One of the most common beginner SDIY questions is which parts are frequently used and worth buying in large quantities to use on multiple projects. The people who ask this question think they're going to save serious amounts of money by buying common parts in bulk, and although I have serious misgivings about the money-saving aspect, the question isn't going to stop being asked, and there are in fact other reasons why keeping a stock of parts may be a good idea. Here are some thoughts on that.
I've already written a lengthy rant on The vanity of "Having A Lot Of Parts" and although it bears repetition, I'll only briefly summarize here. Some parts, like maybe fixed resistors, really are commonly used and likely to turn up in many projects. But those parts are cheap anyway. They cost just a small fraction of the total price tag of a project. Most of the real cost of building any given project will be in parts that you cannot reasonably stockpile because there are just too many fiddly details to them. If you're building someone else's designs, and especially if you're using circuit boards someone else laid out, then you run into problems like that illustrated in the following picture.
Two Bourns multiturn trim pots; 3296W-series on the left and 3296Y-series on the right. Multiturn trim pots in general are frequently used parts that you'd need for many different projects and might want to buy in bulk and keep a stockpile. Can you even see the difference between the two photos? It may not be obvious at first glance...
The footprint, or arrangement of pins on the bottom where they plug into the board, is different between the two series. There are actually several other footprints available too. These two are the most common; but they're both common. A given board that calls for multiturn trimmer pots will probably call for one of them, but there's no consistency as to which one. If you have a circuit board designed for one footprint and you try to plug in the other, it won't fit. (It's technically possible to design a board that could take either, but that's not a common practice because it means crowding out space that could be needed for other things, creating problems routing the board.) If you stockpiled one series of multiturn trim pots and you want to build a project whose board was designed for the other, too bad; you're stuck buying new trim pots of the other type and you can't use the ones in your stockpile. And this is a costly mistake to make because those little trimmers cost over $3 each if you get the name brand in small quantities; less if you're getting the Thai knockoffs, but they're still going to outweigh the cost of things like fixed resistors where it's easier to get away with keeping a stockpile.
As long as you're building from others' designs, you're going to constantly run into this kind of issue. The really common parts are never the expensive ones. The real money that goes into building an electronic project is in parts that have these kinds of fiddly detail issues to them killing the idea of stockpiling to save money. Similar issues contribute to why "PCB and panel" is usually a bad idea.
You have to synthesize
However, that stuff is a problem when you're building from others' designs. Things change a lot if you aren't. You can both save money and make things more convenient for yourself by keeping a stock of "common" parts if you're in a position to do your own designs, or at least to modify existing designs to suit the parts you have. You can arrange to make the parts you have genuinely "common" if you take a hand in the design process. As Kompressor says, You Have To Synthesize - by which I mean, create your own designs.
Suppose everybody who used multiturn trimmers like those shown in the photo agreed to use the 3296Y-series footprint. Then you could just stock those and be sure they would work. Problem solved, right?
You'll never be able to convince everyone designing modules in the world today to standardize on a single footprint, and even if you could you can't go back in time and convince all the designers of the past to do their designs with the standard parts you want to settle on today. But within a controlled environment, like your own lab, you can easily enforce this kind of standardization. I actually do get away with buying multiturn trimmers in quantity and Saving Money by keeping a parts inventory, because I'm only building North Coast modules and I have decreed that all North Coast modules will be designed for the 3296Y-style footprint; the same trimmers go into a Leapfrog VCF and a Dual Octave Switch. I similarly use only a couple different styles of jack sockets, a couple of form factors of panel potentiometers, usually 5.08mm-spacing film capacitors, and so on. You can do the same if you're doing your own designs.
It's not even necessary to do entirely new designs yourself. You can work from a schematic you got off the Net and do your own layout to use the parts you have, whether it's for a full etched PCB, a stripboard or perfboard, or something dodgy with point-to-point wiring. As you get more advanced with the electrical side, you can even get into stuff like substituting values of parts. Replace a 100k pot with a 50k pot because that's what you have? Will that work? Usually... so you have to learn how to answer that question, but once you know, it means you can make better use of parts you have on hand.
