Acoustic 360 Fuzz

[Barcode80]

So, I've been throwing around the idea of doing as Viva ANalog did and build up a pedal version of the fuzz section of the Acoustic 360 bass head. Here is the original schematic:
http://analogguru.an.ohost.de/193/schematics/Acoustic_360_Fuzz_c-fin.gif

And here is my redraw for clarity (used 2n2222 in place of the low gain 2nd stage transistor, 5088s everywhere else):


So I have a few questions. First, what would you guys expect to be seeing voltage-wise on those transistors with the straight 25v biasing? It looks like (from my DMM readings) that the 680K resistor drops the voltage considerably, but that last transistor stage sees the full 25V (presumably gain recovery, high voltage for headroom?).

Also, what changes would you need to make to run this at 9V? I would think you could just run it off 9V, but it seems like the biasing with that 680K would drop the transistors to turn-off levels.

I have a layout for this that I would happily share, but first I'd like to explore a 9V version. Right now I have a 25v charge pump built up and working, so my initial plan was to run it off the original voltage.

Thoughts?

[PRR]

But it already runs on 9V. Near 'nuff. The funny connection of Q1 amounts to a 7V Zener. And Q4 only passes teeny signals, does not need big supply.

The only substantial change needed for 9V use is to omit Q1 and change R10 to about 5K.  

[PRR]

> biasing with that 680K

R10? 680K or 68K?

[Barcode80]

> biasing with that 680K

R10? 680K or 68K?

sorry, 68K. I mistyped.

But it already runs on 9V. Near 'nuff. The funny connection of Q1 amounts to a 7V Zener. And Q4 only passes teeny signals, does not need big supply.

The only substantial change needed for 9V use is to omit Q1 and change R10 to about 5K. 

I had a feeling that was the case. I'm curious about it only passing "teeny signals" at Q4 though. It's the output stage? Sorry, just trying to understand.

I knew a zener would be a good replacement for that stupid transistor polarity protection. Probably they had a surplus...

Q3 seems like it would produce no signal with only .8V at the collector. Am I wrong? Though I guess that's also the voltage at the collector with a 25v supply (given the combined resistance of about 215K between supply and Q3).

I'll give your changes a shot and post the results. Thanks for the input!

Anyone else?

[Gus]

The first two transistors make a fuzzrite etc type fuzz.  The fuzz circuit runs at about 7volts as PRR posted.  Make R10 a jumper remove Q1.  Power the circuit with 9VDC adjust Q4 biasing. This is kind of like PRR posted.

 Q4 is an emitter follower run at about 25VDC.  This circuit was part of a amp and the preamp section ran at 25VDC. One wants the fuzz section to have limited output before clipping that is why they maybe ran it at 7VDC.

If you search this has been covered before IIRC.

[Mark Hammer]

The first two transistors make a fuzzrite etc type fuzz.  

I had never really looked at it all that closely before, but when I looked at it, before getting to your post, I came to the very same conclusion.  Q7 and Q8 provide the same cascaded gain stages arrangement one sees in the Fuzz-Rite, Shin-Ei FY-2, and Orpheum (among whatI am sure are many others we don't know about yet), and the Attack control pans between the output of the first and second stages. 

Of special interest to some here is the use of a gain recovery stage after the "Gain" control.  Some folks in past who had built the FY-2 had complained of low signal output.  The post-level-setting recovery stage provides a boost that would be suitable for the FY-2, and follows in the tradition of the gain recovery stage on the the output of the Big Muff.

Things to play with here include the values of C25/26/27.

[PRR]

> I knew a zener would be a good replacement for that stupid transistor polarity protection.

It's not polarity protection. The power amp had a high voltage, they had a known-good low-voltage fuzz, they had to drop it down a LOT. Could be resistors, but drop will vary with parts tolerances. Could be a tight-spec Zener, but the exact value is NOT critical.

