Module wear and aging
2022-10-13 electronics environment
There's a lot of confusion and worry over the idea that modules wear out or age over time, and what can be done to prevent or deal with that process. It's a part of human life that many things including our own bodies do change with use and time; and if you've spent a lot of money on musical instruments, then it's reasonable you might be worried about wearing them out or hurting them. Electronic equipment and the physics behind it are mysterious to many people. We know that invisible things like "voltage" can possibly damage electronics; it's not obvious what those invisible things are, or where they come from or how to protect the equipment; we have some idea that things in general can wear out, but may be unsure of exactly how that applies to electronics; and so modular users, and especially beginners, end up with a lot of confusion, worry, and misconceptions about wear and aging of equipment.
I had the idea to write an article about these issues, but as I was drafting it I found that some topics were spiralling out of control with many points I wanted to make, while others were hard to say anything intelligent about in the first place. It was hard to organize it into a single well-balanced article. I now think it'll work better as a series of loosely-connected articles; this is the first of them, and should serve as an introduction as well as covering some smaller points that didn't fit well into other parts.
The overarching point I'd like to make is that there are two ways modules fail: there is wearing out, from normal use over a long period of time, and there is sudden failure, from situations that are not normal use. Discussions of either of these often seem to get mixed up with the other, but they are two quite different issues.
I'm indebted to a modular user I once saw start a thread on a forum about how he'd destroyed a module by plugging it in backwards, to the point that it could no longer be meaningfully repaired and would have to be replaced completely, and how he thought this was part of a "throw-away culture" and "planned obsolescence." Although I strongly disagreed with his point of view on such things (and told him so!), that thread highlighted the important and interesting issue that wear over time in normal use is different from sudden failure because of a one-time catastrophe. When to make the call to replace instead of repair is also interesting to think about, but not part of my plan for things to discuss in the present series.
It's a basic rule that when electronic equipment wears out gradually over time, usually it's because of straightforward, simple, physical and chemical effects that are not directly related to electricity - like friction, corrosion, chronic overheating, liquid evaporation, and even radioactive decay. These are the same ways any physical object could wear out.
Conversely, when electronic equipment suffers damage from complicated and subtle effects, and from directly electrical effects, then it tends to fail in a sudden catastrophic way, immediately when the cause of the failure is applied; not gradually over time. These kinds of sudden catastrophic effects might include a severe overvoltage; a static-electricty "zap"; short-circuiting a high-power output; reversed power; a "missing rail" condition on certain ICs that have a specific vulnerability to that; or even the shutdown of an Internet-based service or "app." These are specific to electronic equipment and may be more mysterious to the average person.
There are exceptions to those general rules in both directions, but exceptions, as the name implies, are exceptional. You shouldn't expect to see modules in a synthesizer routinely failing gradually over time because of subtle electrical effects, because that just isn't how equipment tends to fail; just as when you hear hoofbeats in most parts of the world, you should expect that you're more likely hearing horses, not zebras.
Friction and pots
Things that move are subject to friction, and moving parts can wear out through friction in a synthesizer module just as they would in any other context. Every knob on the panel of a module has some mechanical device behind that panel registering its motion - often a potentiometer, sometimes an encoder or a switch. All these devices are subject to wear. There are shafts turning in beartings, and wipers sliding along tracks. On the microscoping level, every pair of surfaces that move in contact with each other are constantly remodelling each other, breaking off atoms here and pushing them into new places there. The general trend is usually that tight things get looser.
Potentiometers are rated for a certain number of "operations" (usually defined as turning the knob all the way from one end of the track to the other) and it may be surprisingly low. I've seen lifetime ratings as low as 5,000 operations for panel pots used in commercial synthesizer modules, although that is unusually low; more like 30,000 would be typical. Carbon composition pots tend to be shorter-lived than other kinds, and that's a problem because they are the cheapest kind of pot and are therefore the most popular.
If you operate the pot 100 times per day, which is a lot but plausible if you play your synthesizer in a professional or fanatical-hobbyist scenario, then you will hit 30,000 total operations in 300 days; that's less than a year.
