Was burn-in ever a thing? Maybe with tubes where stuff "cooks off" the cathode when it's first turned on. That I have measured. The THD of my DG300B dropped by 6 dB from fresh tubes to four hours in. It was easily measured. That said, this is for a sample size of N = 2, so not scientifically valid, but a data point, nonetheless.
I would argue that any source or amplifier that needs burn-in is a failed design. First off, drift and aging are completely uncontrolled processes, so why would you hang the performance of your circuit on that? Secondly, if the components drift or age enough to cause performance shifts on time scales measured in hours or days you really should revisit that circuit and either redesign it or choose some better parts. These parameters are specified for components, ya know...
For electromechanical components I can maybe see an argument for burn-in. Maybe. Maybe the suspension on a woofer needs a little loosening up. But any speaker designer I've talked to has said that this loosening up happens in a few seconds of use. So basically you run a sine sweep and the speaker is broken in.
I also recall reading an article where a PA speaker repair guy spoke of break-in. He measured the speakers when they went into the field and when they came back for repair years later. Obviously, he must have compared the still-working drivers against their performance when new. He found no change in the measured parameters. This after years of (ab)use in some cases.
To me, this indicates that break-in happens between our ears. I'd further argue that manufacturers' claims of requiring a specific break-in protocol is 200% marketing. If I break my speakers in with Brahms they won't be able to play Beethoven, ever? Gimme a break! Also: Why would I buy a speaker like that? What if I discover new music that I like?
That said, I can't exactly blame the manufacturers for playing into the break-in narrative. Many customers expect that hifi gear requires break-in, so it presents an opportunity for the manufacturer to say "our gear is special. It requires this specific protocol that you MUST follow" (yes, a manufacturer here made such claims). And, also, the customer will get used to the gear over time and likely like it more (Mere Exposure Effect).
A different way to look at this is to approach it from an engineering perspective. I'm not a mechanical engineer, but I have an extensive electrical engineering background. In EE we always aim to make circuits that perform consistently over time. Drift and aging are bad things because they lead to failures. Customers get upset when their circuits drift out of spec and needs servicing every three months or fail within a few years. Why would mechanical engineers be any different? I would be extremely surprised if ME's design their systems to age to the specified performance. Rather I would expect a mechanical system to be designed to meet spec at the beginning and continue to meet spec until it is considered worn out (i.e., out of spec). If a part wears out in three months and the consumer expected to get 10+ years out of it, consumers become upset and engineers get spoken to by the pointy-haired boss.
More like test = N*time = N*$. It's a serial process after all. It's not like the photolithographic process used to manufacture ICs where you can make thousands of ICs with one flash of UV light. (OK, more than one flash but you get the idea). In the test & measurement world there's a limited number of test setups so testing thousands at a time would be insanely expensive.
It wouldn't surprise me if many manufacturers - especially of insanely inexpensive gear - leave the quality control to the customers. Most products will work when assembled correctly and most products are assembled correctly. It's cheaper to just swap out the defective ones when the customer complains than it is to test them.
Tom
I worked for a small instrumentation equipment company many years ago (I left in 1993!) and we were getting early hard field failures on some of our products so the company invested in a very large soak oven and everything would go in there for 24 or 48 hours (I cannot recall for exactly how long) at elevated temperatures and it pretty much solved the issue - products would, for the most part, show their failures up early and the stuff was re-worked and then re-soaked.
But, I doubt doing this would break in an amplifier and make it sound better.
Motivated reasoning is a cognitive and social response, in which individuals, consciously or unconsciously, allow emotion-loaded motivational biases to affect how new information is perceived. Individuals tend to favor arguments that support their current beliefs and reject new information that contradicts these beliefs.[1]
Motivated reasoning, confirmation bias and cognitive dissonance are closely related to each other.
https://en.wikipedia.org/wiki/Motivated_reasoning
Also fits what Kahneman said about jumping to conclusions: https://www.apa.org/monitor/2012/02/conclusions
System 1 is a storyteller. It tells the best stories that it can from the information available, even when the information is sparse or unreliable. And that makes stories that are based on very different qualities of evidence equally compelling. Our measure of how "good" a story is—how confident we are in its accuracy—is not an evaluation of the reliability of the evidence and its quality, it's a measure of the coherence of the story.
No doubt you will get some support from people who like your story whether or not its exactly true. At least you told a coherent story, which is probably enough for some people.
Last edited:
What people seem to mean when they use the term "burn in" is, IMO, really a non issue. While components all experience drift, usually accelerated at the start of their life cycle, the effective degree of change just doesn't matter for something perceived on such a gross scale as audio or similar - excepting, possibly, a mechanical process. All of which is not to say that "burn in" or more accurately called "ageing" is not something that can just be ignored - just not in systems with such large tolerances. In the world of precision metrology it is a very real and is something that must be considered. So, the effects of ageing on your very expensive piece of test equipment is certainly a concern but typically not in your amplifier. At least that is the way it should work... but when something is constructed with low quality components or just is a bad design then anything goes...
Absolutely Hal!