Thus, despite my disparaging remarks about people trying to "save money" with stockpiled or salvaged parts, it is possible to do it... as long as you're willing to dig a little deeper into creating and modifying designs. You just can't get good results stockpiling parts for completely ready-made designs. Looking at a complete kit, like the ones I happen to sell, and saying "I want to buy just part of that and save money by already having the other components" is a fool's errand.
There is another reason to keep an inventory of parts, and it's especially relevant when you're getting into design and experimentation. Quite often when you're playing with a new circuit, you want to try different values for something. This came up for instance with the Transistor ADSR - it has a resistor in the input which sets the voltage level at which the envelope will trigger, and the first value I chose for that turned out not to give the results I wanted. I had to try a couple more before settling on the final choice, and I was only able to do that because I had all the values I wanted to try, in stock. It went a lot better than my development of the Fixed Sine Bank, where I needed to try several different values of Zener diodes to set the output level, I didn't have a selection of them in stock, and I ended up having to make many small orders from distributors to get different Zener values until I found the right one. When you're doing development of new designs, it's really valuable to have a selection of different values of components in stock.
I'll say once more: this is NOT a way to save money on regular builds of pre-existing designs! It's a way to save effort on development of original designs. You want to have a few of many different values on hand so that at any time when you want to try a given value, you'll be able to put your finger on that value without rushing off to buy it. You save money, maybe, on buying quantities a little greater than your immediate needs, but you'll probably lose more on buying parts that you won't actually use. You are paying for convenience - but the convenience is worth paying for when you're doing design work.
Take a look at my article on preferred values for resistors and capacitors. Those series of numbers provide a good guide to useful values of these components. If you get, say, all the E24 values in a range, then you'll know that for any value you might want in that range, you'll have a part with a nominal value within about 5% of your desired value, and that's usually close enough. In practice, you often want to choose parts from the shorter lists. I keep a stock of E24 resistor values but I usually design with E12 values when I can; that means there are that many fewer values I'm likely to end up needing to buy in larger quantities. For capacitors I aim for E6 or even E3 values. At the extreme, the power-of-ten values (1k, 10k, 100k, 1M) are so popular it may actually make sense that someone could "save money" by stocking up on them even for others' designs.
With resistors in particular, there are a lot of commercially-available kits on the market that provide a few (maybe 10, maybe 20, maybe 100, you have choices depending how much you want to spend) of all the values in some E-number series and a given range. If you're looking to set up a lab for doing experimental designs, it makes a lot of sense to get one of these because they're often priced much lower than what you'd spend buying small quantities of many values one at a time. Just be aware that you will inevitably end up getting some values you are unlikely to use, try not to waste too much money on those, and think carefully about which kit you want.
I kickstarted my lab a few years ago with a kit that contained 100 each of all the E24 values from 10 ohms up to 1M, in 1% metal film 1/4W through-hole. That has served me pretty well. However, as is common with metal-film resistor kits, it goes further down into the low end and not as far into the high end as I'd really like. In synth DIY, it's rare to use any resistor values below about 1k except maybe for the occasional LED or output current limiting resistor. I expect that many of the values between 10 ohms and 1k (two full decades) will never be used in my lifetime. Meanwhile, I do sometimes want resistor values above 1M, for things like high-frequency trim in oscillators, frequency-setting resistors in LFOs, and so on, and the kit just doesn't cover that range. (Very few kits do.) I've had to supplement it by buying additional resistor values in the beyond-1M range. So my ideal lab kit, which I don't think anybody is selling, would actually go from 100 ohms to 10M.
There are those who would claim that it doesn't make sense to buy 1% resistors in E24 values; the E24 series is really intended for 5% resistors and 1% resistors in E24 values end up giving uneven coverage with gaps between the values. My thinking, however, is that I'm really treating these as higher-quality 5% resistors. For decades the standard was 5% carbon-film resistors. In most circuits, when you calculate a value you want, that's about how close you have to hit it to get the circuit to work. Having tighter tolerance, so that the circuit is more predictable, is really nice, and in this day and age, at hobbyist quantity levels, 1% resistors don't really cost much more anyway and it's worth having them. But it's not really necessary to keep a stock of very many more values (going to E96 for best coverage of 1% resistor ranges would quadruple the size of the kit) just for the rare cases when I need to hit a calculated value more precisely during development. It's also a fact that I don't want to be designing with E96 values any more than necessary, at all - they're much harder to source as single values when it comes time to buy larger quantities for production. And if I do ever try to misuse my inventory as a cost-saver for building others' designs, well, others also design with E24 (usually E12) values and my chances are better with those.
