• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required as is 20 years of participation in forums (not all true). Come here to have fun, be ready to be teased and not take online life too seriously. We now measure and review equipment for free! Click here for details.

Devialet Phantom Reactor 900 Measurements (now with Spinorama)

napilopez

Major Contributor
Joined
Oct 17, 2018
Messages
1,385
Likes
4,447
Location
NYC
#1
Another recently measured speaker I thought I'd share. It's an impressive showing from Devialet, at least in the frequency response department.

EDIT: Quasi-anechoic spinorama/CTA-2034A in this post.

Horizontal Response:

Reactor Horizontal Response.png


Note the graph goes down to 10Hz instead of the usual 20. Gated measurements(6.5ms) taken on tweeter axis from 1m away. Nearfield bass spliced below 200hz. Bass is corrected for baffle step, so it seems that bass hump from 150 to 25 Hz or so does actually exist, unless I screwed something up. (Anything different about the Devialet's layout that would dramatically alter the baffle step that I'm missing?).

In any case, impressive as it is to get 20hz out of a pair of 4-inch woofers, bass extension on an active speaker like this doesn't mean much without corresponding to a certain SPL level. Unfortunately, I haven't had the chance to blast music at crazy volumes to test for this, but my impression is that if you listen in the nearfield, your room is small, or you generally don't listen too loud, a pair of phantoms will have all the bass you need.

As for the vertical response, I've only had the chance to measure a couple points so far, but I suspect it won't be much different from the horizontal response given the trend so far and quasi-spherical design. It still looks very flat out to 30 degrees:

Reactor Verticall Response.png


Update: Posted this later in the thread, but figured I'd update the OP too. Here's a graph showing the stereo in-room response at my listening position 9 feet away. The back of the speakers are about 1 and 2 feet from the front wall (my wall is uneven).

I did not take special care in positioning, nor is this response averaged, so don't pay much mind to the overall room curve. This is mainly to show how the bass compresses at higher SPL levels. Purple is max volume, yellow is about 50 percent I believe.

Reactor Compression.png


For a pair of tiny speakers, I think this is really impressive. Though I listen from a pretty normal distance, I have a fairly large listening room. I'm in a studio loft, the speakers are basically radiating into the entire apartment. The space the speakers are in is roughly 13x30x16, not counting the hallway. Speakers are about 1-2 feet from the rear wall. So it's likely the SPL levels could be higher in an average room.

Notably, compression seems to only affect the sub-bass at 60 Hz and below, so even at max volume you are getting ample bass thump for a speaker this size, especially considering equal loudness curves suggest you need less bass at higher SPL levels. In any case, even the yellow line is already a fair bit louder than I typically listen, so the reactors still have some headroom for dynamic peaks.

Other thoughts:

It's pretty amazing that you can have a 2-way speaker this small, with a pair of 4-inch woofers, that can genuinely be considered 'full range' -- at least at lower volumes/nearfield. It has very wide dispersion too, with seemingly seamless driver integration.

It seems a really neat little speaker especially if you listen nearfield where compression in the bass isn't an issue. Makes me wonder what might've gone wrong with the bigger Phantom's design, with their weird huge dip at ~3000kHz. The Reactor has a dip there too, but it is much more shallow.

There's some noise when using the line in at higher volumes though not something I noticed until I was measuring. Haven't checked if that hiss happens with the optical input as I don't currently have an optical souce.. Seems the smoothest response is at about 15 degrees off-axis.

Devialet makes a great case for active speakers here. I can't tell you much with regards to distortion as that's not something I'm much experienced in measuring and interpreting, but everything else looks good.
 
Last edited:

Juhazi

Major Contributor
Joined
Sep 15, 2018
Messages
1,202
Likes
1,663
Location
Finland
#3
napilopez, can you please measure it up to 90 or even 180deg off-axis? The ball shape with boxer woofers should make it to have very smooth baffle diffraction variations, very diffrent from conventional speakers. Your estimated baffle loss compensation might not apply here.

