Fuzz, Transistors and taking my knowledge to the level slightly above nothing.

[rocinante]

Hey all,

Ive spent roughly the last year lurking on here and trying to learn how to build and fix pedals, but now im really wanting to understand a bit more about how everything works and how to debug builds.  In particular a problem ive run across a few times when building fuzz effects is a huge amount of noise or oscillation or static when i ramp up the texture/gain/whatever pot up the last few percent. It generally makes a fuzz sound when playing, but then when there is no guitar input, there is some aweful stuff going on.  Anyway im not asking for a particular solution to a particular circuit, but more of a theoretical vague question, i hope that isnt too infuriating. (although i will use the tim escobedo tripple fuzz that i am working on as an example).



I had to sub a BC327 for the 2n3906, but otherwise everything is per schematic.  And the noise occurs when i shunt the mpsa18 emitter to ground (i know i could probably just add in a small resistor but i figure if it is designed like that, then it is something i am doing wrong).

Basically i am figuring this has something to do with biasing transistors, especially when i have to sub a transistor or something for one that i actually have.   So i have played around with a bunch of things trying to figure out the solution myself, but at the moment it is all just numerous stabs in the dark.  Ive also been trying to read as much as i can about transistors and biasing, but i am no EE and have no training apart from what i can find on here and the net.

So my Questions:
1) Is this type of noise typically related to the transistors and misbiasing
2) what can you typically do to rectify the differences when you replace the recommended transistors with another.
OR
3) What are the things to look out for on the datasheet when looking at similar transistors? Eg. what is going to make it 'not work' or not function as intended.

I hope these questions havent been asked too many times and im not asking too much without enough knowledge.  Im just trying to make a practical link (in my mind) between the datasheet and the effect it has on the circuit.


Thank you in advance!
Steve

[digi2t]

From one "non-EE" to another  ...

I just finished a clone of the Spaceman Effects WOW Signal Fuzz. The front end uses a similar arrangement, but with a low gain germanium transistor after the trippler. You can find all the info from my build here; http://www.diystompboxes.com/smfforum/index.php?topic=99850.0 . I found that it was VERY fussy about which two transistors you used for the trippler. I ended up using a 2N5087, and a BC546B. The gains have to be matched as close as possible. As soon as I got out more that 6 or 7 points on the gain differential, the decay starts to gate. Match the gains, and the front end is a REALLY nice fuzz.

I know it doesn't exactly answer on the oscillation, but it might help you out with the trippler. Oh yeah, you did read the part where Tim wrote, "And at max gain, it's completely nasty"? Just checking...

Insofar as the oscillation is concerned, from my reading, there could be many causes. Amoung them; feedback (some of the output signal getting into he input), power supply noise feeding into the output, transistor being used has too high of a gain for the application, just to name a few. It's really a jungle out there, so when I've encountered the problem, I try to be somewhat linear in tracing the problem, always starting at the power supply, and working my way in.

The other thing I've found is that most of these circuits nave been built to death, so if you're having a particular problem, chances are, you're not the first. Some disgruntled builder has probably written something, somewhere, about his/her misadventures in Fuzzland, so Google becomes an indispensable tool. 

[rocinante]

Hey,

Thanks for the quick reply.  That spaceman build was awesome.  Ill have to go back and read all your results towards the end of the build.

Oh yeah, you did read the part where Tim wrote, "And at max gain, it's completely nasty"? Just checking...

Haha yeh i did, and when i play it sounds just that.  Its just the noise when there is no signal that worries me.  My figuring is that you wouldnt buy a pedal in a shop that buzzes incredibly loudly in normal operation even if you do turn the pots right up (except maybe the fuzz factory ), so i figure it probably isnt meant to do what it is doing.

It's really a jungle out there, so when I've encountered the problem, I try to be somewhat linear in tracing the problem, always starting at the power supply, and working my way in.

