Perlisten speakers

[Dennis Murphy]

That's a gated measurement with pretty low resolution. Would be hard comparing that to our NFS measurements.

James is able to get an 8 ms window measuring outside on a tower. For some reason, he smooths the Spin graphs to 12 dB/octave, whereas his standard off-axis waterfall plots are smoothed at 24 dB/octave. The latter are likely to pick up any resonance that amounts to anything from an audibility standpoint, and the main advantage of the Klippel plots is below 250 Hz, where James is forced to rely on ground plane measurements. That's not to say that greater resolution isn't desirable, but the Audioholics plots probably capture the essence of the speaker.

 

[amirm]

The posted plot starts at 250 Hz and hence my assumption that it is a simple gated measurement. Just looked at the review and he says:

"The Perlisten S4b speakers were measured in free-air at a height of 7.5 feet at a 1-meter distance from the microphone, and the measurements were gated at an 11-millisecond delay. In this time window, some resolution is lost below 250 Hz and accuracy is completely lost below 110 Hz. Measurements have been smoothed at a 1/24 octave resolution."

So definitely low res. I am also not sure of the 1/24th octave resolution in higher frequencies. It is just too smooth there.

 

[Ata]

It turns out it's the best bookshelf speaker in the world objectively. and would give 99% of floorstanders a run for their money. Thanks @hardisj for another amazing review: https://www.erinsaudiocorner.com/loudspeakers/perlisten_s4b/

Very curious what would 4x the money buy as compared to the current passive bookshelf champion (KEF R3) in terms of preference scoring!

 

[Matthew J Poes]

The posted plot starts at 250 Hz and hence my assumption that it is a simple gated measurement. Just looked at the review and he says:

"The Perlisten S4b speakers were measured in free-air at a height of 7.5 feet at a 1-meter distance from the microphone, and the measurements were gated at an 11-millisecond delay. In this time window, some resolution is lost below 250 Hz and accuracy is completely lost below 110 Hz. Measurements have been smoothed at a 1/24 octave resolution."

So definitely low res. I am also not sure of the 1/24th octave resolution in higher frequencies. It is just too smooth there.

It is a simple gated measurement but the resolution loss is primarily below 250hz.

I’ve already seen true anechoic and even NSF measurments of this exact speaker. There is virtually no meaningful difference above 300hz.

 

[Matthew J Poes]

Very curious what would 4x the money buy as compared to the current passive bookshelf champion (KEF R3) in terms of preference scoring!

The biggest difference I know you would see is a massive different in maximum output. That wouldn’t show up in the preference score since the output isn’t a part of that.

The finish quality is better as well. This is sold to the luxury product market and it’s finish is closest to something like a Revel Ultima Salon type of speaker.

Best I can tell it seems the Perlisten would measure better in most ways but probably less bass. They are a sealed compact monitor so they don’t produce a lot of bass. They are intended to be used with a subwoofer.

 

[Matthew J Poes]

I'm guessing this sort of vertical directivity finese is only possible in a 3-way where the mid-range and the tweeter are in MTM configuration?

The directivity above ~1khz is highly controlled in all Perlisten speakers as a result of the waveguide beam forming design. It’s evident in the measurements.

 

[Ata]

The biggest difference I know you would see is a massive different in maximum output. That wouldn’t show up in the preference score since the output isn’t a part of that.

The finish quality is better as well. This is sold to the luxury product market and it’s finish is closest to something like a Revel Ultima Salon type of speaker.

Best I can tell it seems the Perlisten would measure better in most ways but probably less bass. They are a sealed compact monitor so they don’t produce a lot of bass. They are intended to be used with a subwoofer.

Regarding finish, yes it would not be hard to best the R3 options, though I personally find the all black finish very attractive.

SPL output of R3 is 110db based on specs vs 111db for the S4b. If this is not an apples to apples comparion, how would S4b achieve much better real SPL physically? Its woofer is barely larger and mid-woofers are tiny area wise...

Based on the spinoramas my untrained eye would expect slightly better directivity and flatter FR.