The dip around 3kHz with Phantom Golds should be easy to eq, and it might have been done already by Devialet. Their amps have built-in eq too. I have heard the Phantoms Golds only and they were amazing!


https://darko.audio/2019/08/a-short-film-about-the-devialet-phantom-reactor-900/

I made a dsp-active 3-way speaker with SEAS MR18 coaxial in a ball, and it measured very nicely too, with moderate dsp adjustment. Here measurements of the mid-tweeter ball. My speaker has downfire 10" bass, it is a clone of Gradient 1.4
MR18 MT lr4 0-90 9ms 112.jpg
 
Last edited:

daftcombo

Major Contributor
Forum Donor
Joined
Feb 5, 2019
Messages
2,429
Likes
2,211
#4
Thanks!

On-axis response peaks a little in the highs. Doesn't it sound bright, even listening at 15° off-axis, since the on-axis response is louder than the off-axis?
 

Soniclife

Major Contributor
Forum Donor
Joined
Apr 13, 2017
Messages
3,359
Likes
3,414
Location
UK
#5
Thanks!

On-axis response peaks a little in the highs. Doesn't it sound bright, even listening at 15° off-axis, since the on-axis response is louder than the off-axis?
What drive unit are they using for the mid range upwards? And do you know the crossover point to it?
 
OP
N

napilopez

Major Contributor
Joined
Oct 17, 2018
Messages
1,385
Likes
4,447
Location
NYC
Thread Starter #6
napilopez, can you please measure it up to 90 or even 180deg off-axis? The ball shape with boxer woofers should make it to have very smooth baffle diffraction variations, very diffrent from conventional speakers. Your estimated baffle loss compensation might not apply here.

The dip around 3kHz with Phantoms should be easy to eq, and it might have been done already by Devialet. Their amps have built-in eq too. I have heard the Phantoms Golds only and they were amazing!


https://darko.audio/2019/08/a-short-film-about-the-devialet-phantom-reactor-900/

I made a dsp-active 3-way speaker with SEAS MR18 coaxial in a ball, and it measured very nicely too, with moderate dsp adjustment. Here measurements of the mid-tweeter ball. My speaker has downfire 10" bass, it is a clone of Gradient 1.4 View attachment 39215
Impressive! Funny you mention measuring out to 180-degrees; the Reactor is the first such speaker I'd done so for. But I was mainly doing it to calculate an early reflection's curve, so I left the above graph up to 75 degrees in the interest of making it easier to compare with my other measurements.

Here's the graph from 0-90+180 degrees. As you mention, it's quite smooth all the way. I also threw in an estimated Early Reflections curve, based on the horizontal response. I'll update it once I've done a full suite of verticalmeasurements, though it's a bit tricky because of the round shape (normally I just place a speaker on its side and rotate; doesn't work here with the side-firing woofers)

Reactor Horizontal Response 180 ER.png

Thanks!

On-axis response peaks a little in the highs. Doesn't it sound bright, even listening at 15° off-axis, since the on-axis response is louder than the off-axis?
It does not sound bright. It's a very small peak, and if you are listening at 15 degrees off-axis, you can't hear the direct sound until it's been reflected (and therefore attenuated) by multiple surfaces. So it wouldn't be louder than the direct sound at 15 degrees once it actually reaches your ears.

What drive unit are they using for the mid range upwards? And do you know the crossover point to it?
Soundstage mentioned the crossover as 525 Hz. I'm not sure where they got that figure from, but it seems to line up with my nearfield measurements. It's quite the unusual setup, enabled by the magic of DSP and active crossovers.
 

Attachments

Last edited:

Juhazi

Major Contributor
Joined
Sep 15, 2018
Messages
1,202
Likes
1,663
Location
Finland
#7
The ball has smooth baffle loss/step, but the bass must be practically omnispherical, because of two drivers on opposing surfaces and crossed around 500Hz. I bet the xo was chosen to get maximally smooth change in dispersion.

My cone-coaxial and Devialet's dome-coaxial have remarkable difference in directivity between 1-10kHz! They should sound different too. It would be funt to have the passive Gradient 1.4s and Phantoms side by side for listening test!


napilopez, step response? It would reveal xo type, I suspect LR4 for mid-tweeter but woofer-mid might be LR2.
 

