Which speaker would you prefer?

[<eargiant]

Some more information. Both speakers are 3-way designs and rear ported. Tweeter and mid spacing is virtually identical between the two.

I've had the speakers measured in plot A & B in here and they both sounded the same to me (same positions, same rooms, etc.).

This has me scratching my head. Is it possible that one of the graphs is misleading or incorrect? Which one is closer to the truth?

 

[ski2xblack]

I doubt either is misleading or incorrect, simply the results they got under the conditions (slightly different, as noted) they measured them in, and very limited in scope.

You still seem to be failing to realize that those single on axis measurements don't provide any sort of comprehensive basis to judge either speaker by, just a little slice of it. That was Lovin's point back at the beginning of the thread. To get even remotely "closer to the truth" as you put it would require a whole slew of additional data.

 

[<eargiant]

Lot's of informative stuff has been discussed here regarding measurement techniques. Thank you all for you contributions.

Time for the Easter Egg. There were a couple of times I thought Beave found it during the "hunt". Maybe with this additional information someone can shed light on why there is such a big difference between the high frequency plots on the two graphs.

Happy Easter & Happy Passover!

Speaker A

Speaker B



Available Measurements:

Speaker A
https://www.soundstagenetwork.com/index.php?option=com_content&view=article&id=839:nrc-measurements-polk-audio-lsim703-loudspeakers&catid=77:loudspeaker-measurements&Itemid=18

Speaker B
https://www.stereophile.com/content/polk-lsiiim703-loudspeaker-measurements

 

[AcuDefTechGuy]

In reality I think they are probably equal from below 9kHz, which is the salient point.

Ha! Now that we know they are the same speaker (of course, we’ve seen different measurements of the same speakers before), I think I should get extra credits for saying they are probably equal from below 9kHz.

I’ve recall different measurements of the NHT SuperZero that were quite different. In one measurement, the listening window was +/- 0.9dB, which is by far the best I’ve ever seen. In another measurement, it was +/-3dB.

So that’s why we all know not to put all hopes into one measurement.

I prefer to see on-axis within +/-3dB, smooth off-axis, sensitivity above 90dB/w/m, and impedance > 3 ohms.

 

[Beave]

Yes, I knew they were the same speaker as soon as I saw the thread.

I've spent WAY too many nights combing through the Soundstage measurements and Stereophile measurements. I recognized the speaker immediately.

 

[Beave]

I already explained the difference in the high highs (that tweeter is highly directional in the highest highs, and the two measurements are on a slightly different axis).

If you look at the NRC's on-axis plot (that also includes 15 & 30 degrees off axis), you can see that they too show a bump in the top octave on their on-axis plot, similar to, but a little smaller than, JA's measurement. It quickly drops off, by 15 degrees. Their plots of 45, 60, and 75 degrees show just how much that tweeter drops off in the top octave.

Stereophile's figure 6 also shows just how directional the top octave is in the vertical plane. It changes pretty substantially every 10 degrees or so.

 

[<eargiant]

I already explained the difference in the high highs (that tweeter is highly directional in the highest highs, and the two measurements are on a slightly different axis).

If you look at the NRC's on-axis plot (that also includes 15 & 30 degrees off axis), you can see that they too show a bump in the top octave on their on-axis plot, similar to, but a little smaller than, JA's measurement. It quickly drops off, by 15 degrees. Their plots of 45, 60, and 75 degrees show just how much that tweeter drops off in the top octave.

Stereophile's figure 6 also shows just how directional the top octave is in the vertical plane. It changes pretty substantially every 10 degrees or so.

Good stuff-thank you

I hope that enthusiasts here and elsewhere will remain objective and look beyond the limited and sometimes solitary (only performed by one entity) speaker measurements that are available to us and remember that there's always more than meets the eye (or ear ) when it comes to speaker measurements. A little height adjustment, toe-in or toe-out (and slew of other variables) will change that measurement dramatically.

Think twice before you call out a gross "imperfection" in a loudspeakers measured response. That was the whole point of this thread.

 

[Dennis Murphy]

Ha! Now that we know they are the same speaker (of course, we’ve seen different measurements of the same speakers before), I think I should get extra credits for saying they are probably equal from below 9kHz.

I

Well, we've actually gone around on this point in a different thread. The plots below 9 kHz really can't be reconciled. Other things equal, the NRC plot should show less of a bump in the bass than the Stereophile plot, because JA splices a near-field plot onto the quasi-anechoic, which will always result in a bass hump if the the bass response is actually flat. But we get the opposite result here. If the NRC chamber shows a bass bump, then JA should get an even bigger one with the splice. As for the difference in the high frequency response, of course NRC's Listening Window plot will be different than the Stereophile on-axis plot. Showing those two plots just wasted a lot of Internet ink.

 

[TLS Guy]

I have come to this late. However as I have often said you need much more than FR. If the FR impersonates the Rocky Mountains, strike it off the list.

However the rub is that is an easy matter to design and make a speaker that has a text book FR that is unlistenable to the point where you could not even understand human speech.

I think all of us who have done this for a long time develop our preferred techniques. This relates to bass loading, transient response, where to use constant voltage versus constant power at a crossover point, and host host of other issues and especially bandwidth power response.

I have always taken the late John Wright's advice to heart. For a truly finished speaker you must listen to it over a prolonged period of time. Once the speaker you feel is good, then you make small changes no sooner that every three months. Be patient enough to allow a couple of years at least to pass before you consider a speaker finished.

A speaker designer must really know what natural instruments sound like.