As I've mentioned, things wearing out like this do not tend to suddenly fail catastrophically. There will not be a moment when you make the exactly 30,000th operation of the pot and suddenly a shower of sparks comes out of the panel and the whole thing blows up, with everything fine up to that point. Instead, you may notice that the knob gets a little looser, twisting with less mechanical resistance, over time. The first stages of that process may actually be an improvement: pots often feel too tight when they're brand new. Eventually, maybe it's looser than you want, and then maybe the operation seems a little flaky - like sometimes there are intermittant jumps in the apparent setting even though you turned it smoothly. Maybe over time there starts to be a little more noise audible in the signal controlled by that pot whenever you turn the knob. Eventually, this flakiness is significant enough to really be a problem and the pot can be called worn out.
Just saying in the data sheet that the pot is expected to last 30,000 operations doesn't mean that all pots of that model will wear out at that point. Usually, it's a minimum: the manufacturer is promising that very few will wear out any sooner. Probably, most will last much longer; but they don't promise that. All pots will eventually wear out if you use twist them enough times. In applications that really require extremely long lifespans, designers may opt for some other technology that involves less sliding friction than a pot, such as an optical encoder.
What is to be done about this? Ultimately, nothing can be done if you use the equipment forever; you some day will have to replace things that wear out. You can postpone it by using a longer-lived component in the first place. I normally use conductive-plastic instead of carbon-composition pots in North Coast Synthesis products. Conductive plastic is arguably just a different kind of carbon composition, but it embeds the carbon particles in a harder, smoother substrate compared to traditional carbon, and it allows for more sliding friction with less mechanical wear. The P260T potentiometers in the Leapfrog VCF are rated for one million operations and should take many years to wear out even under very heavy use. Some other North Coast modules, such as the Middle Path VCO, use P0915N pots, which are rated for 100,000 cycles - not as much, but still significantly more than the popular carbon-composition type used in many Eurorack modules. Cost is one reason for choosing one type of pot over another, but physical size is also important; in a module with many pots, using the most long-lived type might put both the price tag and the overall physical size outside what the market will bear.
Looking hard at the expected lifespan of panel pots raises the interesting general question of how much lifespan users really need. I mentioned above an estimate based on someone playing their synth heavily every single day, but most customers do not really do that; and the lifetime of a pot scales directly with how many times you literally twist the knob. Just like having eyes smaller than our fingers, many of us have unrealistic ideas of how heavily we will really use our equipment, and for how many years. It may not really make sense to pay a lot of money up front for potential lifespan that, honestly, we will never really need. As with many things in engineering, there are important trade-offs here and maximizing the specific variable of lifespan may not really be the right answer to the overall question.
Sometimes synthesizer users use the term "trimmer" to describe a potentiometer with a plastic shaft sticking through the panel, without a bushing or separate knob attached. These pots are popular because they take up less space than other types, and everybody wants a smaller module panel. They may possibly be cheaper, though that is not always the case and I think physical size is usually the more important reason for choosing them. And they are also a cause of complaints because people think the lack of a bushing makes them "wobbly." I've heard reasonable arguments both for and against the wobbliness claim and am not fully convinced either way; but it's true that I avoid this kind of pot in my own designs, and I mention that all my pots have bushings in my store pages because I think customers care about it. The word choice of calling bushing-less pots "trimmers" annoys me because that word means something else.
Properly speaking, a "trimmer" is a potentiometer (or in some contexts, especially in radio work, a different component like a variable capacitor or inductor) designed for trimming a circuit - that is, for more or less permanently adjusting the circuit to handle variations in component tolerance, V/octave calibration, and so on. You adjust a real trimmer only rarely, usually only once as part of the manufacturing process.
