Revel Salon2 vs Genelec 8351B - Blind Test Preparations

[Neale]

I am going to be doing a controlled blind test of these speakers soon. Opinions on best test procedure are welcome ..

Have you told us somewhere what amp is driving the Salon 2's?

 

[ctrl]

I’m open for donations of such apparatus (Complete with the necessary acoustically transparent but visually opaque curtains, electromechanical automation, controls, and end-to-end installation work, etc. necessary for it to be practical.)
Otherwise I will have to find the least bad compromise.

A stable turntable need not be expensive. You only need some wood, many steel balls (and balls of steel when your life partner sees the constructions for the first time) and the necessary space to store the two turntables.

Diameter of 1-1.2m with an extension for very deep speaker cabinets should be sufficient to place two speakers back-to-back on a turntable.
To measure a loudspeaker on a turntable with a diameter of 0.6m and an extension so that even deep speakers can be set up flush with the axis of rotation, it looks like this:

A manual can be found here:
http://www.uibel.net/bauvor/bv_tutorials/dt/dt.html

On my turntable the milling for the steel balls is placed further outside and to support the extension of the turntable there are ball rollers under the extension.

All you have to do is drown out the spinning noises during the ABX test

 

[Chrise36]

It is not so pricey if you can do a bit of diy...You can take some ideas from this blog http://croweaudio.blogspot.com/2019/03/beyma-tpl-150h.html?m=1

 

[echopraxia]

A stable turntable need not be expensive. You only need some wood, many steel balls (and balls of steel when your life partner sees the constructions for the first time) and the necessary space to store the two turntables.

Diameter of 1-1.2m with an extension for very deep speaker cabinets should be sufficient to place two speakers back-to-back on a turntable.
To measure a loudspeaker on a turntable with a diameter of 0.6m and an extension so that even deep speakers can be set up flush with the axis of rotation, it looks like this:

View attachment 79282


A manual can be found here:
http://www.uibel.net/bauvor/bv_tutorials/dt/dt.html

On my turntable the milling for the steel balls is placed further outside and to support the extension of the turntable there are ball rollers under the extension.
View attachment 79285

All you have to do is drown out the spinning noises during the ABX test

I admire your setup there, but I’m also curious: how do you deal with switching back and forth rapidly when the listener wants to go over a part of the song repeatedly on each speaker? I’m not sure how Harman structures the test procedure, but in many cases this ability for participants to switch back and forth quickly seems to be helpful and desirable.

 

[KSTR]

Given the fixed constraints of your setup, I'm sceptical that true ABX protocol will give meaningful results. What I can contribute is that we tried to do something very similar to what you are doing, just that it were two pretty similar, sometimes even identical, small 2-ways or horizontal 3-ways stacked on each other, tweeters facing each other.
Since the task of ABX is to identify X as either A or B, it strongly depended on the timing of the test protocol for the success rate we could tell speaker A from B this way. If the switch-over is quasi gap-less (10...100ms breaks) it was always possible to identfy the correct speaker simple by that when switching over from, say, A to X it was not hard to tell if the sound changed or not and by that X can be identified.
With much larger gaps (up to a day) things got more difficult but still we could often associate the perceived location on the acoustical blindfold to each speaker and identify the unknown X, played as the first thing the next morning, actually the blindfold did not make much difference to full-sighted (but of course not knowing the selection). Perhaps it would have been better to physically blind the listeners...

From that, with these two quite different speaker and large listening distance (full reverberant field) I would guess that even when acoustically aligned you will be able to tell speakers apart alone from their different directivity and again even when the in-room response is somewhat EQ'd to bring them closer together.

Maybe a preference test would be a better choice for test protocol than ABX or AB? Or just taking notes what people think of different aspects of playback of A vs B, not knowing which is playing and with no switch-overs to avoid cues (or at least with a long pause).

Anyway, I do encourage to proceed with the test as planned because no matter how it ends still it increases experience and knowledge for everybody involved and it's fun, last not least.

