Do Audio Speakers Break-in?

[MAB]

I would reject the report because the people involved apparently haven't looked into the industry at all, where such things have been intensively researched for well over 50 years.
Industrial membranes and certain surrounds are very similar to loudspeaker drivers in terms of function and materials. The measurements show a stable operating point after approximately 50-200 hours, depending on the material, even more. However, the stresses are higher, and lower stresses prolong this initial aging process.

The driver measurements I see indicate the break-in effect occurs in the first few moments of operation, is measurable but very small and inconsequential to the reproduced sound. Perhaps link the actual research you mention.

Based on this, the stated 3 hours are far too short, and the result is very questionable in my opinion.

No, the primary material property changes occur in the first few moments. Andrew Jones discusses this:

https://www.audiosciencereview.com/forum/index.php?account/alerts

Vance Dickason discusses this in his books; that while driver break-in is measurable, it happens early in operation of the driver, and are tiny:

"As can be seen, the changes in box performance are trivial."
Vance Dickason, Loudspeaker Design Cookbook, 6th edition, Chapter 8, Section 8.20 (Break-In) page 155

Dickason indicates the real reason to test drivers prior to installation, defects and out of spec units. Which is my experience, as can be seen below if you carefully read the posts.

I only skimmed the whole thing, but perhaps I missed something, and they also conducted tests after a few hundred hours.

Yet you dismiss without evidence.

Did they also consider the different materials and material pairings? For example, the two extremes in midrange and woofer drivers? Foam surrounds in conjunction with very soft spiders, or thick rubber surrounds with very hard spiders? Or what about cone materials like polypropylene, glass, Kevlar, carbon fiber fabric, paper, sandwich compounds, etc.?
All these materials change to a greater or lesser degree. Some become softer, for example, fabric-based materials, while some plastics become harder and more brittle due to the outgassing of plasticizers and solvents. Not to mention the effects on resonances.

Perhaps show this. I was really surprised to see this old urban legend of driver break-in was alive and well at ASR, despite the observations of Dickason and Jones and other speaker experts, and despite how easy it is to actually test this. I was also surprised at how many members cling to the idea that break-in is a significant issue.

So I tested, some are in this thread. Others are here:

Speaker Break-In and Environmental Measurements

The subject of speaker break-in comes up often. It's also called run-in. It's different than aging, or wear-out. I see people posting YouTube videos of woofers being deformed into the shape of a potato-chip, while the 'tuber yells at me to make sure and break in my drivers before ever...

www.audiosciencereview.com

I show that speaker break-in is tiny, in fact on the order of room temperature fluctuations and smaller than drive unit heating due to operation. In two cases, driver to driver matching is worse than break-in effect. At some point I will try to curate all of the driver break-in tests I have done, fortunately I have many drivers from building speakers, and always test them when new, since discrepant drivers are more common than many people believe, for example is very expensive Seas unit with improper adhesive from the factory:

I tested it's replacement that was shipped to me, it had break-in that was tiny.

So yeah, driver break-in happens but is tiny. And as Dickason says, the main reason for doing break-in testing is to weed out discrepant or defective drivers, which involves testing before and after break-in. It's certainly unfortunate since driver break-in is one of the old tricks in the book used to shame customers into keeping bad sounding speakers. "It will always sound better after break-in."

 

[ahofer]

The driver measurements I see indicate the break-in effect occurs in the first few moments of operation, is measurable but very small and inconsequential to the reproduced sound. Perhaps link the actual research you mention.

No, the primary material property changes occur in the first few moments. Andrew Jones discusses this:

https://www.audiosciencereview.com/forum/index.php?account/alerts

Vance Dickason discusses this in his books; that while driver break-in is measurable, it happens early in operation of the driver, and are tiny:

Dickason indicates the real reason to test drivers prior to installation, defects and out of spec units. Which is my experience, as can be seen below if you carefully read the posts.


Yet you dismiss without evidence.

