Q: A lot of semiconductors like (insert your part number here) are incredibly hard to find. A lot of these have also been discontinued. Many times a part doesn't always have an NTE/ECG equivalent replacement. What should I do - or is there anything I can do?

A: Keen's Laws are the rules I use when I have that and other similar problems.

These laws come out of the observation that in many cases almost anything will do if it's somewhat similar to the specified part. This is true for at least two reasons:
1) many semiconductors inherently have a lot of variation, so EE's have developed techniques for making the variation not matter much. That not only opens the door for variation in the specified part, but also for many other parts that have even vaguely similar operation.
2) today's semiconductors are very, very good compared to the needs of audio circuits like we use in effects. Many modern devices, even ones with old part numbers are made with modern processes, and as a result have frequency responses far better than the same device from 20 years ago.

But on to the Laws.
1. Use what you can get, subject to some rough and ready rules. See Resistors, Capacitors, Inductors, Transistors and Opamps below. If you have a circuit that needs a specific NPN transistor, and all you can find is a 2N3904, try it. Chances are that if it's not a high power application, it will work mostly. If you need a 3.3K resistor and don't have any of those, notice that 3.3K is about 1/3 of 10K and make one up by paralleling three 10K's that you do have, or a 2.2K and a 1.1K in series. There are places where this kind of approach will not work, but it does work surprisingly often.

2. Get it running now, and put the perfect parts in later. In most cases, there will only be one magic, special part you can't find, and then it's really a matter of not being able to find the part *right now*. Build assuming that you will replace the replacement part with the exact thing when you can. If you still need to.

3. Resistors: Ohm's law rules! Resistors are the most standardized part. You only have to ensure that you get the resistance right, the power dissipation right, and in high voltage situations, the voltage rating right. Whole books have been written on Ohm's law, and I'm working on an article to explain it in some detail from the effects view. To do anything except paint-by-the-numbers electronics, you need to know Ohm's law backwards, forwards, and upside down.

However, for this short version of the Law, you need to know how to series resistors (add the resistances) and parallel them (Rp = 1/((1/R1)+(1/R2)+...)) This reduces to Rp = R1*R2/(R1+R2) for only two resistors. You can usually make up any odd value you need this way.

4. Capacitors: Get enough! For capacitors, get at least the right amount of capacitance, and at least the right amount of voltage. Getting enough capacitance can take the form of finding any capacitor with the right value and a voltage equal to your power supply voltage (9V for most effects) or greater. If you don't have capacitors big enough, you can parallel caps to get more. They add when paralleled, like resistors in series. As an aside, the calculation for capacitors in series is just like the calculation for resistors in parallel to get smaller values. But most times paralleling will do the trick.

Remember - get enough capacitance, and enough voltage. The effect of getting too much capacitance will only hurt you in tone control or filter sections, where too big a capacitor will move the critical frequencies down. But the device will still work.

5. Inductors: Don't! Take a tip from the pros. EE's usually avoid inductors like the plague. There's only three uses for inductors in the music world that make up well over 99% of the usage. Those are (a) in the Vox wah circuit (b) in graphic EQ's (where they are usually replaced by gyrators) and (c) filters in tube amp power supplies. Count on some misery getting the right inductor. Avoid them unless that's the only thing that will do.

6. Transistors: When in doubt, use a 2N5088. Or a 2N3904 - or a 2N4401 - or a BC549 - or... find cheap, available NPN and PNP devices that you can get easily and use them. In the USA, 2N3904 (NPN) and 2N3906 (PNP) are easily available and can be found almost everywhere. I personally prefer 2N5088 (NPN) and 2N4250 (PNP) because they have slightly better gain and lower noise. 2N4401 (NPN) and 2N4402(PNP) work almost as well. Find good parts you can get, and then figure out why they **won't** work, not how to find magic parts from Timbuktu.

For small signal MOSFETs, the problem is easier. Get 2N7000 or BS170 for N-channel and BS250P for P-channel.