The distinction is important to the question of wear, because a real trimmer, intended for trimming or semi-permanently adjusting the circuit and then meant to be left alone, may be rated for far fewer operations than a panel pot intended to be turned repeatedly during normal use. Some real trimmers are rated for a lifetime of only 100, or even 10, operations. When they wear out, it will happen in the same way other potentiometers wear out, just much sooner. If somebody put one of those on a front panel and encouraged users to tweak it repeatedly during normal use, the modules would wear out almost immediately.
Fortunately, professional designers don't really do that. When commercial modules are built with bushing-less pots on the front panel, those are almost never trimmers at all; they are simply bushing-less panel pots, with lifespans comparable to other panel pots. They'll be rated for at least thousands of operations, usually tens of thousands of operations or more, the same as other panel pots. I just worry that calling these panel pots "trimmers" may lead some DIYers to buy other items that supplier catalogues call "trimmers" because they really are, and then put those on a front panel and be disappointed.
Other components besides pots can wear out by similar mechanical means. Switches of different kinds are subject to sliding friction or metal fatigue. Jack sockets include spring-like contacts that flex every time a plug is inserted, and the plug itself slides over the contacts as it goes in and out. These kinds of components all have their own ratings for how many times the manufacturers promise they can be used before wear will become an issue, and the numbers tend to be in the same range as the lifespans of panel controls: thousands to tens of thousands of operations. That's usually enough, and in practice, when people notice modules wearing out from use as opposed to other effects, it tends to be the potentiometers that go first.
I have a personal aversion to slide switches, the kind with the little plastic tab, because I've had too many pieces of equipment where those in particular became flaky before the rest of the controls, no matter what the data sheets might say about their expected lifespan. I would rather have a (much more expensive) toggle switch with a long lifespan; but in principle, it's possible to get slide switches with reasonably long lives, and many high-quality modules do have slide switches on them. Other than those, I usually only expect to see issues with things like sockets wearing out when they are being used outside their normal application area. For instance, the common "header" connectors often used for Eurorack power, and used in North Coast Synthesis modules and many others for board-to-board connections, tend to be rated for hundreds of plugging and unplugging cycles. Not thousands. In their intended use, that's plenty; nobody is completely disassembling a modular system and separating boards within a module multiple times per day. But it might become an issue if, for instance, someone were to invent a custom synth format with header connectors on the front panels as a cheaper substitute for Eurorack jacks. And I have seen DIY projects built that way.
Play hard, wear harder?
Great rock'n'roll should hurt, and it should change your life, not necessarily for the better.
- David St. Hubbins
There's a fascinating misconception that I've only really encountered a couple of times, but I have encountered it more than once, and I'd like to talk about it because it's aesthetically appealing even if it makes no physical sense. This is the idea that your modules will wear out sooner depending on the style of music you play. If you play soft, gentle, melodic music with your modules, they'll last longer than if you play harsh dissonant experimental noise. I mean, I almost want that to be true! It would make so much artistic sense.
But for the same reasons that having a module powered up doesn't cause it to wear out faster, running different kinds of signals through a circuit while it remains within its normal operating specifications also won't cause it to wear out faster or slower, and even if you break the rules, the consequences are more likely to show up as catastrophic failure than as accelerated wear.
There is nothing moving with the signal except the electrons. A component like a transistor responds only to voltage and current, microsecond by microsecond. If that transistor happens to be in an amplifier through which you're playing a harsh, severely distorted heavy metal lead, then the level of voltage and current will be jumping around from one microsecond to the next in an irregular pattern - but as long as the values are within the maximum ratings at each point along the way, the transistor is under no more stress than if it were sitting at its average operating point, corresponding to silence, all the time. In a typical "class A" circuit, the average amount of power dissipated by the transistor, which determines how hot it'll get, is determined by the biasing and does not change with the signal level; and even in more complicated setups where signal levels do affect heat dissipation, it's only the overall loudness of the signal that's likely to have an effect, not more subtle issues of its content.
Similar considerations apply to other components: only very simple physical considerations are relevant to wearing things out. How gritty the signal sounds to a human has little relevance; and to the extent program content matters at all, it's usually just a question of sheer volume, not style.