I like to have decent stocks of smallish electrolytic capacitors for power supply filtering: 10µF and maybe 22µF, rated for the full Eurorack power voltage (in practice this usually means 35V, as the next level comfortably past 24V for +-12V power supplies; it would cover the +-15V of other formats too, though just barely). I also use many 0.1µF axial ceramic capacitors for bypassing ICs; usually two of those (one on each power rail) for every DIP IC, which is slightly overkill but better than having too few. I wouldn't invest in a range of power-filtering capacitors because the cases where you really need other sizes than these are rare enough it makes sense to wait and buy caps specifically for those cases when they come up. I avoid putting audio through electrolytic capacitors, but if you are willing to do that you may want some electrolytics in this range or a little larger (maybe 33µF and 100µF) for AC-coupling signals.
For op amp compensation: I don't keep all of these in stock but it might make sense to keep the E6 values from 10pF to 100pF (in addition to those two, the series would be 15pF, 22pF, 33pF, 47pF, and 68pF) in disc ceramic or MLCC; disc ceramic is cheaper if you can find it, but they're falling out of use. Among those you can find most of the compensation capacitors you're likely to want for synthesizers. Capacitor kits, like the resistor kits, do exist but they're likely to cover a lot of values you're unlikely to use for audio.
For capacitors in a synthesizer that are really used for audio (filtering, VCO integrators, and so on) it makes a lot of sense to use polyester film. I've standardized on 5.08mm (0.2 inch) lead spacing in my own designs; it's convenient for breadboarding, doesn't take up a lot of space on the PCBs, and most capacitor values are available in that spacing. As with the ceramic caps, there are capacitor kits available. I use a lot of the E3 series from 0.1µF to 1.0µF (that is 0.1µF, 0.22µF, 0.47µF, 1.0µF) but that's partly because I'm building Fixed Sine Banks which use these relatively large film capacitors. Other audio applications often use smaller values, like from a few hundred pF on up. This is a wide enough range, and the applications are specialized enough and the actual capacitors expensive enough, that I would hesitate to buy a range of values in advance, and would just buy them piecemeal when I had a specific application in mind. Unless you can get a really good deal on a "kit" there's just too much likelihood of waste.
I don't think it makes much sense to stockpile potentiometers unless you're really confident of being able to redesign things to fit your inventory. There's just too much that can go wrong with the physical design, and pots are expensive, so it's really a shame to buy a bunch of a specific type of pot and then find it won't fit in the place you want to put it and you have to spend a bunch more money on other pots that are very similar to the ones you already bought but can't use.
Nonetheless, if you want to live dangerously and you accept that You Have To Synthesize: 100k linear and audio taper are the most common panel pots. Pick a manufacturer and series according to your quality and price preferences and design your circuits around that. Very many circuits use or can be adapted to use 100k panel pots. Other power-of-ten values (10k and 1M in particular) are also nice to have for occasional situations where 100k won't do.
For trimmer pots you may need more distinct values. The main thing, as I mentioned earlier, is to pick a consistent PCB footprint. I really like the Bourns 3296Y-series footprint: it doesn't take up too much board space; it fits on breadboards and prototyping PCBs with 0.1 inch grid holes; I think having the pins in a triangle instead of a line makes for a slightly stronger, more rigid assembly; and there are many options both cheap and expensive for trim pots fitting this footprint from different manufacturers.
The 1N4148 diode is ubiquitous. It probably makes sense to grab a thousand of those. Note that projects calling for 1N914 diodes can use 1N4148; the only difference in specification is that the 1N4148 has a lower maximum amount of noise allowed, so any diode that meets the 1N4148 spec also meets the 1N914 spec. I also use a lot of 1N5818 Schottky diodes. These are appropriate for reverse-voltage protection on Eurorack power supply connections and can often substitute for the other kinds of protective circuitry that other designers like to use.