 

[echopraxia]

Given the fixed constraints of your setup, I'm sceptical that true ABX protocol will give meaningful results. What I can contribute is that we tried to do something very similar to what you are doing, just that it were two pretty similar, sometimes even identical, small 2-ways or horizontal 3-ways stacked on each other, tweeters facing each other.
Since the task of ABX is to identify X as either a or B, it strongly depended on the timing of the test protocol for the success rate we could tell speaker A from B this way. If the switch-over is quasi gap-less (10...100ms breaks) it was always possible to identfy the correct speaker simple by that when switching over from, say, A to X it was not hard to tell if the sound changed or not and by that X can be identified.
With much larger gaps (up to a day) things got more difficult but still we could often associate the perceived location on the acoustical blindfold to each speaker and identify the unknown X, played as the first thing the next morning, actually the blindfold did not make much difference to full-sighted (but of course not knowing the selection). Perhaps it would have been better to physically blind the listeners...

From that, with these two quite different speaker and large listening distance (full reverberant field) I would guess that even when acoustically aligned you will be able to tell speakers apart alone from their different directivity and again even when the in-room response is somewhat EQ'd to bring them closer together.

Maybe a preference test would be a better choice for test protocol than ABX or AB? Or just taking notes what people thing of different aspects of playback of A vs B, not knowing which is playing and with no switch-overs to avoid cues (or at least with a large pause).

Anyway, I do encourage to proceed with the test as planned because no matter how it ends still it increases experience and knowledge for everybody involved and it's fun, last not least.

Yeah, I think that’s fair. Overall it seems we are all converging pretty much exactly on the procedure I’ve used in my last two blind tests (ironically perhaps, since all the initially proposed changes here have been from feedback from those threads): speakers spaced equally apart (AB ——— AB), and just take notes about preferences and record a preference vote for each track.

Main difference this time around is I will be measuring and ensuring bass response is nearly identical, as rigorously as I can (whether I accomplish this with subs or not, I will see what I can do). And this time, in addition to each person choosing their own tracks, I plan to at least at minimum include a standardized set of songs, like from here: https://www.audiosciencereview.com/...sic-tracks-for-speaker-and-room-eq-testing.6/

 

[Balle Clorin]

The difference in elevation will make a difference, take a look at the spinoramas above and below axis

 

[KSTR]

The AB ---- AB arrangement, near-/midfield in a wide room, is what we mainly ended up for jugdements and preference, while we used ABX/AB when the same fixed raw speaker cabinets could be used, testing different amplifiers or crossovers. "We" refers to the staff of a german manufacturer of active studio monitors, 10 years ago, I will add.

 

[echopraxia]

The difference in elevation will make a difference, take a look at the spinoramas above and below axis

The only speaker that would hypothetically be operating at a nonstandard height is the Genelecs, whose vertical off axis response is excellent. It is also designed as per Genelec manual to be elevated up to 15 degrees and pointed downward, if necessary. This would not be a problem in terms of spinirama.

What is a concern is that the mere physical presence of the Genelec could change the way sound propagates out vertically from the Salon2’s, if stacked. And of course, the height of the sound source will be different, but this is much less perceptible than the horizontal placement changing — and in this case, we just compromise and choose one or the other here.

However, the concerns that the mere presence of the Genelecs stacked on top could negatively influence the vertical response of the Salon2’s are valid enough that I hesitate to do the stacking method. I need to measure the Salon2’s with and without the Genelec on top,
to see if there is any measurable difference.

So I suppose my test procedure is converging back to exactly what I did before, if I don’t stack. But I am still very tempted to test stacked. Audibly to me, stacking just sounds way more similar than if I don’t stack them, due to the left/right shifting issue.

I already know no matter what I do, plenty of people will feel compelled to post and explain why I did everything wrong. And if I do it they way they would suggest, another group of people would explain why I should have done it the other way instead.

So I fully accept that not everyone will be satisfied by the procedure. Just am hoping here to get thoughts and converge to the best compromises.

 

[echopraxia]

In a very pragmatic sense, I still gravitate towards stacking, vs separate locations in the horizontal plane — because stacking subjectively sounds the most similar to me when I switch back and forth. So, I am very tempted to use it because it simplifies so much, and also does not seem to cause any audible issues to my ear. I will do more measurements and listening tests soon, though.