 

[Matthew J Poes]

Regarding finish, yes it would not be hard to best the R3 options, though I personally find the all black finish very attractive.

SPL output of R3 is 110db based on specs vs 111db for the S4b. If this is not an apples to apples comparion, how would S4b achieve much better real SPL physically? Its woofer is barely larger and mid-woofers are tiny area wise...

Based on the spinoramas my untrained eye would expect slightly better directivity and flatter FR.

The simple SPL answer is that the 110dB spec is wrong. Look at the measurements obtained by Erin.

That is the larger R5 tower. Look at the compression happening at 102dB. It's already showing 1 dB of compression by 102dB at numerous frequencies.

By comparison, the Perlisten is only at -.5 or -.6dB in that same range. I asked Erin if he was going to push it to the standard -3dB of compression and he was too worried about possible damage and being responsible.

Kef and Perlisten don't spec SPL the same way as there isn't a great standard. While CEA has one, THX uses a totally different one and numerous designers have told me they aren't huge fans of the CEA spec for maximum spl.

The long term compression test Erin did is also an interesting comparison showing a bit of what is going on:

Note that all the long term thermal compression is happening in the tweeter range. That isn't ideal. It isn't fixable.

This shows no thermal compression in the tweeter, it is all in the woofer. Add an 80hz highpass, as intended by design, and a lot of that compression is addressed. Again, this is not comparing an apples to apples speaker either, this is a tower speaker vs a bookshelf, and the bookshelf is nearly or absolutely besting it in output.

 

[Matthew J Poes]

Regarding finish, yes it would not be hard to best the R3 options, though I personally find the all black finish very attractive.

SPL output of R3 is 110db based on specs vs 111db for the S4b. If this is not an apples to apples comparion, how would S4b achieve much better real SPL physically? Its woofer is barely larger and mid-woofers are tiny area wise...

Based on the spinoramas my untrained eye would expect slightly better directivity and flatter FR.

Probably the estimated in room response is the best comparison

vs

 

[Matthew J Poes]

oops, just realized Erin did the R3 too.

Ok that looks very similar to the perlisten.

but he didn't test it's spl unfortunately.

I think the Kef has better vertical directivity being a coaxial design. The Perlisten and Kef look otherwise very similar, but the fact that the bookshelf bested the R5 in SPL I would guess that the S4B would have that as its primary benefit.

 

[blueone]

That said, there is actual content down there in some modern music. It’s not super rare anymore, and turning on a high pass 30Hz filter via Roon shows that those ULF really can enhance the experience.

That's what I'm measuring with a spectrum analyzer. Fusion jazz is the best example I can think of outside of cathedral organ music, but even a concert grand piano in a small to medium size venue is producing output in the 25-30Hz range. For example, the solo piano arrangement of Pictures at an Exhibition, like this one (my favorite, actually):

Mussorgsky, Liszt, Wagner, Barry Douglas - Pictures at An Exhibit - Amazon.com Music

Mussorgsky, Liszt, Wagner, Barry Douglas - Pictures at An Exhibit - Amazon.com Music

www.amazon.com

has substantial deep bass response.

[Edit: I should probably clarify what I mean by "substantial". In this context I meant that if you have a roll-off below about 30Hz you'll notice it in playing back the recording. I did not mean high level deep bass like you get from a modern action movie.]

 

[ctrl]

I’ve already seen true anechoic and even NSF measurments of this exact speaker. There is virtually no meaningful difference above 300hz.

I'm a little ambivalent about it. We can compare the measurements (above 250Hz) of Audioholics (AH), ErinsAudioCorner (EAC) and the manufacturer (MA).

If we compare the on-axis measurements, the differences between AH and EAC up to 9kHz are not very big. Above 10kHz it depends very much on the quality of the individual mic calibration file, because in this frequency range the microphone capsules usually show the biggest deviations.