JFETs present a bigger problem. N-channel JFETs vary all over the map, and by an almost intractably large amount. The big problem is that the variation of Vgs and gm values is huge for these parts. This is one place that you may have real problems. However, almost all JFET circuits are set up with trimmer resistors to adjust the operating conditions for the JFET because of the large variation in the JFET. I personally stock 2N5485 and 2N5457 JFETs. BF244C is almost exactly the same as the 2N5485.

On all transistors watch the pinout like crazy . Assume that the device you actually have in your hand is probably different from the device specified in the effect and go look up the pinout for what you actually have. Never assume that a replacement has the same pinout - check it! This one step would probably eliminate half the non-functional first-timer's pedals. The data sheets are available on the internet in almost all cases. Go do the search. Find them.

7. Opamps: Get the pinout right.
Opamps are parts that are deliberately nonspecific. At the low voltages found in effects, usually any opamp will work, at least to some degree, if you get the pinout to match. Match the functions on pins - that is, wire the pin on the actual device in your hand to the hole in the board where that function goes. If you are replacing the 4739 (Craig, why did you do that to us???) find the hole where the noninverting input of the 4739 was to go, and then find the noninverting input pin on the device in your hand. Run a wire between them. Now do the inverting input. Now the output; now the power pins. By the time you get done, you should have a working device. Pinout can be viewed as kind of similar to those word jumble puzzles - look at it all kinds of ways and unscramble the pins so functions match.

8. Beware Germanium! Germanium transistors and diodes are exceptions almost by definition. They have even more variation than silicon parts. In germanium there is no guarantee that even the real, official, specified part number will work like you want it to. Assume you're going to have to specially select, match or interview germanium parts.

Will these Laws work all the time? Absolutely not. There are places where only the right part will do, and that part works magic. But those occasions are rare.

Take your own fate into your hands. A real part that's almost what you need and is sitting in your hand works about a hundred thousand percent better that the perfect part that you can't get.

Q: So theoretically I could use any PNP or NPN to substitute for any other PNP or NPN respectively??

Yes - with caveats!

=> it has to be small signal circuitry, not power applications; most FX work is like this

=> it has to be low frequency, like audio; most FX work is like this

=> it has to have enough Vceo capability; almost every bipolar made will work with 9V batteries, though

=> it has to have enough current capability; almost every bipolar made will work with under 50ma or so, and the vast majority of all effects applications will be less than that

=> you have to get the pinouts scrambled around right!! No transistor will work with its base, emitter or collector leads in the wrong hole!

=> it may not work perfectly in some cases and in a very few cases, it won't work at all.

It's that last statement that you have to be philosophical about. As I noted, having any NPN in a corresponding NPN set of holes has a hugely better chance of doing something you like than not having a transistor there at all.

There are places where this will clearly not work - power applications where the transistor will go up in smoke; unforgiving applications that rely on some specific feature of a nonstandard device; high frequency applications; applications which absolutely require super high gain, low noise, or something more esoteric, like very low Rceon. These apps all exist, but mostly they're not how we use transistors in effects.

Subbing any old transistor into a circuit has to be viewed as experimental - it may work, it may not. What I'm observing is that for a large majority of times, it will work well enough to at least let you have some neat-o sounds, and let you get along until you can get the perfect part.

Many times, it will work so well you never have to find the original/perfect part. For instance - transistors in the Tube Screamer, Big Muff, Foxx Tone Machine, Superfuzz, and to some extent the D-comp/Ross compressors and almost any wah circuit. Doggone near any modern NPN with a gain over 100 will pop in there and work fine forever. Will there be slight variations in tone? Sure. Will it be dramatically different? Absolutely not.

This still leaves you with a quandry, right? OK, so it works most of the time. When will it *NOT* work? More to the point, can I pop any old NPN into my Belchfire Mechablaster clone and have it sound like laser beams on ice?

Sorry - I can't tell you. What I can do is tell you that you'll do a lot better trying a 2N3904 (oops, 2N5088!) than moaning about not being able to get CZ45y7's from Inner Elbonia. There is a fair chance it will sound *better* than with the original CZ45y7's, too.

As a last thing to think about, I can tell you that almost never does the maker of a commercial effect carefully interview parts in a tone-tasting, analog of wine tasting. What they do is buy the *cheapest* part that will work most of the time - and hire engineers that will give them that choice.