You can not design a good speaker if you only know amplified music of the popular rock pop variety.

You really need to have a wide knowledge of solo, especially and including human voice, of chamber music, the symphonic and choral repertoir. I firmly believe you can not design a great speaker unless you do.

I think you will be very lucky to choose a speaker that you really like from a set of measurements. I think you can know ones you won't like, but not ones you like. You have to listen.

These days that is very difficult, if not next to impossible. That is true to such an extent now, that it just about compels those who want really good sound in the home to become expert speaker designers and builders.

 

[TechHDS]

Can't comment without seeing the scales, also, let's at least include the distortions vs frequency graphs, if not the impedance and phase ones.

After your post pretty much every other one, is mute without the other graphs. I have upper low end speakers, sound decent at modern volume levels. Crank them up get them loud they distort. But I knew that about Polk M70s, was what I could afford at the time.

 

[PENG]

Good stuff-thank you

I hope that enthusiasts here and elsewhere will remain objective and look beyond the limited and sometimes solitary (only performed by one entity) speaker measurements that are available to us and remember that there's always more than meets the eye (or ear ) when it comes to speaker measurements. A little height adjustment, toe-in or toe-out (and slew of other variables) will change that measurement dramatically.

Think twice before you call out a gross "imperfection" in a loudspeakers measured response. That was the whole point of this thread.

This is a great thread! You did a good job in making your point. I also hope that you now realize we are actually quite an objective group on this forum, as you can see that no one has made any definitive conclusions on those "two" speakers. Most comments cited things like:

- more information needed, e.g. smoothing/resolutions of each plot off axis/listening window, distortions, impedance, phases CSD.
- probably sound similar.
- both look pretty good.
- "assuming everything else equal..."
- the elevated above 9 kHz response might be due to the mic positions being different....
- difference due to anechoic and quasi-anchoic....

So overall, no one was willing to pick or jump to conclusion, without more info or making a lot of assumptions/caveats, partly because both look really close and good, except that elevated band. As an example, if the LF response look very different, we wouldn't worry about the marking on the scale, or if there are clear sign of resonances in one but not the other, we (at least I) would not have raise the point about difference in smoothing and resolutions of the graphs.

If anyone would have jumped to conclusion based on just one FR plot (of each speaker), one would have been called out by any of us, that's just my guess of course. If we have not done so in the past, we should, from now on.

 

[Dennis Murphy]

Well, we've actually gone around on this point in a different thread. The plots below 9 kHz really can't be reconciled. Other things equal, the NRC plot should show less of a bump in the bass than the Stereophile plot, because JA splices a near-field plot onto the quasi-anechoic, which will always result in a bass hump if the the bass response is actually flat. But we get the opposite result here. If the NRC chamber shows a bass bump, then JA should get an even bigger one with the splice. As for the difference in the high frequency response, of course NRC's Listening Window plot will be different than the Stereophile on-axis plot. Showing those two plots just wasted a lot of Internet ink.

When I actually took a look at the Stereophile article, I realized that their plot and the NRC plot are more comparable than I indicated above. The Stereophile plot isn't just on-axis--it's the average of an on-axis and 15 degrees off axis on either side. The NRC plot is that plus 15 degrees off above and below axis. So it averages 2 additional off-axis plots, and that will help account for the lack of a rise in the top octave, although probably not entirely. I also reexamined the measurements NRC did for my BMR monitor. They showed a bump in the bass that I know isn't there. Interestingly, the BMR is also a 3-way monitor with a rear port, and the explanation is probably the same--the delay in the arrival time of the port information causes reinforcement and cancellation effects with the woofer output that don't occur in a normal environment. The bottom line is that both the NRC and Stereophile measurements can be confusing once you get to the bass region.

 

[Irvrobinson]

A speaker designer must really know what natural instruments sound like.

You can not design a good speaker if you only know amplified music of the popular rock pop variety.

You really need to have a wide knowledge of solo, especially and including human voice, of chamber music, the symphonic and choral repertoir. I firmly believe you can not design a great speaker unless you do.

I agree, and I'd take it a step further, really knowing whether a speaker is accurate takes recordings you made yourself. You need to know what the live sound should be. I've been an advocate of this for a long time now. In your case, you have a large library of recordings you made, but for the rest of us all it really takes is a handheld digital recorder that costs a few hundred dollars. And you don't need to record Yo Yo Ma either. Just tapping on a cymbal, playing a scale on a piano, recording anyone playing a violin live is good... an upright bass can be very revealing. A flute reveals a lot about the lower range of tweeters. A piano seems to reveal everything. One of the most identifiable sounds to me as real or not is a wooden drumstick on a dark ride cymbal.

If you want to see how inadequate most audio systems really are, record a drum kit in a residential room. Most systems won't play loud enough with low enough distortion.

Non-musical sounds are also surprisingly useful. Whack a large, heavy cardboard box with a wooden spoon. It makes a surprisingly unique sound.

I think humans have an instinct for what sounds live and what is obviously (and wrongly) reproduced when they know what the real thing sounds like.

 

[Art Vandelay]

Good stuff-thank you

I hope that enthusiasts here and elsewhere will remain objective and look beyond the limited and sometimes solitary (only performed by one entity) speaker measurements that are available to us and remember that there's always more than meets the eye (or ear ) when it comes to speaker measurements. A little height adjustment, toe-in or toe-out (and slew of other variables) will change that measurement dramatically.

Unfortunately though, brand bias will always prevail, and where possible many people will seek to use or misuse published measurement data as evidence to cement their position or sustain a particular narrative.

It's just human nature.