Do you just trace along the circuit with a probe?? How do you go about checking that each is working as described (especially when you havent got voltages to go off??).

Thanks again!  Ill get onto transistor matching a bit later tonight.

[drolo]

Another non EE...
Try to put a smallish cap (like the 220p on the other 2, testing to find the value that wont choke too much treble) across the C and B of the MPSA18. I'm not sure it works in your favor here but it does miracles in fuzz face type circuits. Worth a try...

[Kesh]

My limited understanding:

Oscillation happens through positive feedback. That is, the signal is somehow fed back to a point prior to it being amplified. When this happens you get a vicious circle, leading to whining, ringing, howling etc. It's called *Positive* feedback as the signal being fed back reinforces itself. It's exactly the same principle as leaning your guitar against your amp and letting it sing.

By the way, if you send the negative version of the signal back you get negative feedback, which calms and controls a circuit. This is a common way of controlling how much gain a circuit gives. In your circuit the 5M resistor between collector and base of Q1 is negative feedback. (The collector signal is the mirror image of the base signal, so it's negative feedback). This is the reasoning behind adding a cap across a transistor to control high frequency whistling.

A common way for unwanted positive feedback to happen is for the signal to travel back somewhere you don't expect it. The power rail is one way, and putting caps across  the power rails can help damp this signal down. Another way it can feed back is wires, or traces on a pcb/vero, running so close together they are effectively capacitors, so signal can jump from one to the other. This is called stray capacitance.

[rocinante]

Hey,

Thanks David!  I'll keep that in mind when im testing these circuits.

@Kesh:  Thanks for the explanation.  I was sitting here trying to work out what that 5m Resister did, so that is awesome.  And regarding the feedback that is good to know.  not a problem anymore on this particular circuit, but definitely has been.  Thanks!

I ended up finding a BC557 in my stash somewhere which is, according to fairchild semiconducters, a much closer match to the 2n3906 and that fixed the problem right away.  

[EDIT:  It actually didnt fix it properly.  When the 10kpot is shunting the emmitter to ground - sounds great.  When i have the resistance up - i have no volume and very little gain. garrgggghhh]

But so i learn something other than matching numbers - is it the hfe that is most important when switching out these transistors.  
For example:
2n3906: Hfe= 30(min) and 300(max) IC = -100mA, VCE = -1.0V
bc327: Hfe = 100 (min) and 630 (max) VCE= -1V, IC= -100mA
BC557: Hfe = 110 (min) and 800 (max) VCE= -5V, IC=2mA
[All fairchild transistor datasheets].
In this case the test conditions were different for the bc557 & the others.  It seems to me harder to compare?  Even looking at the graphs below they were tested at 5v and 1v respectively.

Im still somewhat confused.

But thanks for your help and patience with me!!

Cheers,
Steve

[LucifersTrip]

So my Questions:
1) Is this type of noise typically related to the transistors and misbiasing

it could be...but there are other factors like shielding, clean wiring, proper grounding, using a metal enclosure, etc... many fuzzes can be noisy in the last couple %. limiting resistors (so you can't hit 100%), taming caps (what drolo wrote), re-biasing, etc...are possible solutions

2) what can you typically do to rectify the differences when you replace the recommended transistors with another.

in fuzzes, what means the most is hfe and leakage and since you're probably working with silicon now, then just the hfe. when you change the hfe, the bias changes, so you will simply have to adjust the surrounding resistors. before building a fuzz (or any circuit), there are working voltages. you simply adjust to get close to those voltages

3) What are the things to look out for on the datasheet when looking at similar transistors? Eg. what is going to make it 'not work' or not function as intended.

hfe or leakage way out of range

[senko]

Fuzz pedals can be difficult to debug.

I'd suggest building an Audio Probe:
http://www.diystompboxes.com/pedals/debug.html

Figure out where the problem is by finding the noisy components and isolating the problem.  Then experiment with other values or use EE techniques to cut down on noise. 

