Focal Aria 906 Speaker Review

[daftcombo]

One of my two pairs is glossy white on stands.

 

[Ron Texas]

Don't worry about that: the "Dark Walnut" finish on 906s is nothing but vinyl!

View attachment 69068

But, there's also some "Prime Walnut" finish, which is claimed to be real wood:
View attachment 69064

That kind of puts it in the same class as the Elac DB-62 for looks. What concerns me is it doesn't clearly say dark walnut vinyl. That might be another reason why I like piano black.

 

[muad]

When measuring the harmonic distortion, at 8kHz you would see a spike for third order harmonic distortion (HD3, 8kHz*3) and at 12kHz increased values for HD2 (12kHz*2), caused by the 24kHz resonance of the break up tweeter.
The resonance at 24kHz can easily be 15dB above the average sound pressure level. Accordingly, the harmonic distortions HD2, HD3,... are "boosted" at their fundamental frequencies.

The harmonic distortions themselves would not be audible at 24kHz, but the possible intermodulation distortions are partially in the audible frequency range.

UPDATE:

As a small example, here is the measurement of the harmonic distortion of a loudspeaker at 90dB. The tweeter used shows an intensive resonance at 27kHz (Seas DXT tweeter).
At 9kHz increased values for HD3 and at 13.5kHz increased values for HD2.
Since the harmonic distortions of this tweeter are at a very low level, the increases in harmonic distortion due to the break up of the tweeter membrane remain within acceptable limits.
View attachment 68994

If there's no musical information up that high, would there still be a resultant hd2 and hd3 during playback?

 

[daftcombo]

Can this be used for any improvement using convolution?

 

[daftcombo]

Instead of trying to fill a dip at 3.000 Hz that we can't see on that estimated in-room graph, wouldn't it be better to try to fill with a low Q the broad dip between 1.000 Hz and 4.000 Hz on the basis of that curve?

 

[GelbeMusik]

Instead of trying to fill a dip at 3.000 Hz that we can't see on that estimated in-room graph, wouldn't it be better to try to fill with a low Q the broad dip between 1.000 Hz and 4.000 Hz on the basis of that curve?

Why would You?! It's posted: Estimated ...

Other room, other result, for sure. Do You know the room model the spinorama uses? Is Yours alike or different, ceiling height, width etc? It is about two dBs in total max. Despite all the clarity the spinorama brings to the topic, people shall be warned to take all the estimates and averages too literally.

 

[Count Arthur]

Does anyone else not like pale speaker cones, especially with a black dust cap in the centre - I feel like I'm being watched.

 

[RichB]

I find this perfectly normal from a marketing POV. Let's face it: the vast majority of speakers manufacturers mentions the "need" of burn-in time in their manuals or communication... At least, "20H" seems fairly reasonnable, when you can see "100H" for the Buchardt S400, and so on.

I just checked at Revel or Elac DBR manuals, there's no mention about that. But JBL HDI manual (Harman/Science-based company) claims strange things about bi-wiring (not only bi-amping!):
View attachment 69013

Sounds like the same kind of marketing BS to me. Hard to blame Focal (neither JBL) on this...

Full disclosure, I bi-amp my Salon2s but I don't understand the notion of bi-wiring at all.
Bi-amping separates the amplifiers and reduces load on near zero out-of-band. It also reduces the possibility of crossover interaction.
You can get some headroom, which is about 1 dB in my system. Not a lot, but if the bass clicks the upper end is unaffected.

Bi-wiring just sells wire.

Perhaps the author is confused. There is no issue bi-wiring with the staps in place.
For bi-amping, your looking at fuse replacement and possibly a trip to the amp repair shop.

- Rich

 

[daftcombo]

Does anyone else not like pale speaker cones, especially with a black dust cap in the centre - I feel like I'm being watched.

I listen grills on.

 

[RichB]

It is nice to see a set a measurements at low and high (96 dB in this case) that includes distortion.
The sounstage measurements also have a nice chart showing deviation from the lower power measurements which show where speaker lose linearity.

Can these be made part of the standard protocol?

- Rich

 

[napilopez]

This is a weird one, though highly anticipated by me. The CSD seems to show that 800Hz peak/resonance, yet the port output must be combining with something that the tweeter is doing to make that happen, because it's not as pronounced as that 60Hz peak at the port and CSD.

That directivity, weird....

When you consider the nearfield measurements are not compensated for baffle step, some of the bump may be explained just by the fact the woofer output is decreasing while the port's response is actually rising in this region. Here's the difference with the nearfield measurements once you apply baffle step correction for their respective radiating areas.

Keep in mind this is very likely not fully accurate as I don't know the distance Amir measured the port and woofer. Ideally, a nearfield measurement of a 6.5-inch woofer (circa 5-inch pistonic diameter) would be done at about 0.25 inches, for example. I guessed amir didn't do it quite that close, so i estimated it from a bit further. The port measurement would be done right at the opening. Still, it's likely good enough for a quick evaluation.