If it turns out that stacking has no measurable effect on the Salon2’s in-room response, I will want to understand more why it is not a better compromise than each speaker being placed in a different location (several feet apart) in the horizontal plane.

 

[Juhazi]

Hi, long distance in large room means you will be listening to the reverberant field (room) instead of the dominance of direct sound of DUTs. Then eq'ing this room response makes comparison pointless.

But that's just me, do what you like!

 

[thewas]

The AB ---- AB arrangement, near-/midfield in a wide room, is what we mainly ended up for jugdements and preference, while we used ABX/AB when the same fixed raw speaker cabinets could be used, testing different amplifiers or crossovers. "We" refers to the staff of a german manufacturer of active studio monitors, 10 years ago, I will add.

Would also recommend above arrangement and due to

Hi, long distance in large room means you will be listening to the reverberant field (room) instead of the dominance of direct sound of DUTs. Then eq'ing this room response makes comparison pointless.

only EQing in the modal region (and mainly "cutting" only peaks due to modes if placed far enough from boundaries).
If you want to compare also with EQ above the modal region I would recommend doing it on the anechoic LW but there things get tricky at loudspeakers with not perfectly smooth directivity where at higher listening distances also partially considering the sound power can be advantageous, but this is leaving too many open variables, so I personally would restrain only to bass correction.

 

[Blumlein 88]

There's no point in doing this in stereo. Stick with one speaker.

I believe hearing a difference in stacked mode will be trivial. Playing pink noise for moment will make it obvious without doing a blind test. So maybe a preference test is on order.

Now I believe the Harman shuffler required 4 seconds to switch. Perhaps incorporating a 4 second delay is enough that vertical stacking no longer becomes obvious whether you are listening to the top speaker or not. Yet it is within echoic memory. Of course Harman tested for preference. Still be good if you didn't know which speaker for preference.

 

[Blumlein 88]

While your doing this it might be nice to record some music. With the speakers stacked, I'd put the Umik 2 meters from the speakers. Record a few song snippets so we can listen at home. Any vertical shift won't be picked up that way, but the FR or tonality will be. Might be interesting.

 

[LTig]

I don't know the size of the effect, I'd do some measurements of the Genelec below 200 hz like you already have, and with the inputs to the Revel Shorted. Those cones can absorb, resonate and slightly alter the sound from what is in the room. Shorting inputs should fix that.

If you find a large difference, you'll need to short the Revel during testing. Hopefully testing response you find the difference is too small to matter.

As long as you keep the Revels connected to the power amp and the power amp is switched on its very low output impedance (you should not use a SET or tube amp) should be close enough to a short to show the same effect of damping resonances.

 

[q3cpma]

There's no point in doing this in stereo. Stick with one speaker.

That's one of the only things from Toole I find dumb, as the argument is completely wrong. The idea that mono shows more MOS difference -> correctness is not backed by anything. It's the same as making people wear ball and chain during a 100 m sprint because it makes the variance in times larger; it makes more evident who's the strongest, but not the fastest.

 

[Alexanderc]

Shouldn’t an audible difference in the Revel with a speaker stacked on top be measurable? If so, could you measure it both with and without the Genelec on top and see if there is any difference?

 

[sigbergaudio]

I have two questions: What is the objective of the test, and why are you applying EQ?

 

[Blumlein 88]

That's one of the only things from Toole I find dumb, as the argument is completely wrong. The idea that mono shows more MOS difference -> correctness is not backed by anything. It's the same as making people wear ball and chain during a 100 m sprint because it makes the variance in times larger; it makes more evident who's the strongest, but not the fastest.

Carrying a weight would reduce differences in a timed sprint. Toole and I disagree with you.

Care to elaborate on the thinking here. If stereo listening made the test more discriminating would it not also increase the spread of difference judged by listeners?

 

[waynel]

I would test in stereo, otherwise the Salon's wide dispersion will be dramatically advantageous.

The Salon’s wide dispersion IS advantageous! The reasoning for mono testing is covered in Toole’s research and documented in his book.