Sometimes, the problem lays in the power supply filtering.  If you're using a wall wart AC Adapter, this could be a source of problems. 

Just remember, this is part of building your pedal-building skills.  I've been doing it for about 4 years now and I still have problems like yours; I just know how to go about fixing them after spending countless hours troubleshooting. 

[brett]

Hi
I'm no expert, but that schematic has some weird things about it.
One thing in particular is probably going to stop it working properly.  So the next question is whether the base current is enough to turn the transistor on. That 5M resistor is keeping the base voltage and current *very* low. You'll often see a resistor more like 470k in that position. Why? because it provides feedback and stabilises the collector voltage.
I suggest putting a 1M or 2.2M resistor in there and seeing if it gets more stable.
cheers
PS Also, with the very high value Rc, the gain pot may alter the bias on Q1 enough to mess things up.

[LucifersTrip]

That 5M resistor is keeping the base voltage and current *very* low. You'll often see a resistor more like 470k in that position.

one of my all-time favorites:
http://tonebender.webs.com/orpheum-SILICON.GIF

[Kesh]

One thing in particular is probably going to stop it working properly.  So the next question is whether the base current is enough to turn the transistor on. That 5M resistor is keeping the base voltage and current *very* low.

base voltage turns transistors on, not current, and if the transistor was off there would be 9V across BE, which would mean it was on, so it can't be off.

[Ronan]

In particular a problem ive run across a few times when building fuzz effects is a huge amount of noise or oscillation or static when i ramp up the texture/gain/whatever pot up the last few percent. It generally makes a fuzz sound when playing, but then when there is no guitar input, there is some aweful stuff going on.

That can be solved with a small resistor in series with the pot, a value of maybe 5% of the pot. A lot of DIY circuits don't have finesse built in, if you turn a pot too far and it sounds bad, then turn it back down again. Its better that way than "wanting more." A circuit can be tuned to taste, the beauty of DIY.

[duck_arse]

another non ee ....

if you are working with a collector biased amp or a feedback pair, like the front-end of the tripple or a fuzzface, different hfe will produce different voltages around the circuit. if you use compound biasing, a resistor to supply and a resistor to ground at the base, you set the dc voltages, and the transistor can't do much about it.

is your noise noise or feedback? any high gain fuzz circuit will produce noise out with no signal in, and gain up.


I say compound biasing, but I mean "voltage divider biasing", now that I've found that book.

[rocinante]

it could be...but there are other factors like shielding, clean wiring, proper grounding, using a metal enclosure, etc... many fuzzes can be noisy in the last couple %. limiting resistors (so you can't hit 100%), taming caps (what drolo wrote), re-biasing, etc...are possible solutions

in fuzzes, what means the most is hfe and leakage and since you're probably working with silicon now, then just the hfe. when you change the hfe, the bias changes, so you will simply have to adjust the surrounding resistors. before building a fuzz (or any circuit), there are working voltages. you simply adjust to get close to those voltages

hfe or leakage way out of range

Thanks for all the info - i had a good look around and i couldnt find a decent alternative to the 2n3906 in my box of crap so i thought i'd order some more transistors and give it another go another day. But im going to get straight on that orpheum!!  Seems like a fun circuit to play around with (any tips ??)

I'd suggest building an Audio Probe:

I use this one.
IMG_2157[1] by frpedals, on Flickr  I did run it around the circuit and i think it's the incorrect transistor instead of the 3906.  Im assuming that matching the 3904 with the 3906 will fix the problem, but i dont have one to test out. 

Everyone else - thank you for your help!

I didnt really want to just put a limiting resister on the pot as i figured the circuit was meant to function without the extra resistor and i wanted to figure out why exactly mine wasn't, but i understand that is a totally acceptable way of accounting for that extra bit of noise.

I think im just about out of questions... for now.

But seriously - thanks everyone!
Steve