At that time (maybe now), a jellybean Si transistor was cheaper than a Zener. Especially if you _already_ have crates of jellybeans in stock. ESPecially if you have been selecting transistors and have a pile of hFE-rejects.

The C-B breakdown is reliably 6 or 7V. With the B-E junction in series, 7V or a bit more.

(Interestingly, this junction-pair is also well compensated for temperature, but I think the user would freeze/faint long before any difference in fuzz level is observed.)

> Q3 seems like it would produce no signal with only .8V at the collector.

The collector can swing down to 0.1V from emitter. So the negative signal swing can be 0.8V-0.1V= 0.7V. This is larger than a guitar signal. If this all goes into a guitar-amp input, we may have to knock it down a bit.

> only passing "teeny signals" at Q4 though. It's the output stage?

The attack pot selects 0.7V (approximately!{*}) signal from Q2 or 1.9V signal from Q3. The gain pot knocks that down an optional amount. I do not know pot values, there may be additional knock-down. So the signal into Q4 is well under 2V, probably under 1V. Q4 is run unity gain, so its output is 2V or 1V. For Q4 to stay very-clean, we need significantly more than 4V of supply voltage to it. But less than the 30V rating of the transistor. 25V was available. Use it.

{*} I don't know what the original transistors were. The design is semi-sensitive to Beta. The 5089 may be higher Beta than the designer found in his bins, so voltages and clip-levels may be different. If you can get voltages from a happy unmolested original, you can zone-in by tweaking R3 R6.

Working with 25V, Q4's distortion is almost unmeasurable, OTOO 0.001% THD. Working with 9V, Q4's distortion is higher, OTOO 0.05%-0.1% THD. Since this follows stages with 100%-1000% THD, we may not care about point-oh distortion or even several-percent distortion in Q4.

Q4 has large supply-crap rejection. Q2 Q3 have nearly none, and work with low-level signals. There's also a high-gain 3-stage sneak-path from Q4 to Q2, and such sneak-paths can motorboat. That's why I suggest retaining C1 and R10, just scale R10 for less drop. 5K and 100uFd will filter wall-buzz on the power feed, and also swamp 3-stage sneakage cheaper than the brain-pain needed to calculate stability factors. With a clean solid power feed, even a LM317 with 100uFd on it, Gus may be right that R10 may be zero and C1 is moot. It's a 60-cent difference and not _my_ money, so I tole you to do it. If you gonna build hundreds of these, you should check for possible cost savings.

> gain recovery stage after the "Gain" control.  

Is an emitter follower, no? Voltage gain is unity (0.98+).

It may appear to have "gain" if the gain pot has large value and the unbuffered load is a low value (lower than this buffer's ~~500K).

BTW, the input at R1 is a low impedance for guitar, ~~70K falling toward 22K. Heavy loading may be useful to mellow the string overtones before the fuzz splatters them all over the spectrum. This only works against a naked pickup, or somewhat against a guitar's volume pot, hardly no high-cut against a buffered source. Also this loading may be aimed more at bass pickups and spectrum more than guitar (duh).

[Barcode80]

> I knew a zener would be a good replacement for that stupid transistor polarity protection.

It's not polarity protection. The power amp had a high voltage, they had a known-good low-voltage fuzz, they had to drop it down a LOT. Could be resistors, but drop will vary with parts tolerances. Could be a tight-spec Zener, but the exact value is NOT critical.

Okay, bear with me, I'm not quite getting you on this one. C1 and Q1 seem to be nothing but polarity protection and filtering, not voltage reduction. What am I missing? Q1 just goes between supply and ground, I don't see the reduction in voltage to the circuit...

At that time (maybe now), a jellybean Si transistor was cheaper than a Zener. Especially if you _already_ have crates of jellybeans in stock. ESPecially if you have been selecting transistors and have a pile of hFE-rejects.

The C-B breakdown is reliably 6 or 7V. With the B-E junction in series, 7V or a bit more.