The manufacturer measurement uses heavy smoothing and might based on a golden sample.
(Note, the scaling is only 30dB)

When comparing the listening window (LW), the differences are also not very large between EAC and AH, but diverging now, however, already above 7kHz.
However, if you follow the individual progressions of AH/EAC and MA, you could come to different tonal conclusions.
The "regression line" might already turn out differently because of different slopes.
(Note, the scaling is only 30dB)

When comparing the sound power (SP) frequency responses, you can hardly believe that it is the same speaker. SP is an "averaged" frequency response over all angles and one would expect that smaller deviations would be smoothed out by the "averaging".

Between 500Hz and 1kHz all measurements show differently pronounced humps, between 1kHz and 2kHz all measurements are unanimous and between 2-5kHz AH and MA are similar, EAC falls a bit out of line.
Above 5kHz the SP curve seems to be completely arbitrary.

To better see the trends, I have in the second diagram offset the SP frequency responses, but plotted them at standard 50dB scaling.

Based on these SP frequency responses, I would attribute different tonal characteristics to the speaker in each case. For example, 1dB deviation in the SP frequency response, over an octave, has clearly audible effects on a loudspeaker.
From that point of view, these could also be speakers from three different manufacturers.

It is, of course, a matter of defining what "meaningful differences" are. But it is not the case that the frequency response curves in the CEA-2034 visualizations of EAC, AC and MA would lead to completely identical interpretations of the measurements.

 

[Colonel7]

Since we're talking surrounds and 80 Hz design with subwoofers in mind Not in same class for SPL and distortion (and directivity) but food for thought and yes I do realize no one shopping for Perlisten is looking at low budget and DIY

 

[Dennis Murphy]

The posted plot starts at 250 Hz and hence my assumption that it is a simple gated measurement. Just looked at the review and he says:

"The Perlisten S4b speakers were measured in free-air at a height of 7.5 feet at a 1-meter distance from the microphone, and the measurements were gated at an 11-millisecond delay. In this time window, some resolution is lost below 250 Hz and accuracy is completely lost below 110 Hz. Measurements have been smoothed at a 1/24 octave resolution."

So definitely low res. I am also not sure of the 1/24th octave resolution in higher frequencies. It is just too smooth there.

I believe that's what I said, only I underestimated the resolution because I was relying on his results for previous speakers. 11 ms is huge for this kind of measurement (I can only get about 6 ms), but I agree that the plots look too smooth for 24 dB smoothing. I'll ask him about that.

 

[Spocko]

I just updated my review. I added a couple photos. I added a little bit more in the wrapup section. I also updated the nearfield measurements (since I took the speaker apart, I disconnected the mids and then tweeter to get a proper singular measurement of the two) and I also measured the impedance of each driver in the waveguide array.

Most importantly, did you improve the speakers for the owner by adding exotic caps and replacing copper wire with hand drawn silver under a full moon?

 

[Spocko]

The biggest difference I know you would see is a massive different in maximum output. That wouldn’t show up in the preference score since the output isn’t a part of that.

The finish quality is better as well. This is sold to the luxury product market and it’s finish is closest to something like a Revel Ultima Salon type of speaker.

Best I can tell it seems the Perlisten would measure better in most ways but probably less bass. They are a sealed compact monitor so they don’t produce a lot of bass. They are intended to be used with a subwoofer.

Matt, the SPL limits suggest that these bookshelves has output that is definitely Dominus worthy when paired with a subwoofer from 100Hz down - the perfect match for my dual 18" JTR!

 

[Matthew J Poes]

Matt, the SPL limits suggest that these bookshelves has output that is definitely Dominus worthy when paired with a subwoofer from 100Hz down - the perfect match for my dual 18" JTR!

Sure, it is quite capable, but it doesn't meet the Dominus spec. I just want to be sure people are aware of that fact. It can be a little confusing. The Dominus spec isn't just an SPL spec.

But this speaker absolutely seems to be unique in offering both such good measured performance and such high output capability, giving it a good ability to operate in larger rooms as an LCR.

