ErinsAudioCorner

[richard12511]

I get it. I do. But view it from my perspective... a dude with no professional ties to the industry and no extra money to blow (no pun intended) on a potentially destructive test.


If someone sending me a speaker doesn’t mind me going full tilt on it at the risk of permanent damage then I’m game. Otherwise, I’m using some my own willingness to push the envelope. Shooting for 96dB at 2 meters is reasonable.

I’ve also been trying to use the CEA-2034 spec for Max SPL (Tone bursts) but it takes literal hours and puts me in a state of fear. Prime example: with this particular speaker I was at 40v input which put the 1 meter output in to the mid-teens.. and it still didn’t exceed the distortion threshold for all frequencies. That test also took 3 hours. That was the fourth iteration. I had well over 8 hours in that testing. And it still wouldn’t resolve at all frequencies. So at that point I either keep pushing it, and run the risk of damaging a $900 speaker, or I provide an incomplete test. I definitely wasn’t willing to do the former. I don’t know anyone who would unless they were just looking to burn money.

Now, again, if someone wants to send me a speaker and gives me their blessing to potentially wreck it, or provide financial backing I will accept the challenge. Until then, however, there’s only so much stress my heart can take relative to the speaker stress. These compression tests are my alternative. At the output levels I am providing I think it’s a reasonable one, too.

I am not saying “no, never”. I am just saying “not today, sir”. I am looking at other options for the max SPL testing but my concerns still stand.

I totally get it. I dream of the day when a totally altruistic billionaire who doesn't mind losing all his money and who's only goal is to advance audio science emerges.

 

[richard12511]

That said, I can send you an Infinity Beta 20(I'll pay for shipping) to test. No need to return it, and you can compression test it to death. I've got 7 more, and the science is worth more to me than any one of them is.

 

[Juhazi]

That said, I can send you an Infinity Beta 20(I'll pay for shipping) to test. No need to return it, and you can compression test it to death. I've got 7 more, and the science is worth more to me than any one of them is.

it will be just one model... It is easy to do compression test at home, because we are looking at change of spl (delta). Response smoothness/validity is irrelevant

Good morning, I read through last pages, some opinions:
- 25dB/a decade of Hz is best ratio. Span of x and y can be anything... CEA-2034 combines two charts, actual responses/spl and DI, that's why it starts from 0dB.
- response beyond 90deg is very important and hard to find!
- 0-180deg response raw vs. normalized to on-ax hmm... please both!
- impedance, phase, step response and distortion at several voltage/dB by spectrum is important to people who understand speaker design and challenges of multiway speakers
- compression info can reveal bad decisions in construction, but most speakers will do fine, for home use. Subwoofers and PA gear are more critical about compression

Audio.com.pl uses 15dB/decade and looks too bad...

 

[edechamps]

@hardisj: can't you run a compression test that stops as soon as the speaker reaches the onset of compression? Seems like you would be way less likely to damage the speaker in that case. Presumably you'd need to run the speaker way past compression to damage it.

Of course, one problem with that plan is that the onset of compression is dependent on frequency… one way to work around that could be to run one compression test per portion of the frequency range (say, one per octave), instead of full sweeps. Or something more sophisticated where the test signal automatically adapts to avoid areas where compression was already reached in the previous steps…

 

[hardisj]

Before this gets completely off the tracks I'll state this and then I'm moving forward.

The goal of the distortion and compression testing is to see the performance of the speaker at normal and a bit above normal operating levels. That is why I provide the data at varied levels. Most people listen to music in the 75-85dB range. My experience also shows most who say they listen to music louder are either not using a calibrated SPL meter and/or not talking full-range where typical house curves have the subwoofer anywhere from 3-15dB higher than the midrange. I tend to listen loud: in the 85-90dB range. But 90dB at full spectrum is quite loud. FWIW, in my home theater I use JBL Pro Home Cinema speakers... literally from the old theater in town... consisting of a 15" midrange and 2" compression driver with 98dB and 118dB @ 1w/1m specs respectively. It gets loud. Using this real world example: My listening position is about 9 feet, but I'll round up to 3 meters. Doing the math...
80dB @ 3 meters is = 86dB @ 2 meters and 92dB @ 1 meter.
90dB @ 3 meters is = 96dB @ 2 meters and 102dB @ 1 meter.

Thus, my desire to test at output at 96dB to 102dB. In my experience, that's adequate to find faults that should cover the overwhelming majority of listeners ... and then some.

