Is there a " better " 2N3904 ?

[HOTTUBES]

I have a few Overdrive build doc's that call for the trusty 2N3904 , but i have read that many opt to go with different trannys in those locations for various reasons ...

Is there a better option interms of transistors in these locations ?
Does it make a difference in the end result tone wise ?
Or , is this just more crazy talk ?

[Blitz Krieg]

jelly bean

[Mark Hammer]

Elektor used to (and may still do so) use the TUP-TUN-DUG-DUS nomenclature, where TUP is "transistor, universal, PNP", TUN = "transistor, universal, NPN", DUG = "diode universal germanium", and DUS = "diode universal silicon".

One of the implications of that is that there are many circuits where the particulars of the device don't really play any role in the performance of the circuit, and only the general features of the device really matter.

It's like that for a lot of circuits here.  Although, to be fair, some things can depend on the biasing of the transistor for their sound, and in those instances the particulars of the transistor can matter.  However, if the circuit includes any adjustable components for tailoring the biasing, then the particulars don't matter so much because you'll be adjusting anyway.

Certainly one of the circuits that tends to misdirect attention towards the putative importance of specific transistors is the Big Muff Pi, that has changed which transistors it uses multiple times over the various issues.  People presume the tone of any given issue is a direct result of the transistor number, completely neglecting that other gain-setting components from that issue are also different.  I suppose if one wishes to replicate a specific issue of that pedal without having to dicker around with finding the resistor values to recreate the sound with different devices, then you'd want to use the exact same transistors.

[GibsonGM]

Would a 5088 offer lower-noise performance?

Not that I've ever worried in that regard, myself, just putting it out there...doubt I'd personally be aware of any improvement from changing it...

[Tony Forestiere]

5089 is a lower noise high gain at the cost of a tad more current consumption.

[PRR]

> a tad more current consumption.

"Current consumption" *should* be entirely under the designer's control. True, if the designer was not fussy, in many cases the high-hFE part will increase current. But it can't increase a lot or the circuit will stop working right.


Modern 2N3904, 2N5088, 2N5089 will all give essentially the same hiss in the same circuit.

Very slightly lower hiss in very high impedances, so it *may* be worth trying a 2N5089. However it should never be a "Wow!" difference.

2N5089 *used* to be sold as "low noise" because some Silicon foundries had dirty Silicon, with extra hiss, and the 2N508x parts were sold with the promise of good clean Silicon (as well as high gain and insignificant parasitic resistances for high-impedance working). However the fantastic rise of performance of ALL Silicon parts means there isn't any "dirty Silicon" around today; if you make million-transistor devices on bad Silicon you drown in defects and rejects.

These and similar transistors should bias-up and work-the-same in ANY well-designed circuit under 10mA (10mA is ample for nearly any small-audio stage).

Our problem is that many pedals are not "well designed". I suspect some of the classics were cobbled in the prototype stage to work-with whatever box of 1,000 transistors that Mike could get cheap.

[alanp]

Some circuits can be hugely affected by what transistor you use. Madbean's Hipster was my first exposure to what a difference transistors could make.

[Mark Hammer]

But that may well be because the surrounding components were selected for one transistor, and you used a different one.  Those same surrounding components could be modified in value to complement a different transistor.

Think of it like eggs.

Eggs come in different sizes, with somewhat different yolk sizes.  The "recipe" may call for 3 eggs, but the other ingredients listed are adjusted in quantity to suit some idealized egg, and may not suit the particular three you have in your fridge.  Were you able to adjust the quantities of those other ingredients flawlessly, the finished product would turn out wonderful.    But if you don't know enough about the perfect ratio between the volumes of this, that, and the other ingredients, it comes out lousy.

Don't blame the eggs.

You can often make a circuit work with a wider array of transistors than you'd think....as long as you know how to adjust the quantities of all the other ingredients.

[nocentelli]

The biasing arrangement also determines how tolerant the circuit will be to a transistor swap: As gus frequently points out, if you just have a collector resistor and grounded emitter, with or without a B-C resistor, the particular transistor used (or different hfe values of the same transistor model) will affect the sound: The big muff, by way of contrast, has an emitter resistor and a base-ground resistor as well as collector-supply and B-C resistor - I've sat with a bmp on the breadboard and swapped transistors for every npn silicon I owned and noticed very little difference. It's also worth noting that the heavy clipping from the diodes will further mask any slight differences.

[amptramp]

Some circuits can be hugely affected by what transistor you use. Madbean's Hipster was my first exposure to what a difference transistors could make.

A production circuit should be able to make use of any transistor within the published, guaranteed specifications for that device.  The usual method is to use enough feedback to maintain the bias and collector currents for any Hfe within the spec values.  If you do not do that, you have to select transistors for operation in the circuit which is possible for a custom one-off design but not for production.  In cases where you are trying to get a non-linear response, you may have to forego what would be proper biasing for a linear circuit, so you may need select-on-test parts.

Transistors now are good enough that there is little difference in noise between devices with the same part number, but if you are using older devices, particularly germanium (since germanium cannot be passivated), anything can happen.  Silicon transistors are passivated by growing an oxide layer on the device which is essentially quartz and is non-conductive, sealing the device inside.  Germanium oxide is conductive, so there is no corresponding process that can control leakage current.