I just traced the measurements with VituixCAD, so there's no native phase information. VituixCAD generates minimum phase though, which I think should be fairly accurate for these as they are individual driver responses, but I'm not sure. Summing by eyeballing the levels I get this:

Which is pretty close to amir's measurement without me going out of my way to try to match them...

Though I don't know the exact measurement conditions so who knows. Again, I'm not calculating the actual scaling for each driver and don't have proper phase information.

 

[ctrl]

Of course, since the harmonic content is way off human perception frequency wise. ...
Ja, there's a lot of confusion.
...
Such is not: "not" to be expected with these spikes from far away resonances. They do not: "not" copy down into the audible band.
...
Two measures are available, to save Your day. (1) do the IM test (2) mostly bad behavior culminates at some frequency, so that HD2 / HD3 / HD4 etc peak at the same frequency; in the far less critical case the HDx peaks appear all at different frequencies

As said in Post #72:

The harmonic distortions themselves would not be audible at 24kHz, but the possible intermodulation distortions are partially in the audible frequency range.

As already mentioned the break up sound pressure level of the tweeter can easily be 10-15dB above the reference sound level.

If the level of harmonic distortion at the base frequency is already high (e.g. due to a bad drive design), the break up can easily lead to very high harmonic distortion. This can result in 10-20% HD2 or/and HD3.

UPDATE: This is not correct, as @GelbeMusik pointed out, no IM is generated.
The interaction with the fundamental frequency then causes IM distortions which are partly in the audible range.

For the Aria 906 with an assumed breakup at 22kHz this would result in increased HD2 at 11kHz and HD3 at 7.3kHz.
So it could cause, if I understand it correctly, e.g. IMD at 22-7.3=14.7kHz and 22-2*7.3=7.4kHz when the tweeter is excited at 7.3kHz.

The IMD at 7.4kHz will result in beating with the fundamental at 7.3kHz. Whether this will still be audible in such a high frequency range is of course doubtful.

If there's no musical information up that high, would there still be a resultant hd2 and hd3 during playback?

No matter at what fundamental frequency, you will always have harmonic distortion. So at 7kHz base frequency you have HD2 (at 14kHz), HD3 (at 21kHz),...The harmonic distortions are always given in relation to the base frequency.

In the case of a tweeter with 22kHz break up frequency, we expect an increased HD3 at 7.3kHz (with 22 kHz).
If then 20% HD3 is produced, your dog will probably cut off his ears, but we humans will not be affected.
But as described above, IMD is still being created, some of which may be in the audible range again and thus be relevant for us.

 

[ironhorse128]

My 906s live in my living room on IsoAcoustics Stand. The white finish looks awesome.
The 906s are great. I just do not like them as much as my M16s.
PS. I have to protect the left speaker with a mouse pad as my wife likes to put things on top of my living room speakers.

 

[ctrl]

The broad bump in the midrange seems it would be troublesome. With our discussion on the Buchardt S400 high-Q midrange peak getting a lot of attention in your recent review, I would expect this speaker's wide-Q bump to be more noticeable to the ear. Luckily, I assume it can be EQ'd down rather easily.

The CSD seems to show that 800Hz peak/resonance, yet the port output must be combining with something that the tweeter is doing to make that happen, because it's not as pronounced as that 60Hz peak at the port and CSD.

Here I have opened a thread about resonances of the tested loudspeakers.

 

[napilopez]

Posted this in @ctrl's thread but thought I might as well here too that the bump around 1kHz seems to be a result of not fully compensating for the baffle:

Overlayed over amir's measurement:

 

[Zvu]

@napilopez It almost always is the case with standmount two way speakers. Incorporating full baffle step compensation in serial coil on midwoofer can lead to dip in lower mids. I guess that they considered this to be lesser evil and it was to expensive in that price range to solve that problem. It can easily be solved with one series RLC.

 

[amirm]

Keep in mind this is very likely not fully accurate as I don't know the distance Amir measured the port and woofer.

The port was measured with the microphone at the baffle. With the woofer, it was a few millimeters away. I have to take care to not hit anything!

 

[napilopez]

The port was measured with the microphone at the baffle. With the woofer, it was a few millimeters away. I have to take care to not hit anything!

I assumed as much! I dented one of my R3 review unit's Uni-Q when trying to do nearfield measurements

 

[QMuse]

Instead of trying to fill a dip at 3.000 Hz that we can't see on that estimated in-room graph, wouldn't it be better to try to fill with a low Q the broad dip between 1.000 Hz and 4.000 Hz on the basis of that curve?

It would.

 

[amirm]

@napilopez It almost always is the case with standmount two way speakers. Incorporating full baffle step compensation in serial coil on midwoofer can lead to dip in lower mids. I guess that they considered this to be lesser evil and it was to expensive in that price range to solve that problem. It can easily be solved with one series RLC.

Which we can fix for free with software filter.