It is easy to diverge from practical to "experimentation". I love the latter as much as anyone here (or more; look at what I've gone through to get here). But there's a practical limit of what makes sense. For example, could I run IMD tests on a loudspeaker? Yes. Does it really need to happen on a loudspeaker? No. On a drive-unit? Yes. Why? Implementation: the drive-unit already has been implemented in a loudspeaker. An IMD test may reveal some information but in the grand scheme of things that doesn't tell us more more than what distortion and compression tests permit. I would rather spend my time quantifying metrics that tell me what I want to know as opposed to spending time getting data that doesn't further the purchase decision.

Finally, I am still working with the Klippel TBM module to test max SPL. As I mentioned previously, this test takes HOURS. It also needs a very quiet setting. And, as I said, there is a "pucker factor" because 40v fed in to a $900 mid-sensitivity speaker is downright scary, if I am being honest. This test is not trivial which is why that data has not been presented for this speaker. Having tested drive-units for years I can honestly say that testing a subwoofer for max SPL is a walk in the park compared to testing a loudspeaker. That's no disrespect to those who do it. Just being realistic. Most subwoofer drivers are a couple hundred bucks and are built not to fail at ridiculous output levels. Manufacturers know a subwoofer will get thrashed. But a loudspeaker manufacturer isn't expecting their audience to run their bookshelf to 130dB wide open. Having said that, I am still working on this and I hope that I can, at some point, provide such data.

Most importantly:
I am working hard to provide data to you guys. I believe all my efforts documented here in this thread are proof of that. So, please, if you don't mind, be considerate of your requests and my ability to provide them while also balancing a day-job and family. Understand, there's a practical limit to what I am willing to do with someone else's gear. I hope you all can respect that and appreciate the data I am providing rather than being concerned about what I am not providing and trust that if I can provide data of merit I will.

Thanks,
Erin

 

[hardisj]

it will be just one model... It is easy to do compression test at home, because we are looking at change of spl (delta). Response smoothness/validity is irrelevant

Good morning, I read through last pages, some opinions:
- 25dB/a decade of Hz is best ratio. Span of x and y can be anything... CEA-2034 combines two charts, actual responses/spl and DI, that's why it starts from 0dB.
- response beyond 90deg is very important and hard to find!
- 0-180deg response raw vs. normalized to on-ax hmm... please both!
- impedance, phase, step response and distortion at several voltage/dB by spectrum is important to people who understand speaker design and challenges of multiway speakers
- compression info can reveal bad decisions in construction, but most speakers will do fine, for home use. Subwoofers and PA gear are more critical about compression

I can provide all the FR you want. But the on-axis response and normalized spectrogram, IMHO, is far more usable than a whole lot of lines that I can't read through. Just my take.

I planned to provide impedance sweep at two levels: 0.10v and 2.83v. Maybe higher. The former is pretty much the standard because it is low enough to not drive the speaker in to non-linear range. But I think it would be more interesting to at least see what happens at higher voltages. I plan to lay these on top of each other assuming it is still legible (considering it will have phase & magnitude for each sweep voltage).

Audio.com.pl uses 15dB/decade and looks too bad...

Not only that, but the data capture is bad. The midrange is gated too low and apparently improperly (the dip/peak pattern below 2kHz is a tell-tale sign); it looks like they allowed reflection in to their result. Or the windowing is incorrect for their purpose (rectangular, half, hamming, blackmon, etc). You can't get anything useful from the midrange response due to their capture method. They need to use multiple measurement points and stitch them all together if they want to get anything of merit below the high-frequency range.

 

[MZKM]

That or they just meant to write 100:1 rather than 10:1.

I think you meant 1000:1 (20-20K); typos sure are easy.

 

[hardisj]

I think you meant 1000:1 (20-20K); typos sure are easy.

LOL.

 

[hardisj]

That said, I can send you an Infinity Beta 20(I'll pay for shipping) to test. No need to return it, and you can compression test it to death. I've got 7 more, and the science is worth more to me than any one of them is.

I'm OK with that. Shoot me a PM and we can talk about it.

 

[napilopez]

I think you meant 1000:1 (20-20K); typos sure are easy.

LOL.

While I am the king of typos (just ask my co-workers who have to edit my articles) 100:1 is correct . With the scaling intended/shown, a 25dB span = 20-200Hz, and a 50dB span = 20-2,000Hz. That's why I drew the red lines!

 

[hardisj]

While I am the king of typos (just ask my co-workers who have to edit my articles) 100:1 is correct . With the scaling intended/shown, a 25dB span = 20-200Hz, and a 50dB span = 20-2,000Hz. That's why I drew the red lines!

Oh, boo. I was hoping for fisticuffs.

 

[maty]

Hopefully the day will come when you measure up a coaxial I am interested in: KEF Q150.