I wanted Erin to actually test the limit of output on these (but admittedly I might have been the one then paying for any damage). From what Dan has told me, these are capable of more output than the official rating. That rating is based on a specific standardized specification that they used across the line. That both the 2nd and 3rd harmonic are less than 3%. That is a pretty high bar. The 3dB compression test is likely a little beyond that. That is quite loud for a modest sized bookshelf.

 

[Godataloss]

If port noise is reduced by means of some foam, that's an impressive piece of engineering, avoiding the expense of a passive radiator.

Goodman's did it in the 60's.

 

[Matthew J Poes]

Goodman's did it in the 60's.

I can't seem to find where that comment is coming from, but the foam included with the Perlisten S7T and other ported models is to block the port. It would stop port noise because the port wouldn't work anymore. It's intended to allow the speaker to be a sealed design for better integration in a bass managed system with subwoofers. The S7T is somewhat unique (I don't mean to imply that it is the first or only speaker like this, just not a common approach) because when ported it acts more like a QB3 alignment. That is, its output doesn't roll off at 4th order right away like most ported designs but gently slopes at more of a 2nd to 3rd order design before eventually transitioning to the steeper 4th order. In room this means you don't get as much bass below 40hz as you might with a more traditional 4th order alignment, but you do get more extension in room and it integrates with subwoofers better. It also provides some excursion control, so with a really capable processor like the Trinnov or Storm, you can highpass the S7T at say 25hz, low pass the subwoofers at 80hz, and allow them to substantially overlap to good effect. While this can be done with a traditional alignment, you usually would need to either highly reshape the response of both and/or turn the sub down more than is desirable. I happen to like this approach myself, but very few commercial speakers offer it, since most people like their bass.

However, as to the actual point, I had done some experiments with reticulated foam in the ports of subwoofers to see if it could reduce port noise. I looked at two things, a reduction in port resonances, a reduction in port chuffing. I got the idea from Earl Geddes and he directed me on how to conduct the tests. I did all the tests with a low noise calibrated measurement microphone and nearfield measurement of the port itself. Ultimately, I didn't find it worked very well for either. Port resonances were only marginally reduced, not enough to matter in many cases. Good port design in the first place turned out to be more important. It also didn't help port chuffing. In fact, best I could tell from this crude method of measuring chuffing, it made it worse.

What it did do, which I did like, was lower the Q of the tuning. This allowed a small extension of the LF response at the expense of a bit of output around tuning. similar to the QB3 alignment mentioned above. I found that if I placed a plug of 30ppi reticulated foam in the port that covered 2/3 or more of the length, I could closely mimic the Qb3 alignment. However, as noted, it seemed to make port chuffing issues worse.

I ultimately gave up. I really hate ports to be honest. For all their good, they are very hard to get right. A lot more bad than good often. As Amir has shown many times here, ports often have a resonance that shows up in the on-axis response. In the best speakers its benign, but in a lot of average to cheap or poorly designed speaker, its a major response flaw. Chuffing can be a huge issue. I've probably built a dozen DIY subwoofers at this point and half of those were with the same driver, just different enclosure designs. I found I could either get no noticeable chuffing within the limits of the driver but a huge audible honk from the port resonance, or get rid of the resonance but have noticeable chuffing. Even driver suspension noise seems to be more audible through the port in these designs. So the best design was a balance of the two. Optimizing the port with a very sophisticated flare is good, but no such flared ports exist for DIY. I made mine using the heat gun and PVC pipe method, which was OK at best. My current DIY subwoofer has a slightly flared slot port and it chuffs. At the end of the day, I much prefer a sealed design with enough displacement to meet my needs. No port problems.

 

[hardisj]

Most importantly, did you improve the speakers for the owner by adding exotic caps and replacing copper wire with hand drawn silver under a full moon?

Of course! I'm currently welding in 0 gauge wire to the crossovers.

I wanted Erin to actually test the limit of output on these (but admittedly I might have been the one then paying for any damage).

Yep. 'tis true. I was thiiiiiis close to testing it but with supply chain issues likely causing significant delays in repair/replace (nevermind the fact we'd have to get it cleared with the owner), we (myself and Matt) weren't willing to risk it.