Perhaps show this. I was really surprised to see this old urban legend of driver break-in was alive and well at ASR, despite the observations of Dickason and Jones and other speaker experts, and despite how easy it is to actually test this. I was also surprised at how many members cling to the idea that break-in is a significant issue.

So I tested, some are in this thread. Others are here:

Speaker Break-In and Environmental Measurements

The subject of speaker break-in comes up often. It's also called run-in. It's different than aging, or wear-out. I see people posting YouTube videos of woofers being deformed into the shape of a potato-chip, while the 'tuber yells at me to make sure and break in my drivers before ever...

www.audiosciencereview.com

I show that speaker break-in is tiny, in fact on the order of room temperature fluctuations and smaller than drive unit heating due to operation. In two cases, driver to driver matching is worse than break-in effect. At some point I will try to curate all of the driver break-in tests I have done, fortunately I have many drivers from building speakers, and always test them when new, since discrepant drivers are more common than many people believe, for example is very expensive Seas unit with improper adhesive from the factory:
View attachment 497294
I tested it's replacement that was shipped to me, it had break-in that was tiny.

So yeah, driver break-in happens but is tiny. And as Dickason says, the main reason for doing break-in testing is to weed out discrepant or defective drivers, which involves testing before and after break-in. It's certainly unfortunate since driver break-in is one of the old tricks in the book used to shame customers into keeping bad sounding speakers. "It will always sound better after break-in."

Thank you. My confidence level has increased about this.

 

[krabapple]

I would reject the report because the people involved apparently haven't looked into the industry at all, where such things have been intensively researched for well over 50 years.

But some people right here at ASR have followed the 'you need hours of burn-in' clown show for years, decades even -- and the results have remained consistent: no good evidence that it audibly matters at all.

Show us where 'the industry' demonstrated otherwise, using blind listening test results and ruling out normal unit-to-unit variation -- not marketing text.

Maybe it's you who has not kept up.

 

[Sal1950]

Perhaps show this. I was really surprised to see this old urban legend of driver break-in was alive and well at ASR, despite the observations of Dickason and Jones and other speaker experts, and despite how easy it is to actually test this. I was also surprised at how many members cling to the idea that break-in is a significant issue.

You just can't argue with "I hear it, so it is so" beliefs.
Like religion, the audiophile community has it's distorted views of reality that are near impossible to address.

 

[a4eaudio]

It is interesting that this thread is at 2K replies. Unfortunately, I don't think the title of the thread is the intended question. It is a KNOWN FACT that T/S parameters of a brand new, never used, speaker driver (at least woofer, probably not tweeters) change after break in. There have been numerous experiments that show the T/S parameters continue to change well after a 5-15 minute break-in period...the TS parameters after 48 hours break-in are usually even a little different than after 24 hours. The answer to the title of this thread is an unambiguous, undisputable "YES" - audio speakers break in. An Amirm's post #1 clearly shows a difference.

I think what was MEANT to be asked was "do audio speakers undergo any AUDIBLE CHANGE after breaking in." I don't know of any good study examining this and I would not be surprised at all it there is no audible difference. But consider these two errors...(1) speakers need to be broken in and you don't break them in, in which case you are NOT getting an accurate representation of the speaker or (2) speakers do not need to be broken in, you break them in anyways, and it didn't matter. Why not just go with option 2? After all, we are on a site that daily reports levels of SINAD and speaker distortion that are not audible.

 

[MAB]

Thank you. My confidence level has increased about this.

Toole has commented on this as well:

Dickason points out that with the change in Fs after break-in, Fs/Qts stays constant. You can see this even in the one example I presented of a driver with a large change in Fs, a HiVi micro woofer. This class of small woofers is greatly affected by any change in spider flexibility. Andrew Jones talks about spider in the video interview, that the lacquer on the spider fabric becomes flexible after the first flex or two. Can even be done by hand. Won't be present on any driver that is pre-flexed at the factory in any way. Even these cheap HiVi maintain constant Fs/Qts ratio between pre and post break-in.

