Burning In

Brian Dunning’s usually excellent Skeptoid podcast this week contains a blooper on the subject of burning in equipment. This podcast was a collection of short answers to student questions, one of which was:

Hello Brian, my name is Julian and I am from Malaysia, and my question is: Do burning in headphones improve sound quality?

His answer:

No, burn-in of headphones and other audio equipment is just one more dimension of the snake-oil world of high end audio, akin to super-duper speaker wire. Burn-in is the process of turning on new equipment, sometimes under extreme conditions, to reveal defects. It’s a common, and almost always worthless, tacked-on optional extra by some retailers of electronic equipment. Once in a while burn-in will reveal a defective component, thus saving the customer the trouble of taking the device home to discover it on his own; but as far as burn-in actually improving the performance of consumer electronics, then no, there is no evidence or plausible reasoning behind this.

Obviously I agree with most of this, but there are two problems. First, Brian seems to misunderstand what ‘burn in’ means conventionally amongst audiophiles. It is not to expose defects. Indeed, it isn’t even to push equipment hard. It is running the equipment normally for some hours, or more typically, tens of hours. Some audiophiles claim that this improves the sound of the equipment.

The second error he makes is in equating the equipment about which the question was asked — headphones — with electronic equipment. These are very different beasts. Headphones are, like loudspeakers, electro-mechanical devices. They convert electrical energy into acoustical energy (which is simply mechanical energy in the form of compression waves in air). They do this by using a linear electric motor to push a cone or diaphragm of some kind. This is designed to be extremely stiff in the frequency range of operation, but nothing is perfect and to some degree or other it flexes. This cone or diaphragm is supported by a suspension system, called a ‘spider’ in the case of a loudspeaker. This is often a stiff fabric rendered springy by means of concertina folding. The edge of the cone or diaphragm is typically also surrounded by some material in order to locate it in space. In speakers this is usually either some kind of foam, or a soft rubber.

All these moving bits change their state with use. The assumption is that they are stiff to begin with and loosen up with use, and this seems to make sense. The assumption also seems to be that they loosen up from their initial stiff state to a normal operational state fairly quickly — say, within dozens of hours of use — and maintain this state for a long time. Think of a tipped over ‘S’ curve.

Whether or not this has any perceptible effect on sound I do not know because I have never done a test. The procedure for such a test would be easy enough. Take two sets of same-model loudspeakers. Use a panel of listeners of sufficient variety and number to generate statistically valid results. Have them listen to both sets of speakers new out of the box, but only briefly, and score any differences in sound.

Run one of the sets of speakers extensively for a lengthy period to burn them in. Get the listening panel back and, double blinded, compare the barely-used and heavily-used speakers again. Score any differences in sound. If this second score is significantly greater than the first score, then the existence of an audible effect would have been established.

Some people swear that burn-in does make a difference in their experience. I suggest that it’s almost impossible to tell without some formal protocol as outlined. You cannot compare the sound of a device from one day to the next in the way that most people do these things. All you can do is compare the sound of a device with your memory of how it previously sounded, and I’d strongly suggest that you shouldn’t trust such a comparison.

But that doesn’t take away from the fact that it is almost certain some physical changes do take place in loudspeakers (and headphones) when they are run.

Electronics? Cables? No way.

This entry was posted in Audio, How Things Work, Imperfect perception, Mysticism. Bookmark the permalink.

9 Responses to Burning In

  1. Rod Williams says:

    Seems like a good question for the Mythbusters to tackle.


  2. alebonau says:

    electronics ? most definitely. if ever changed capacitors with fresh ones youd know how much they change in sound over first 24 hours. They do indeed settle down with time.

  3. treblid says:

    I just got new cables.. 😛 Will get back to you guys in 6 months time and say if cables require burn in (if my past experience is right, bass will be more defined, top end less harsh, and the whole presentation will sound a tad more “real”)…

    Too bad I can only afford one pair at a time.. If only I can buy 2, then I can keep one as is unconnected (control), and the other in use (test). The two sets will be in the same room so any environmental influences should be the same. They aren’t expensive (<$100), but I can't justify spending $100 extra just to satisfy other's curiosity.

    As to electronics. The CD player is a mechanical device with moving parts, the laser will age also. So I can use that to explain why sound changes over time. However I have experienced components sounding different over time too – discrete opamps. Is the soldering the reason behind this?

    The volume knob is also a mechanical device it seems, and an aging volume control seems to introduce noise.