In test and measurement, aging and slight component shifts in value matter to performance. In older equipment it may make a device indicate out of tolerance. These days calibration cycles are around two years, it used to be 1 year (manufacturer recommended). For lab assets it could be 6 months, and there was an aging term to the accuracy specifications.
In audio, bias current and DC offset may drift a little in an amplifier. Certainly for CD players and tuners, component drift can affect performance. Once reset after a couple years it would normally remain in tolerance. I like to check bias currents in amplifiers every 10 years, and of course everyone forgets. In the event that bias drops and distortion is raised, people adjust with it and don't realise it. Or it drifts way up and runs hot. It all just depends.
You can see the total gain in some solid state amplifiers change with temperature slightly. I measure this with almost every amplifier I test. No one complains and they are completely unaware. The change isn't enough to hear for sure, but extremely easy to measure.
Some designs are very susceptible to component drift affecting performance. I agree with Tom, a bad design. But everything is affected by component drift to some extent.
In test and measurement, aging and slight component shifts in value matter to performance. In older equipment it may make a device indicate out of tolerance. These days calibration cycles are around two years, it used to be 1 year (manufacturer recommended). For lab assets it could be 6 months, and there was an aging term to the accuracy specifications.
In audio, bias current and DC offset may drift a little in an amplifier. Certainly for CD players and tuners, component drift can affect performance. Once reset after a couple years it would normally remain in tolerance. I like to check bias currents in amplifiers every 10 years, and of course everyone forgets. In the event that bias drops and distortion is raised, people adjust with it and don't realise it. Or it drifts way up and runs hot. It all just depends.
You can see the total gain in some solid state amplifiers change with temperature slightly. I measure this with almost every amplifier I test. No one complains and they are completely unaware. The change isn't enough to hear for sure, but extremely easy to measure.
Some designs are very susceptible to component drift affecting performance. I agree with Tom, a bad design. But everything is affected by component drift to some extent.
@Markw4 - I suggest you view your own posts through Kahneman's lens. You may find it insightful. I'm keenly familiar with his work and other aspects of cognitive psychology.
And if you really wanted to you could send your audio amp out for calibration periodically. I highly doubt any aging would impact the performance enough to be audible. Maybe measurable, but not audible ... unless you believe it is, of course. But that's your brain and not the circuitry making that difference.
Crappy circuit design will lead to crappy circuits. No surprises there. Just like cost cutting in airplane manufacturing seems to lead to major parts falling off of aircraft mid-flight.
Tom
Well, yeah. Also the performance of most test equipment is pushing the boundaries of physics, hence, the need for periodic calibration. Though, much of the calibration that's done is really just a (re-)confirmation that the equipment still passes spec. At least in my experience from National Semiconductor and TI where we had a quite extensive collection of high-end test equipment, the equipment was in spec at the end of the calibration cycle (usually one year) for the vast majority of samples.
And if you really wanted to you could send your audio amp out for calibration periodically. I highly doubt any aging would impact the performance enough to be audible. Maybe measurable, but not audible ... unless you believe it is, of course. But that's your brain and not the circuitry making that difference.
Crappy circuit design will lead to crappy circuits. No surprises there. Just like cost cutting in airplane manufacturing seems to lead to major parts falling off of aircraft mid-flight.
Tom
I wouldn't consider a bias to be a fault. It's a heuristic (in the psychological definition of the word).
Tom
Tom
Hi Tom,
Agreed, I was a calibration tech for a while. Most good equipment passed certification on a level 6 cal. Cheaper stuff was typically out of tolerance, and often out straight out of the box from the factory. However, the costs to the user of the test equipment if it was out of tolerance could lead to recalls of work, or components on a huge scale. Millions if the affected thing was on a satellite or under the sea when it failed to perform properly.
For audio, failure is the biggest concern. Gross distortion might occur in a tuner or CD/DVD player. I have seen bias current drop to near zero, and that was audible, but over the course of years. Crappy circuits? Yeah. I have also seen bias current climb to stupid high levels, again circuit design.
It is reasonable to at least check your audio equipment on a decade basis, certainly not every year unless you have tube stuff and want to see if it is retube time.
Agreed, I was a calibration tech for a while. Most good equipment passed certification on a level 6 cal. Cheaper stuff was typically out of tolerance, and often out straight out of the box from the factory. However, the costs to the user of the test equipment if it was out of tolerance could lead to recalls of work, or components on a huge scale. Millions if the affected thing was on a satellite or under the sea when it failed to perform properly.
For audio, failure is the biggest concern. Gross distortion might occur in a tuner or CD/DVD player. I have seen bias current drop to near zero, and that was audible, but over the course of years. Crappy circuits? Yeah. I have also seen bias current climb to stupid high levels, again circuit design.
It is reasonable to at least check your audio equipment on a decade basis, certainly not every year unless you have tube stuff and want to see if it is retube time.
I just spent the day re-doing the way we track our equipment for calibration. Included in that are 4 Fluke 5720A's and a stack of Keysight 3458As - all of which we send out for calibration on a yearly basis because of contractual requirements. But looking over 10 years of "as found" data there was only one instance of a failing value and that turned out to be a warranty issue not a drift out of specifications. So @tomchr hit it dead on by saying that the calibration of high end gear is usually just a reassurance.
