VBE = 0.65V @ 1mA

RC1  = 56.4k  (trimmed for VC2 = 4.48V)
RC2  = 10k + 10k
Rpot  = 2.5k

Results:
When Vbat = 9.5V:  VC2 = 4.48V, VE2 = 0.628V, VB1 = 0.599V
When Vbat = 8.5V:  VC2 = 3.54V, VE2 = 0.621V, VB1 = 0.595V
When Vbat = 7.5V:  VC2 = 2.50V, VE2 = 0.613V, VB1 = 0.591V

So the trimmer, RC1, must be adjusted outside of the 47k range.

If keep the same transistor,, use the nominal pot value of 2.2k, and use RC2=10k+1k.
When, Vbat = 8.5V:   Adjust trim RC1 to 7.43k to get the schematic VC2 = 4.3V, and then,
VE2 = 0.84V, VB1 = 0.65V.
When Vbat =7.5V:  Adjust trim RC1 to 34.3k to to get the schematic VC2 = 4.3V, and then,
VE2 = 0.64V, VB1 = 0.60V

For these cases at least the trimpot is within the 47k adjustment range.  From VE2 the values are in the right ball-park to match the schematic but I can't match-up all the voltages without playing with the Pot as well.
-------------------------
OK I had the idea to adjust the battery voltage.
The same transistor, use the nominal pot value of 2.2k, and use RC2=10k+1k.
When Vbat = 8V,  adjust RC1 to 12.7k to get VC2 = 4.3V,  then
VE2 = 0.74V and VB1 = 0.63V.

So that's looking a lot like the schematic.  Does that mean the schematic was done with Vbat = 8V?  I don't know.

[lcv]

Hi Rob
Great job!
The Q1's  collector resistor in my build is 33K.
My LTsice simulation with that resistor , Vbatt=9.5V, Rpot 2.5K and BC560C model taken from this site : https://www.cordellaudio.com/book/spice_models.shtml, gives 4.36V on Q2's collector.

I hope you are planning  to prototype this (it's a 5 minutes work) with some C-grade PNP  and have a listen.
Obviously, keeping 10K+10K  is key to  preserve  the sonic results (whether these are good or bad, that is of course subjective ).
Using lower values would affect the loop gain and consequently some  important parameters  (e.g. input impedance at various Rpot settings and max gain level).

Thanks
Luca

[jgenet99]

Awesome work! I know that the original fuzz face used a much smaller resistor between 9v and output than between output and the collector of q2, which limited the volume of the pedal (they reportedly could on go a little louder than unity) so I assume part of this system is to increase the volume of the pedal and drive the amp? Are there other consequences that would result from the balance of the two resistors between the power and q2?

I know Roger Mayer was a big fan of gain pots that were at least 2k... does anyone know what effect this has on the pedal?

If the capacitor on Q1 was dramatically rounding off the highs, and the output capacitor is 10nf and rolls off the lows, would we then expect this to have a mid-hump?

Is the input resistor intended to focus and smooth the sound a little, sort of like rolling back the volume slightly?

Is anyone able to speculate how a lower gain transistor with the resistance adjusted to make the voltage would differ from a higher gain transistor that naturally biased to 4.5v? Would it just be louder, or would the tone be affected too? Would the emitter resistance largely even out differences in gain? If I had to adjust the resistors to two 5.6k (por ejemplo) to maintain the bias, would it preserve the overall character?

Thanks everyone! I'm learning more every day!

[Rob Strand]

The Q1's  collector resistor in my build is 33K.
My LTsice simulation with that resistor , Vbatt=9.5V, Rpot 2.5K and BC560C model taken from this site : https://www.cordellaudio.com/book/spice_models.shtml, gives 4.36V on Q2's collector.

I checked out the spice model and it matches up reasonable well with the datasheet.  BTW the on-semi and Philips data do not agree for this transistor.   Not a bad collection of models really.  When I run the sim I get similar results to you.

