Noise of various transistors for fuzz?

[jm22]

It appears that there are scores if not hundreds of different NOS transistors that could perhaps be used in fuzz pedals. My question is, if a given assortment of transistors all have about the same gain factor, will some be more quiet and less prone to noise, hum, etc.? If so, is there any way to know which ones will be less noisy than others?

[antonis]

It's called "Noise analyzer"..

But we should shake hands on Fuzzes aren't exactly what we call Hi-end effect..

[R.G.]

I've written several polemics on what type numbers mean (not a huge amount, btw) as regards picking transistors for gain and noise. It can be distressing if you're just learning about electronics.

Type number (2N...., 2SC....) can serve as some kind of hint. Transistors that are intended specifically for low noise applications will often have some kind of noise specification. That's a good place to start. But remember, datasheets talk about whole populations, but each transistor is an individual. There is variation in noise among a specific part number. Worse, a transistors's history is significant. Reverse breaking the base emitter junction >>even once<< can permanently make a transistor noisier.

Sadly, trying them out one at a time is the only good way.

And no transistor is more or less prone to hum. That's out of the transistor's hands.

[jm22]

Thanks for the replies. When you say reverse breaking the base emitter junction, can you more precisely describe what that is? I am aware of two basic styles of transistor, can style, like AC128, and the plastic ones like BC183. Are you saying that over-stressing the legs causes some kind of mechanical failure where the leads enter the transistor? And are both metal can and plastic transistors prone to this?

[bluebunny]

When you say reverse breaking the base emitter junction, can you more precisely describe what that is?

R.G. means wiring up the transistor wrong.

[mozz]

So one of these transistor analyzers, such as a Ebay MK328 or DCA55 could be possibly damaging a transistor while it does the pinout test?

[R.G.]

Mechanical stress is prevented from reaching the chip inside by the overall package. I meant electrically breaking over the junction with a reverse voltage.

All diode junctions will conduct one way, and block current the other way. But they can only block reverse current up to a certain reverse voltage, and they then break down and let current through. When we run diodes in this breakdown mode deliberately, we call them zeners.

The base-emitter junction of a bipolar transistor is no different. It reverse breaks when a reverse voltage big enough comes along. The first transistors were symmetrical devices - the base-emitter and base-collector junctions were nearly identical, including in terms of reverse voltage. This produced low gains and other badnesses, so we learned to make them asymmetrical, by making the base-emitter junction thinner and its reverse voltage lower, getting higher gain, frequency response, lower noise and so on. But this does make it easier to break over.

Most old germaniums are mildly asymmetrical. All modern silicons, as far as I know, are highly asymmetrically doped, so that the base emitter junction breaks at 6V-8V. We mostly use them in circuits that always have the junction forward biased, so that's OK.

But if you wire them up wrong, or do abusive tests with a 9V battery, you can reverse break them. I suspect, but don't know, that component identifiers are designed not to reverse break junctions. But then I didn't design the tester, so some caution is needed.

Breakover damage is cumulative, and depends on how much reverse current you shove into the junction. Little blips won't do much damage, big reverse currents or many little reverse currents accumulate lots of damage. The "damage" is that the junction just produces more hiss than when the device was new.

There are a couple of things that can make bipolars get hissier over time, especially input transistors: input transients and capacitors. Input transistors take whatever the input cable gives them, so unless you protect the input devices, they get little zaps of microsecond breakdowns and over time they get noisier. We're all familiar with old pedals and amplifiers being noisy and fixing this by replacing the input transistor. Input capacitors and emitter capacitors can do much the same at power-off if they hold enough voltage to act like a reverse voltage when power goes down. This isn't a big issue in 9V powered stuff, but when the caps can hold over 8V or so, it can make a little reverse zap every time you turn power off.

The solution is easy - put a reverse biased diode across the base emitter. A 1N914 or 1N4148 hooked up this way will completely prevent reverse breaking the base emitter, and adds only very few picofarads of capacitance to the junction, not hurting the audio performance.

[jm22]

Very interesting reply R.G. Thank you. I might have a follow up question but I'm going to spend some time digesting your post first.

[PRR]

> transistor analyzers, such as a Ebay MK328 or DCA55 could be possibly damaging a transistor while it does the pinout test?

No. (R.G. is not the only one who knows this.)

There's breakdown, but the real damage comes from the Current which flows (and the Power that makes). The "good" testers use microscopic (uA) current which is enough to sort the legs without measurable harm.

[Mark Hammer]

I hope this response is not too sarcastic.  I write it after just watching John Oliver's show, and it's a little contagious.  But with that preamble....

Transistors are made and sold for pennies/yuan.  Even when we pay several dollars for them, because they are NOS germanium, they were originally made for pennies.  And for pennies, quality control is a crapshoot.  If one has dabbled in the synth universe and tried to make something to generate noise, the advice is always to use a socket and try out different transistors of the same type/number in search of the noisiest one.

That's not to say that part-number/family (2N, BC, MPS, etc.) differences are meaningless.  Rather, there is still overlap between them such that a noisier OR quieter unit is always somewhere out there even within the exact same part number, and few recommendations from any of us will absolutely guarantee that a given transistor, identified only by number, will be noticeably less noisy than anything else suited to the purpose.  The manufacturer can certainly aim for and indicate that this or that unit leans in that direction, but again, it is no ironclad guarantee that it is decidedly always less noisy than transistor X or Y.

The other thing to consider is that the noise of the circuit is not exclusively the noise of the transistor.  Other components produce noise, and the transistor has the job of amplifying everything at its input, which could include noise.

None of this is to suggest that being picky about transistor choice is pointless.  But certainly control of bandwidth in higher-gain circuits is something one can intentionally tackle, where noise of specific transistors - unless you have a lot of them and a similar amount of patience for testing them - has more luck involved.

[Rob Strand]

My question is, if a given assortment of transistors all have about the same gain factor, will some be more quiet and less prone to noise, hum, etc.? If so, is there any way to know which ones will be less noisy than others?

Hum and buzz doesn't come from the transistor.   It comes from the environment, layout, wiring and shielding.

Sometimes buzz might appear to be louder with some transistors but that's usually because the circuit has more gain ie. things aren't really the same.   If the circuit has more gain then the signal should also be louder so the signal to noise is about the same.

The noise from transistors is usually the hiss.   

For high impedance circuits, say the first transistor in a pedal,  a higher gain transistor will often have lower noise.   While in theory all transistors with the same gain should be pretty much the same, you do see some differences between transistors in high impedance circuits.   The differences are often small in practice because other parts like resistors also contribute to noise.   For guitar pedals you often go for the sound of the pedal and that may override changing things to reduce the noise.    (Low impedance circuits can also show different noise between transistors but for a different reason.   This aspect is almost irrelevant to pedals.)