Hal
Hal
You don't think Tom's story was a troll?
I know for a fact that while it may contain some truth, it also paints some good people in a false light.
That's all for today. Hopefully you will talk about capacitors and not about manufacturer conspiracies.
I know for a fact that while it may contain some truth, it also paints some good people in a false light.
That's all for today. Hopefully you will talk about capacitors and not about manufacturer conspiracies.
Hi Mark,
No, Tom's story is not. But you dragged in a bunch of stuff right after that really hasn't a purpose. PM me if you can explain so I understand.
Let's leave psychology out of it where it doesn't pertain to the topic directly. That's your go-to.
I don't know that it paints anyone in a poor light Mark.
Manufacturer's conspire to save money at the end user's expense. More so these days. But this was a topic related to reliability and long term performance. I've seen this first-hand often enough across three major industries. It doesn't hurt to be aware of this.
No, Tom's story is not. But you dragged in a bunch of stuff right after that really hasn't a purpose. PM me if you can explain so I understand.
Let's leave psychology out of it where it doesn't pertain to the topic directly. That's your go-to.
I don't know that it paints anyone in a poor light Mark.
Manufacturer's conspire to save money at the end user's expense. More so these days. But this was a topic related to reliability and long term performance. I've seen this first-hand often enough across three major industries. It doesn't hurt to be aware of this.
Okay, I'll argue this. Why bother? How do you know what value is undersized and what's "right sized?" What if the value is so high it upsets the bias for the first five minutes of being powered on? "Oh, but the capacitor's contribution to distortion is 240dB down!"
Not to dismiss Douglas Self's work, but there's high school classes that still the periods and lengths of pendulums, in spite of Newton's work (and not to dismiss his work either). Distortion values of different types of capacitors aren't as consistent as pendulums on Earth's surface.
Hi benb,
So what is your point here? You are not arguing anything specific, and no one indicated any value for you to take exception to. -240 dB referenced to what? Say 1 volt, that's way below the thermal noise floor of everything.
If you don't know what value is undersized, you shouldn't be designing or modifying anything. The definition is easy though. No appreciable signal voltage across the terminals. You can also do this by trial and error by watching distortion as you vary the size of the capacitor. Besides, this is entirely up to you if you're messing with the circuit.
If the value is high enough to upset audio circuit conditions, it's a bad design. For some tests in labs you may need to allow a day or so soak time, but then we are at the hairy edge of physics measuring tiny values using extremely expensive equipment in temperature and humidity controlled areas.
As for variables between capacitors - yeah. What did you expect? You'll get differences between two capacitors next to each other made at the same time, same batch.
So what is your point here? You are not arguing anything specific, and no one indicated any value for you to take exception to. -240 dB referenced to what? Say 1 volt, that's way below the thermal noise floor of everything.
If you don't know what value is undersized, you shouldn't be designing or modifying anything. The definition is easy though. No appreciable signal voltage across the terminals. You can also do this by trial and error by watching distortion as you vary the size of the capacitor. Besides, this is entirely up to you if you're messing with the circuit.
If the value is high enough to upset audio circuit conditions, it's a bad design. For some tests in labs you may need to allow a day or so soak time, but then we are at the hairy edge of physics measuring tiny values using extremely expensive equipment in temperature and humidity controlled areas.
As for variables between capacitors - yeah. What did you expect? You'll get differences between two capacitors next to each other made at the same time, same batch.
The rule off thumb is you make the cap 10x the required value for the LF frequency of interest. So if the calculation says 10uF for 20 Hz, use 100uF and shift the corner freq down to 2 Hz.
If your circuit bias is getting screwed up for minutes because of that, I’d argue you have other, more severe problems to worry about.
BTW, you will not see any distortion at say 1kHz - distortion only starts to rise through an electrolytic when it develops an AC voltage across its terminals so the problem is at LF with undersized cap values. Self did some work on this and the magic number is > 60 or 70 mV.
Make sure a capacitor is the right way round.
With all due respect, how do you know the "stuff" brought in by MarkW4 hasn't a purpose that someone else can see? Why PM you to explain as an aside? Why can't we see his explanation without the seeming threat of being censured? I might have warranted such action but he doesn't seem ever to have warranted it.
What is going on? You seem behaving as peer reviewer of all substance submitted.
My best friend for perhaps 40 years at Western University (and that doesn't mean I am or ever was his as a lowly tech) is a Dr. Kucerovsky a physicist/mathematician/engineer. He received a gold medal as a peer reviewer for submission of materials to a Journal of Physics. It seems important to keep in mind that unlike "The Journal of Physics", the minimum standard for submission to DiyAudio is "Make sure a capacitor is the right way round" and expect having met that standard in this post.
And now I have a certain song by Dead or Alive stuck in my head. "Like a cap, baby, right 'round, 'round, 'round"
Whole lot of wild oats sowed to that one back in the day...