However, it brings up a dilemma!   According to the datasheet the VBE's for a given current are lower than that values I assumed for my simulation experiment.   So while your sim and build match-up relatively well, we end-up with voltages somewhat lower than those marked on the schematic.  On the other hand for my sim experiment I simply chose reasonable parameters for the transistor, then tweaked the collector resistor on Q1 to get the right Q2 collector voltage and *all* the values on the schematic line-up quite well.   So on one hand I definitely think the model on that site is a good representation of what the datasheet says; if you believe the datasheet.  I definitely believe your measurements.    But that leaves us with voltages which don't agree with those marked on the schematic!.    Very puzzling.   

As a side note I noticed the pics of the Stone-Fuzz have the pot set at 23k in one case and 47k in the other.  If we assume those are good settings, then those too line-up with your set-up.   So I guess from that we would conclude the resistors are 10k + 10k!

I'm not sure what transistors I have.  I've got a set I tinker with but there's more packed away.  Given the finer points of the problem it only makes sense to use the exact transistors.

[Rob Strand]

Are there other consequences that would result from the balance of the two resistors between the power and q2?

It's a very messy business trying explain the options.   I guess the ultimate goal is to adjust the resistors to get about 4.5V at the collector of Q2.  The method of adjusting the collector resistor of Q1 to achieve that, like the Stone Fuzz, is a very poor way to achieve that goal compared to just adjusting the collector resistor of Q2.   When you adjust the Q1 collector resistor you can end-up changing the current in Q1 by a considerable amount which can affect the tone (it affects the input impedance and gain).

I can say that if you vary RC1 over not too large a span, then adjust RC2 to set the collector voltage on Q2 the sound doesn't change enormously.

I know Roger Mayer was a big fan of gain pots that were at least 2k... does anyone know what effect this has on the pedal?

I believe it affects the gain.  At first glance it looks like the gain isn't affected since the ratio of the resistors on Q2 are kept in the same ratio.  However, the input impedance of Q2 goes up,  because it is operating at a lower collector current, and that makes Q1's gain go up.  Another way to look at it is the collector current of Q1 mostly flows into Q2's base.  Since the output current of Q2 is hFE2 * base current and the output voltage at Q2's collector is RC2 (Q2's collector resistor) * hFE2  * base current then by making RC2 higher the gain goes up.

When you factor in the different ways of setting the collector voltage of Q2 it starts to get confusing.  Since each method gives you a different result.  I analysed the 2k changes ages ago and I remember you end-up with problems comparing apples to apples.

If the capacitor on Q1 was dramatically rounding off the highs, and the output capacitor is 10nf and rolls off the lows, would we then expect this to have a mid-hump?

To some degree.  As bool mentioned when it clips some high-end comes back but it's very different to when that 1n cap on Q1.    I can't really comment on how the 10nF cap sounds with a 47k pot I don't think I've ever gone that extreme before.    For a fuzz-face with 10n + *470k* sounds  relatively balanced to me.

Is the input resistor intended to focus and smooth the sound a little, sort of like rolling back the volume slightly?

You need to do that experiment yourself.   IIRC the tone starts to take on a different character when you get to about 10k.  BTW, there are fuzz-face variants with pots in that position.

Is anyone able to speculate how a lower gain transistor with the resistance adjusted to make the voltage would differ from a higher gain transistor that naturally biased to 4.5v? Would it just be louder, or would the tone be affected too? Would the emitter resistance largely even out differences in gain? If I had to adjust the resistors to two 5.6k (por ejemplo) to maintain the bias, would it preserve the overall character?

I remember you could get close but nothing is ever *exactly the same* with the fuzz face.  The annoying thing is once you get past the big-ticket items like the bias voltage of Q2 and the input cap, you end-up with a whole heap of small things which are different but (to me) aren't enough to change the general character.  You have to play around with it for several hours and decide if you care about those small differences.   Maybe you like the one of those Frankenstein versions with pots for everything.

At the end of the day I'm not really a Fuzz-Face dude but I have spent many hours tinkering with it.