The Fs/Qts ratio is the same across all of these runs, even due to the effect of temperature in run 3 and as the driver cools back to room, which is why our speakers don't sound dramatically different warm up over the first few minutes of listening. And for sure, most of us don't have arrays of micro-speakers for better or for worse. One thing is apparent from the above plots, out of the four drivers one is defective, likely has too many wire turns on the voice coil (driver C has higher inductance). Which is Dickason's point about break-in testing, it's all about finding defective drivers. This is surprisingly common, as are voice-coil rub and other maladies, is somewhat brand-dependent.

These HiVi micro-woofers do have lots of variation and surprising number of defective units, not surprising if you consider they were $300 for 100 units ($3 each) when I bought them. I regret I didn't track the defects. For sure it wasn't as bad as 25% as the above test above suggests. But something like 5% with VC rub and 5% with Le out of spec is probably in the ballpark. The Seas was very surprising, they were very prompt in shipping me a replacement as I expected, which I tested right out of the box and after break-in...

 

[krabapple]

It is interesting that this thread is at 2K replies. Unfortunately, I don't think the title of the thread is the intended question. It is a KNOWN FACT that T/S parameters of a brand new, never used, speaker driver (at least woofer, probably not tweeters) change after break in. There have been numerous experiments that show the T/S parameters continue to change well after a 5-15 minute break-in period...the TS parameters after 48 hours break-in are usually even a little different than after 24 hours. The answer to the title of this thread is an unambiguous, undisputable "YES" - audio speakers break in. An Amirm's post #1 clearly shows a difference.

Holy cow, someone measured a difference??
Alert the media.

Amir's post #1 take seems rather different than yours:
"The two measurements are essentially identical. This, despite the fact that the 1000 point measurement has much higher order expansion so brings with it more accuracy and some variations. Yet, the response is still the response. No way anything regarding tonality of the speaker changed. The same bass, mids and highs are there."

Maybe magnitude of difference matters? And do we even take unit-to-unit variance into account as the baseline?


Amir's conclusion:
Objective analysis shows that there is no change to the frequency response and hence tonality of the speaker after a few hours of intensive running (full sweeps from 20 to 20 kHz). This completely mirrors research performed at Harman where they tested a new woofer and a "broken-in" one in a real speaker. While physical changes are occurring in the driver, they are at such low level that there is no hope of attaching them to audibility.

 

[a4eaudio]

Amir's post #1 take seems rather different than yours:
"The two measurements are essentially identical.

My interpretation, which is the same as Amir's, just different words... The two measurements are not the same.

 

[MAB]

My interpretation, which is the same as Amir's, just different words... The two measurements are not the same.

Very few people in this thread are saying the measurements before and after break-in are the same. I certainly am not.
The differences are below audibility, and are not a reason to expect a speaker to change sound in the first moments, minutes, or hours of use. That was the point.

Separately, please show these numerous experiments you talk about, with evidence of audibility:

There have been numerous experiments that show the T/S parameters continue to change well after a 5-15 minute break-in period...the TS parameters after 48 hours break-in are usually even a little different than after 24 hours.

Would be good to see these.

 

[bmc0]

The two measurements are not the same

The measurement uncertainty is not given, but I can confidently say that it is not zero. One cannot conclude that any and all differences are attributable to the device under test and not measurement error.

 

[mlsstl]

Didn't read the whole thread but seems odd that the audiophiles who believe in really long breaking periods never seem to worry about their gear getting  too broken-in.

 

[a4eaudio]

Very few people in this thread are saying the measurements before and after break-in are the same. I certainly am not.
The differences are below audibility, and are not a reason to expect a speaker to change sound in the first moments, minutes, or hours of use. That was the point.

That may have been the point, but that is not the question in the thread title.
"Do Audio Speakers Break-in?" The answer seems to be a near-unanimous "yes".
If the answer is "Yes, but we don't know if it is audible" then wouldn't we advise taking the conservative approach and recommend everyone burn their speakers in for 24 hours just in case?