(Interestingly, this junction-pair is also well compensated for temperature, but I think the user would freeze/faint long before any difference in fuzz level is observed.)

True.

> Q3 seems like it would produce no signal with only .8V at the collector.

The collector can swing down to 0.1V from emitter. So the negative signal swing can be 0.8V-0.1V= 0.7V. This is larger than a guitar signal. If this all goes into a guitar-amp input, we may have to knock it down a bit.

Ah.

> only passing "teeny signals" at Q4 though. It's the output stage?

The attack pot selects 0.7V (approximately!{*}) signal from Q2 or 1.9V signal from Q3. The gain pot knocks that down an optional amount. I do not know pot values, there may be additional knock-down. So the signal into Q4 is well under 2V, probably under 1V. Q4 is run unity gain, so its output is 2V or 1V. For Q4 to stay very-clean, we need significantly more than 4V of supply voltage to it. But less than the 30V rating of the transistor. 25V was available. Use it.

{*} I don't know what the original transistors were. The design is semi-sensitive to Beta. The 5089 may be higher Beta than the designer found in his bins, so voltages and clip-levels may be different. If you can get voltages from a happy unmolested original, you can zone-in by tweaking R3 R6.

In fact, I have the unit in my garage. I may do this.

Working with 25V, Q4's distortion is almost unmeasurable, OTOO 0.001% THD. Working with 9V, Q4's distortion is higher, OTOO 0.05%-0.1% THD. Since this follows stages with 100%-1000% THD, we may not care about point-oh distortion or even several-percent distortion in Q4.

Q4 has large supply-crap rejection. Q2 Q3 have nearly none, and work with low-level signals. There's also a high-gain 3-stage sneak-path from Q4 to Q2, and such sneak-paths can motorboat. That's why I suggest retaining C1 and R10, just scale R10 for less drop. 5K and 100uFd will filter wall-buzz on the power feed, and also swamp 3-stage sneakage cheaper than the brain-pain needed to calculate stability factors. With a clean solid power feed, even a LM317 with 100uFd on it, Gus may be right that R10 may be zero and C1 is moot. It's a 60-cent difference and not _my_ money, so I tole you to do it. If you gonna build hundreds of these, you should check for possible cost savings.

> gain recovery stage after the "Gain" control.  

Is an emitter follower, no? Voltage gain is unity (0.98+).

It may appear to have "gain" if the gain pot has large value and the unbuffered load is a low value (lower than this buffer's ~~500K).

BTW, the input at R1 is a low impedance for guitar, ~~70K falling toward 22K. Heavy loading may be useful to mellow the string overtones before the fuzz splatters them all over the spectrum. This only works against a naked pickup, or somewhat against a guitar's volume pot, hardly no high-cut against a buffered source. Also this loading may be aimed more at bass pickups and spectrum more than guitar (duh).

What about the idea of replacing the output buffer of Q4 with a unity gain opamp buffer?  I'm with you on just scaling down the resistor. I'd rather buy bulk resistors than bulk LM317's

[Barcode80]

Okay, I made adjustments based on the above discussion, and came up with a quick and dirty test layout. Second eyes please? I think it checks out in ExpressPCB's net checker unless I missed something.

[PRR]

> I don't see the reduction in voltage

Q1 is working as a Zener Diode, NOT a transistor.

Zener regulators:
http://hyperphysics.phy-astr.gsu.edu/hbase/electronic/zenereg.html
http://www.allaboutcircuits.com/vol_3/chpt_3/11.html
http://www.wisc-online.com/objects/ViewObject.aspx?ID=SSE7005

Transistor B-E junction as a Zener:
http://www.electronicskb.com/Uwe/Forum.aspx/design/6146/diode-connected-transistors
IC Array Cookbook, Walter G. Jung, 1980, page 28, search ABE
Designing Analog Chips by Hans Camenzind, page 1-28, download  3MB PDF file

[Barcode80]

Gotcha.