I think part of the perception comes from human experience with stuff like blowing out speakers. We expect that if you turn the volume up too high, you can damage something like a speaker - but speakers are partially mechanical devices, damaged by a mechanical process. We expect maybe that the people likely to do that sort of thing are the ones who also play harsher music and inflict violence on equipment in general. Think of rockers smashing their guitars (YouTube link). The distorted sounds you get after damaging speakers or other equipment also sound a whole lot like the distortion that's sometimes created deliberately by less destructive techniques in some styles of music, so there may be an expectation that distortion and damage naturally go together. But if you turn up the power amplifier way too high, you can still blow out your speakers just as hard with light adult contemporary; and this effect is caused by power level alone. There is nothing going on here except basic physics.
Now, if your playing style affects how you physically manipulate the modular - if you're flipping switches and frantically wiggling knobs on every beat - then that's obviously going to have consequences of simple physical wear as compared to a different style where you maybe you would just set up a patch and leave it to run for an hour without touching it.
I mentioned radioactive decay as a possible cause of slow failures in electronic equipment. It's not something that comes up very often for today's synthesizer users, but there is one case that may be relevant occasionally in synth DIY, and it's interesting to think about, so worth mentioning.
Neon glow lamps were once popular among hobbyists for doing switching and logic operations in the pre-transistor days when there were few other affordable options. It's especially easy to build a sawtooth relaxation oscillator using a neon bulb as the active element: the circuit is basically just a resistor, capacitor, the neon bulb, and a relatively high DC voltage power supply. The "electric organ" circuit based on neon bulbs, with one such oscillator for each note, is an old standby in collections of hobbyist project ideas.
Neon glow lamps work by passing current through ionized neon gas. When the lamp is actually in operation, the ionization is self-sustaining, but it decays in a fraction of a second after the lamp is switched off. So to turn it back on again, there needs to be a way to get the ionization started. Usually, that's a pulse at a relatively high strike voltage (which might typically be 60V to 100V).
Ambient light - that is, light coming from some other source, not the lamp itself - will partially ionize the gas, with the effect of lowering the necessary strike voltage to turn the lamp on. If you try to turn on a neon lamp in a completely dark environment, the strike voltage may be significantly higher than otherwise, in what is called the dark effect. (This does mean you can use a neon lamp as a light detector, although getting consistent results is difficult.) Logic and oscillator applications especially benefit from keeping the strike voltage relatively low and as consistent as possible. You don't want to have several bulbs near each other in a computer influencing each other through exchange of light rather than through the wiring, nor two adjacent notes in an organ frequency-locking to each other because of an exchange of light.
So to reduce the dark effect, these lamps used to be made with a little bit of krypton-85 mixed into the neon. This is a radioactive byproduct from the nuclear power industry; it produces a small amount of beta radiation which keeps the neon slightly ionized all the time, both reducing the strike voltage on average, and making the strike voltage less light-sensitive.
Such lamps are no longer commonly made at all, and to the extent they are made, the present-day versions usually don't contain radioactive substances. But neither do the old ones, anymore, because the half-life of krypton-85 is only about 11 years. If you try to build a vintage circuit with vintage parts and you obtain some "new old stock" neon bulbs made in 1960, in 2022 more than 98% of the original radioactivity is gone.
Logic and oscillator circuits based on neon bulbs are notoriously finicky and hard to get working. Part of the reason is just that the components are inherently unpredictable. That is part of their charm, from today's point of view. But another part of the reason is that even if you get the original components, decades of radioactive decay have put them far off their original specifications. The bulbs might have been designed, back in the day, with enough radioactivity to keep working well ten or twenty years after manufacture, but not to last all the way into our time. And there is basically no way to prevent or slow radioactive decay. It occurs inevitably with the passage of time, uninfluenced by any effects easily accessible to humans.
In future articles of this series I plan to talk about power-cycling and per-module switches; the effects of heat, especially on capacitors; and power-related insults like short circuits - not necessarily in that order.
◀ PREV Equivalent circuits || Power cycling and extra switches NEXT ▶