You need basic NPN and PNP silicon transistors. The most popular ones are 2N3904 and 2N3906 and these are reasonable choices for general-purpose use. In my own designs I've more or less standardized on 2N5088 and PN200A, which are higher-gain types. A few places the higher gain is useful; I just use these ones across the board because it's convenient to have only a couple of transistors that I use almost everywhere. One of the secrets of modern discrete BJT design is that the specific transistors you use hardly matter anymore, especially at audio frequencies; and in the occasional cases where they do matter, you're going to be buying transistors specific to your project instead of taking them from generic stock anyway.
Despite mentioning above that I ran into a situation where I wished I had a range of Zener diodes to try, I would not actually recommend trying to stock such a selection. They're uncommon enough that most will go unused and you're probably better off just buying specific values as needed. Similarly with other kinds of discrete semiconductors, like JFETs.
The TL074 quad op amp is ubiquitous in synthesizers and it makes sense to keep those on hand. Maybe also TL072 (dual instead of quad version of the same thing). It probably doesn't make much sense to stockpile any other op amp types - in the cases where you need something else, your needs will be specific enough that you'll probably need to buy specifically for your project. At one point I bought a bunch of bipolar-input MC33079 quad op amps thinking I would use them for higher-quality audio applications and then found that because of the input impedance requirements of synth modules, I had very few chances to actually do that. Similarly, I keep a stock of 555 timers for general hobby electronics, but rarely use them in synths in particular.
The standard VCA chips are the LM13700 (as about the only operational transconductance amplifier still on the market in through-hole form) and various manufacturers' successors to and clones of the discontinued SSM2164. It may make sense to stock these for synthesizer experimentation.
I have several times seen a recommendation to keep on hand voltage regulator chips, in particular the 78L05, but I'm not sure how useful that really is. I guess it depends on the kinds of designs you do; if regulating the +12V rail down to +5V to supply digital circuits, or sensitive analog circuits that need an isolated power supply, is something you do a whole lot, then you may use a lot of 78L05s, but if you're in that position you would know. I used the 78L09 for similar purposes in the Leapfrog VCF but haven't found another reason to keep a stock of that chip recently. I do use a lot of TL431 regulator/references in different designs of my own, but I'm not sure anyone else does; it's a matter of taste.
For the most part, I think you should expect to be choosing and buying new chips for each design you do.
As a related issue, assuming you're doing through-hole, it makes sense to stock IC sockets, and that introduces the next item.
Jack sockets to put on the panel for patching. I use Lumberg 1502 03 vertical sockets (similar to the popular Thonkiconn) for boards parallel to the panel, and CUI MJ-3536 right-angle sockets for boards perpendicular to the panel. These, or anyway some kind of jack sockets of your choice, are probably worth stocking.
Header connectors for plugging boards into each other: you can get breakaway single-row male headers that snap to the desired length, and in principle there are female sockets that are also break-away style, but I prefer to just stock 10- and 12-pin lengths and use as many of those as necessary. Double-row headers are also worth having for Eurorack power connections.
Toggle switches are expensive and it may or may not make sense to stock them, but if you do, it certainly makes sense to standardize on a particular line. I like the E-Switch 100-series of miniature toggle switches. Many cheap toggle switches have the same dimensions as these; I've even had good luck ordering no-name versions from Tayda and finding that their unspecified dimensions were compatible with the E-Switch version. I keep stocks of the SPDT three-position (centre is no connection) and DPDT two-position types, because these are the ones used in my current products, but I expect that in the future I'll end up stocking one or two others for use in other designs.
I worry a lot about newbies who think they are going to "save money" by keeping a stock of parts to build third-party designs; that doesn't really work. But it does make some sense to keep an inventory if you will be experimenting with original designs, or if you are willing to cultivate the skill of modifying existing designs to use the parts you have on hand. It works better, in particular, when you are building from a schematic instead of from a full physical and PCB design. I've gone through some of the kinds of parts and value ranges where keeping a stock of commonly-used parts is most likely to be successful. I had originally planned for this entry to also include some notes on storage and organization, but it has gotten quite long, so I think I'll leave that for some future entry.