I am in an impass with the KEF Q100. I do not know whether to modify the filter, adding more slope to the woofer, with quality components or leaving it as it is, improving the components that maybe it needs to improve. Or a new coaxial as Q150 or the old R100 (second hand?).

 

[hardisj]

Hopefully the day will come when you measure up a coaxial I am interested in: KEF Q150.

I am in an impass with the KEF Q100. I do not know whether to modify the filter, adding more slope to the woofer, with quality components or leaving it as it is, improving the components that maybe it needs to improve. Or a new coaxial as Q150 or the old R100 (second hand?).

If you have one, send it to me.

Though, I honestly doubt it has changed enough from the Q100 to warrant it. I haven't paid attention to Kef drivers since 2013, though.

 

[maty]

We have the Atlantic Ocean between us. If I lived in the USA, I would wait for the occasion when the sale price falls to $ 300.

With the $ 300 dollars saved (they now cost $ 600): https://www.amazon.com/KEF-Q150-Bookshelf-Speakers-White/dp/B071P6KQZX?th=1

I would buy them and put your shipping address for you to measure them in detail to know if the crossover is enough or could be clearly improved. In the second case, after enjoying it for a season, you would send him to Danny Richie to try to improve it to the maximum with ALL audiophile? improvements. Then they would come back to you for a new measurement and compare. And finally to my home. But before a new YouTube video as Q100.

YouTube Lets fix the Kef Q100! by Danny Richie

You know that I am very grateful for your measurement of the KEF Q100 driver.

• https://www.erinsaudiocorner.com/driveunits/kef-q100-drive-unit/
• https://erinsaudiocorner.com/driveunits/kef-r300-drive-unit/
• https://erinsaudiocorner.com/driveunits/kef-ls50-drive-unit/

 

[HammerSandwich]

Doing the math...
80dB @ 3 meters is = 86dB @ 2 meters and 92dB @ 1 meter.
90dB @ 3 meters is = 96dB @ 2 meters and 102dB @ 1 meter.

Thus, my desire to test at output at 96dB to 102dB. In my experience, that's adequate to find faults that should cover the overwhelming majority of listeners ... and then some.

Are these average or peak? (80-90 hints at average to me.) How relevant is dynamic testing without measuring peak levels?

The THX standard of 105dB peaks at listening distance seems like a reasonable goal, with an escape clause for speakers that show distress at lower levels.

 

[HammerSandwich]

I'm OK with that. Shoot me a PM and we can talk about it.

Get it to @amirm first. Or did I miss that you both measured the same speaker already? Not same model - same exact unit.

 

[hardisj]

Are these average or peak? (80-90 hints at average to me.) How relevant is dynamic testing without measuring peak levels?

The THX standard of 105dB peaks at listening distance seems like a reasonable goal, with an escape clause for speakers that show distress at lower levels.

Average listening levels. Peak would depend on the media.

My 102dB is 3 dB shy of that goal. If I feel like pushing it a bit more with other speakers then I will. But my baseline goal is 102dB. Anything more will simply be that: more.

 

[HammerSandwich]

My 102dB is 3 dB shy of that goal.

This omits the distance factor.

 

[hardisj]

This omits the distance factor.

I see. 105dB at typical listening distance. And I am assuming, by your persistence, this listening distance is 2-3 meters? If so, then, you're correct; I am not achieving 105dB at that listening distance. I stood nearby for the testing of this particular speaker when I it was getting in to the upper 90's and literally winced at times because I thought "okay, this has got to be cutting it close". It survived. I just don't have the... you know... to push it further without a back-up plan.

I don't know what else I can say to get you to understand why I don't fancy doing such extremely high output tests for a speaker of this nature. Honestly, if you can cover the cost of speakers I test (should I damage one and need to replace it) I will absolutely be happy to do significantly higher output tests. This is not at all a snarky response either. I am legitimately serious. I personally don't have the income that I can risk, though. If you do and you're willing to cover the cost then I'll talk to Mads about it and get his blessing (since these are his demo speakers) and we can go from there. I am literally on the finishing stages of the review now so let me know.

 

[hardisj]

Get it to @amirm first. Or did I miss that you both measured the same speaker already? Not same model - same exact unit.

Nope. I don't know if this is something Amir has tested or not. You can see my request for speakers in a separate thread I made here. https://www.audiosciencereview.com/forum/index.php?threads/looking-for-speakers-for-test.13765/

At any rate, it's ultimately up to @richard12511 what happens to it because he offered it up. I am willing to test it and run it through the full gamut. But I don't necessarily want to bother if it's something Amir has already tested for reasons discussed in the linked thread.