Okay, I built this up, did some tweaking, and ended up with an 8.2K resistor to get the ~7 volts you specify in your calculations. I'm pretty close to your voltages everywhere else, except I'm closer to 5 than 6 on the emitter of Q4. I'm getting pretty gated, buzzy fuzz with very low output and almost no sustain. Not really close to the original sound.

Any thoughts as to what is going on? I'm wondering if I need to omit or reduce R4. I was looking at the Fuzzrite schematic, since the second stage is almost exactly like it, and noticed the 470k value there. The two feedback resistors on Q3 combined amount to 1.15M, which seems awfully high.

[PRR]

"Gated" on either/both sides of the "attack" pot?

You can't omit R4 or C3 will suck signal.

What were the *original* transistors?

[Barcode80]

Yeah, sounds pretty much the same at both ends of the attack pot. The original transistors were 2n2926 for Q1, Q2, and Q4, and 2N1306 for Q3. I also built the 25 volt version with the stock values, with 2n3904s all the way through, and got the same sound. So I assume it may be in the transistor choice.

[PRR]

2N2926 should have a color code.

color -- hFE
Brown - 35-70
Red -55-110
Orange - 90-180
Yellow - 150-300
Green - 235-470

What you got on the original?

2N1306 ?? That's a classic low-noise audio part. Germanium! hFE at 10mA 60-300. Ft 10MHz (much slower than Si parts).

[Barcode80]

Looks like the 2n2926 is only present in the "zener" spot. The others are 2n3391. Otherwise the original schem I copied is dead on to the original.

I pulled the fuzz board out of the original and wired it to my 25v supply, and it sounds SICK! It's awesome. Looks like 2N3391 for Q3 and Q4, and the Ge 2n1306 for Q2. I mistyped on the redraw of the schematic because the hand drawn one in analogguru's archive i based it on is pretty hard to read. R10 should be 6.8K.

All the other values are dead on to the original. I also found a better drawn repro of the factory schem which is much easier to read, and also is exact. Fuzz section is the bottom middle.

So, how would one go about rebiasing Q2 for a silicon transistor?

[PRR]

In a simple clean amplifier, Ge to Si conversion is simple; in a circuit totally designed to work "bad" in a "good" way, it's more art than science.

Since you've posted three different transistor-types, typo resistor values, and two schematics which don't quite agree, I'll leave it to you.

Maybe JC's page can help: http://www.lynx.bc.ca/~jc/pedals360.html

Someone sure liked transistors as Zeners.

[Barcode80]

In a simple clean amplifier, Ge to Si conversion is simple; in a circuit totally designed to work "bad" in a "good" way, it's more art than science.

Since you've posted three different transistor-types, typo resistor values, and two schematics which don't quite agree, I'll leave it to you.

Maybe JC's page can help: http://www.lynx.bc.ca/~jc/pedals360.html

Someone sure liked transistors as Zeners.

Let's not get testy, I've removed all of my schematics in the thread thus far to avoid further confusion. I've verified the part numbers, and the piazza schematic is verified part for part, including those transistor markings and the 6.8K resistor on the supply. I'm currently updating my redraw of the isolated fuzz section, and I'll repost it as soon as I do.

To be fair, a lot of what I posted was based on analogguru's schematic, before I had pulled open the original. I was just trying to make out a very hard to read schematic and made some typos. Geez.

EDIT: Here is the schematic with the adjustments culled from the piazza schem and the original board.



Yeah, it certainly looks like someone liked transistors as zeners a whole lot. Maybe a bulk lot they got cheap? probably convenience more than anything else. I've got some NPN low gain Ge's on the way, I'm guessing I should be able to dial in the bias by ear for Q2 with a trimpot on R7 or R8, eh?

By the way, here are some shots of the original board now that it has been removed from the unit:

[Gus]

Search for "fuzzrite schematic"  There are threads here and other places.