2n2712 transistor replacement?

[bensonix]

I'm building a pink noise generator that uses 2 2n2712 transistors. is there a comparable transistor to use instead? they seem kindof hard to find and the sites that have them they are quite expensive. thanks!

[bensonix]

http://english.electronica-pt.com/components-cross-reference?ref=2N2712  ? any of these ones??

[m_charles]

Pretty sure any of the usual suspects can be used: 2n5088, 2n3904, 2n4401, etc.
BUT, your pin-out will be different. If you have already made the board, you can simply put wire sleeving on the trannie legs, and bend them around to fit the PCB. Not the most pro solution, but it works fine!
You can get one the trannies above at rat-shack

[GibsonGM]

Some say that's a hard to find part, costly.   In the comments section HERE http://www.mutable-instruments.net/forum/discussion/3643/diy-noise-genuse-x0xb0x-as-a-programmable-lfo/p1
they are talking about using any garden-variety NPN like a 2N3904.

[R.G.]

The 2N2712 was at one time a good noise source. It was not designed that way. It was designed for other things and it happened to hit a sweet spot for making noise when its base-emitter was broken over with reverse bias. It's an old design - I had a hard time finding a datasheet for it, which says something. And finding a cross reference won't necessarily help. A cross reference will be listed because it has specs on its datasheet that meet or exceed the specs on the 2N2712 datasheet - and the 2712 datasheet didn't include reverse-bias noise performance as a criteria.

You're probably stuck with trying to find old stock or take pot luck on trying new ones - which is what I suggest.

Build your circuit, then get a representative selection of modern, cheap, easy to get transistors and sub them in. Then listen.

(1) Can you tell any difference between them at all?
(2) Which ones sound better?
(3) Which ones meet the technical criteria for "good pink noise"?

If you can't measure pink noise "goodness", it is unlikely you can do better than swapping on various transistors and picking one you like. The 2N2712 was judged "good" back at some time in the past. With counterfeiting rampant, even if you get a part labeled as such, you can't be sure it's what you thought you were getting.

The better way is to digitally generate pseudorandom noise, then run it through a pinking filter. I suspect that an 8-pin PIC is cheaper than a 2N2712.

[bensonix]

Wow, thanks everyone!

I've already printed my board so i may just order a couple of each for a prototype and put sockets in for the transistors. The pin out situation was something i wasnt even aware of!!

I already use the 2n3904 for white noise generators, I'm just not sure that it will create the frequencies desired from pink noise, or will it?

[m_charles]

RG's post above addresses that. The short answer is yes, socket and experiment.
It's fun to do anyways 😄

[R.G.]

Reverse biased junctions allow charge carriers to slip through when they're given too much voltage to handle.  When this is current limited - as in "fed through a high resistance", not many (relatively, that is) charge carriers get through, and their transit is noisy, much like a few raindrops are heard as individual pings, but the sound of a downpour merges into a roar. The trick is to get the current about right for producing wideband noise.

White noise is characterized by a flat power spectrum, at least across the bandwidth you are looking at. It sounds very hissy. Pink noise has a power spectrum that declines as frequency goes up, but declines at half the rate that a simple RC filter would do. Pink noise sounds more "even" in frequency content to human ears than white noise does.

To get pink noise easily, you start with white noise and put that funny multisection RC filter at the end. If you do the filter right, then when you feed white noise into the input of the filter, pink noise comes out.

And that gets us back to your question - will a 3904 produce the frequencies needed for pink noise.

All reverse-broken silicon junctions come close to the right frequency distribution; this includes most transistor base-emitters, most collector emitters, most zener diodes and most avalanche diodes (which is what most "zener" diodes above about 6-7 volts actually are). They only vary slightly around the ideal one.