Separately, please show these numerous experiments you talk about, with evidence of audibility:

You missed what I said....

...I think what was MEANT to be asked was "do audio speakers undergo any AUDIBLE CHANGE after breaking in." I don't know of any good study examining this and I would not be surprised at all it there is no audible difference.

The measurement uncertainty is not given, but I can confidently say that it is not zero. One cannot conclude that any and all differences are attributable to the device under test and not measurement error.

If I take a measurement in my garage tonight, tomorrow night, and the next night you won't see any difference in the SPL measurement unless my wife turns on the washing machine upstairs. I can't imagine the Klippel NFS has that much measurement error.

Below is the difference of Amir's before and after break-in with the scale +/-1dB. (Sorry if this was already posted, I have not read all 1,700 posts.) "Fresh out of the box" is down 0.25dB almost everywhere from 1.5kHz down. While I can not hear a high-Q peak of 0.25dB, I would not be surprised at all if a speaker 0.25dB higher SPL for over 5 octaves was audible to well trained listeners. And this is ONE speaker. Is there any reason to believe that if Amir ran the same experiment on 100 speakers that there wouldn't be several where the deviation is +/-0.5dB or greater for 2+ octaves? So again, it would seem to me, that it would be conservative to recommend people break-in their speakers.

 

[tineared]

You are lucky they havent burnt to dust by now.

What are you talking about?
They're not even broken in yet...

 

[bmc0]

If I take a measurement in my garage tonight, tomorrow night, and the next night you won't see any difference in the SPL measurement

Try it. The measurements will not be exactly identical (although the error should be quite small).

Fresh out of the box" is down 0.25dB almost everywhere from 1.5kHz down.

There is no mechanism that I'm aware of that would explain a woofer's sensitivity increasing uniformly over essentially its entire bandwidth as a result of mechanical break in.

 

[krabapple]

My interpretation, which is the same as Amir's, just different words... The two measurements are not the same.

That you totally ignored these provisos in my post
"Maybe magnitude of difference matters? And do we even take unit-to-unit variance into account as the baseline?"

tells me you aren't serious.

If I take a measurement in my garage tonight, tomorrow night, and the next night you won't see any difference in the SPL measurement unless my wife turns on the washing machine upstairs. I can't imagine the Klippel NFS has that much measurement error.

Below is the difference of Amir's before and after break-in with the scale +/-1dB. (Sorry if this was already posted, I have not read all 1,700 posts.) "Fresh out of the box" is down 0.25dB almost everywhere from 1.5kHz down. While I can not hear a high-Q peak of 0.25dB, I would not be surprised at all if a speaker 0.25dB higher SPL for over 5 octaves was audible to well trained listeners. And this is ONE speaker. Is there any reason to believe that if Amir ran the same experiment on 100 speakers that there wouldn't be several where the deviation is +/-0.5dB or greater for 2+ octaves? So again, it would seem to me, that it would be conservative to recommend people break-in their speakers.

etc etc ad tedium...

The very most stringent level matching for blind comparisons insists on matching within 0.2dB, but 0.5dB is far more common, and in both cases that level of discrimination is very frequency-dependent. 0.25 dB difference is really, really, really stretching the world 'audible' for a loudspeaker difference.I would be VERY surprised if even 'well-trained listeners' could do it without specialized signals.

And again you haven't even ruled out unit-to-unit variance *at baseline*, much less measurement to measurement.

 

[a4eaudio]

Try it. The measurements will not be exactly identical (although the error should be quite small).

Yes, they exist, they are just too small to see in my default REW settings on my large monitors. (20Hz to 20kHz x-axis, 40dB range y-axis) Probably smaller than 0.05dB.

...There is no mechanism that I'm aware of that would explain a woofer's sensitivity increasing uniformly over essentially its entire bandwidth as a result of mechanical break in.

So you think the Klippel NFS isn't very accurate? The deviations in post 1,732 are just the difference between Amir's two measurements in post #1
(I have assumed that he just left the speaker in the exact same position and burnt it in and re-measured. But I would assume the NFS has pretty precise positioning of the speaker and he could take it off and put in back on and get repeatable results.)