    To suggest there is no burn in/run in, is to suggest electronics never wear out. So it stands to reason my electronics will last forever. I hope this is true :p.

  4. therationalaudiophile.wordpress.com says:

    I don’t actually share your faith in listening tests; if I was trying to do this, it would be with measurements only.

    My objection to the listening test experiment would be that there were too many variables to control. Speakers even from the same production run will have finite, measurable differences between the drivers, and they will be affected by temperature, and even humidity – as will the listeners’ ears. If, on one particular day it is not possible to discern differences between them, but on another day it is, I don’t think this would prove that it was the burn-in that was responsible.

    Another factor might be that the listeners began to become trained to discern differences after the burn-in, but that it was the finite differences between the speakers (that were there all along) they were learning to hear.

    On the other side of the coin, if, as is likely, the listening test failed to show up a difference then it could simply be that the choice of music was ‘wrong’, or the volume setting, or the room’s acoustics, temperature, humidity etc. Of course these conditions could be stated with the findings, but this would still not make the result ‘proof’ of anything.

    I say all this as an armchair scientist who has never conducted a listening test in his life, but I don’t think that’s necessarily a disadvantage..!

  5. You can measure stuff all you like, and I do a fair bit of measurement, but there remains a problem: some people insist that A sounds different to B, even though there are no measurable differences. Equally, tests have been conducted in which A and B produce different measurements, but properly conducted listening trials reveal no-one can hear the difference between them (think high resolution audio vs CD standard, for example).

    Double blind tests remain, for all their weaknesses, the only way to determine if there is a difference in what we mostly care about: the sound. The blinding is essential because time and again tests have shown listeners imagining they hear differences between pieces of equipment, which disappear once they no longer know which piece of equipment is producing the sound.

    Yes, there are all kinds of complications about environmental conditions and possible differences between units. If the matter were really important, then you’d spend hundreds of thousands of dollars doing the test by holding the environment constant and using lots of different speakers from the same batch, randomising them, running in half of them and not running in the other half. You’d need lots of different listeners as well.

    Set up ABX testing, or cable switching as preferred. Level match — or should you? If the speaker sensitivity increases due to running in, that in itself is a valid test. The first aim of the test is to determine whether nor not there is a difference. Only later would come the second part: does the run-in speaker actually sound better?

    One thing you could be confident of: if the test returned no significant difference between running in and not running in, there would be plenty of people quibbling about the test protocol.

  6. therationalaudiophile.wordpress.com says:

    Hello Stephen. Yes, I accept all that except, perhaps, for the premise that it’s actually worth doing it, most of the time! As you say, if it was really important then it would cost a lot of money to do it properly, and if not, then why bother?!

    In order to carry out a test, the alternative setups need to exist already, and I expect that the superiority of one over the other will already be known, pretty well, simply on a theoretical basis.

    The principle of level matching, alone, bothers me. In order to truly match levels, it implies that the signals are identical already! I presume that if we were comparing an amplifier with feedback versus one without, it would not be possible to ever say we had truly matched the levels, except by simplistically stating “We matched levels on a 1kHz sine wave at 1V RMS output into an 8R resistor” or something like that. The same for comparing speakers that may, or may not, compress differently, have different bass roll-offs etc. The results as to whether there was a difference in, or preference for, the sound after the initial nominal matching of levels would surely be meaningless..?

    Can we not reach the answers we are looking for by ‘deduction’ and other indirect methods?

    (Very pleased to have discovered your blog by the way. I’m enjoying rummaging through the archives).

  7. Thanks for the compliment. You ought to put the link to your website in the ‘Web’ box when you’re commenting so people can click on it and go straight there. I’ve had a little look through — pretty busy at the moment — and I love the focus on rationality.

    As for level matching, it doesn’t really matter what you match on. If the two things are so different that the level matches at different respective gains in different parts of the signal, then the level matching was probably unnecessary. For things like electronics, though, it has long been known that if two otherwise identical signals are played, the slightly louder one will generally be perceived as sounding ‘better’. This effect can appear for level differences of not much more than 0.1dB. However if matched, the ability to distinguish the signals can disappear.

  8. “You ought to put the link to your website in the ‘Web’ box when you’re commenting so people can click on it and go straight there.”

    Oops! Didn’t notice that box…

  9. Sorry for the delay … I didn’t realise that you’d be held for moderation again once you put some stuff in the web box.

Leave a Reply

Your email address will not be published. Required fields are marked *