 

[a4eaudio]

... do we even take unit-to-unit variance into account as the baseline?"

And again you haven't even ruled out unit-to-unit variance *at baseline*, much less measurement to measurment.

I don't understand the comments about unit to unit variance. Amir measured a speaker, burned it in, and measured the SAME speaker with a Klippel NFS.

 

[bmc0]

So you think the Klippel NFS isn't very accurate?

The accuracy depends in part on the number of measurement points, which was not the same for the two cases. Amir would know better than me, but that alone may explain much of the difference.

0.25 dB difference is really, really, really stretching the world 'audible' for a loudspeaker difference.

The detection threshold for a midrange resonance with Q=1 is on the order of 0.25dB spectral modification with a pink noise stimulus[1]. Thresholds for music are program dependent and generally a fair bit higher. Wider bandwidth spectral modifications may have somewhat lower thresholds, but yes, it's unlikely that 0.25dB would be detectable with music.

[1] F.E. Toole and S. E. Olive, "The Modification of Timbre by Resonances: Perception and Measurement."

 

[MAB]

That may have been the point, but that is not the question in the thread title.

I'm not really going to debate the title of the thread, perhaps contact Amir about the thread title vs. his OP.

"Do Audio Speakers Break-in?" The answer seems to be a near-unanimous "yes".

Yes.

If the answer is "Yes, but we don't know if it is audible" then wouldn't we advise taking the conservative approach and recommend everyone burn their speakers in for 24 hours just in case?

This seems to be your own conclusion. We do know about the audibility. Just listen for the first 24 hours and onwards, there is nothing to be conservative or radical about at all. And don't expect your speakers to intrinsically change sound in that period of time, or even after that.

The conclusion is "We know that break-in causes inaudible changes, despite the changes being measurable." See Dickason chapter 8, that I referred to earlier, he has a nice example and discussion of a driver with even a 10% mismatch in f(s), with <<0.1 dB change in output. Here is another detailed test using actual drivers measured pre and post break-in, and the resulting changes in output:

Again, ~0.02 dB isn't audible. The author also compares break-in changes with driver-to-driver variability and finds that driver-to-driver variation is much larger than the break-in changes.

Break-in changes are typically negligible, inaudible. That is the conclusion.

 

[a4eaudio]

The accuracy depends in part on the number of measurement points, which was not the same for the two cases. Amir would know better than me, but that alone may explain much of the difference.

An accurate on-axis SPL measurement of before and after break-in is NOT complicated and should be able to be done with a $25 Behringer mic, but I do find the measurement differences "interesting", especially the 1dB dip at 5.5kHz which would seem highly unlikely from break-in.

The detection threshold for a midrange resonance with Q=1 is on the order of 0.25dB spectral modification with a pink noise stimulus[1]. Thresholds for music are program dependent and generally a fair bit higher. Wider bandwidth spectral modifications may have somewhat lower thresholds, but yes, it's unlikely that 0.25dB would be detectable with music....

I put the deviations from post #1,732 into EQ (summed to mono on one speaker that has an F3 of 37Hz) and switched it on and off playing white noise and could not tell a difference. Since I could not tell a difference with white noise I didn't proceed to Fast Car.

But I would still argue that the conclusion in post #1 is inaccurate: "Objective analysis shows that there is no change to the frequency response and hence tonality of the speaker after a few hours of intensive running." There is a clear OBJECTIVE change in the frequency response - the black and orange lines are different from each other by as much as 0.5dB and even 1dB in some places. It is also clear that it is quite small - deviating by about 0.25dB in most places. The fact that I (and presumably most people) can't hear the difference is "subjective". I know a LOT of people on ASR have EQ capabilities, so it wouldn't be hard for people to see if they can hear the difference in post 1.732. However, I'm not sure how convincing the results from a sample of ONE good-quality Revel speaker is.