Arnandsway

Senior Member
Joined
Sep 13, 2018
Messages
431
Likes
366
#8
The ball has smooth baffle loss/step, but the bass must be practically omnispherical, because of two drivers on opposing surfaces and crossed around 500Hz. I bet the xo was chosen to get maximally smooth change in dispersion.

My cone-coaxial and Devialet's dome-coaxial have remarkable difference in directivity between 1-10kHz! They should sound different too. It would be funt to have the passive Gradient 1.4s and Phantoms side by side for listening test!


napilopez, step response? It would reveal xo type, I suspect LR4 for mid-tweeter but woofer-mid might be LR2.
This gradient is a very unusual design. I really like it. Do you have a thread on the gradient-clone build?
 

oivavoi

Major Contributor
Forum Donor
Joined
Jan 12, 2017
Messages
1,457
Likes
1,126
Location
Oslo, Norway
#9
Thanks, @napilopez ! Very interesting. It seems to have somewhat worse off-axis response than the Golds, no? At least comparing your measurements here to the NRC measurements of the Golds. On the other hand it doesn't have the above-mentioned dip around 2 K, even though that dip is pretty easy to eq away.
 

Soniclife

Major Contributor
Forum Donor
Joined
Apr 13, 2017
Messages
3,359
Likes
3,414
Location
UK
#10
Soundstage mentioned the crossover as 525 Hz. I'm not sure where they got that figure from, but it seems to line up with my nearfield measurements. It's quite the unusual setup, enabled by the magic of DSP and active crossovers.
That's quite a range to cover with a single driver, is it breaking up at the top end?
 
OP
N

napilopez

Major Contributor
Joined
Oct 17, 2018
Messages
1,385
Likes
4,447
Location
NYC
Thread Starter #13
The ball has smooth baffle loss/step, but the bass must be practically omnispherical, because of two drivers on opposing surfaces and crossed around 500Hz. I bet the xo was chosen to get maximally smooth change in dispersion.

My cone-coaxial and Devialet's dome-coaxial have remarkable difference in directivity between 1-10kHz! They should sound different too. It would be funt to have the passive Gradient 1.4s and Phantoms side by side for listening test!

napilopez, step response? It would reveal xo type, I suspect LR4 for mid-tweeter but woofer-mid might be LR2.
Keep in mind I'm talking about the new, smaller Phantom Reactor. The Reactor doesn't have proper a tweeter - it uses two 4-inch woofers and then just a single 1-inch-ish unit they call a 'full range' driver handling the rest. Not sure what kind of driver that is honestly. Here's the step response:

Step response.png

Thanks, @napilopez ! Very interesting. It seems to have somewhat worse off-axis response than the Golds, no? At least comparing your measurements here to the NRC measurements of the Golds. On the other hand it doesn't have the above-mentioned dip around 2 K, even though that dip is pretty easy to eq away.
While it's always somewhat risky to compare measurements from different sources, I don't see why the Reactor's off-axis measurements look worse than the golds; if anything, they look better to me.

As for the dip, if you EQ it away, then you introduce a peak in the off-axis response. Some would probably help, but you wouldn't fully fix the issue.

Side note: I'm not nearly as smart/experienced as the folks over at the NRC, but I did notice when taking my measurements that the Phantom Reactor is particularly sensitive to its placement on the measuring platform, likely because of the quasi-spherical shape and low-sitting highs driver. I saw a smaller but similar dip in the Reactor's response when there was just an inch or two of a 'lip' in front of the speaker's base. Adjusting my rig so that tweeter saw less of the platform eliminated the issue. If you buy these, it's important to make sure the base is pushed all the way to the edge of their platform.

Again, I don't claim to know more than the NRC/soundstage, but it just seems like such an egregious mistake that Devialet managed to let a ~6dB dip into its flagship speaker's frequency response.

Anyway, I've been spending some more time listening today before I run a final suite of measurements and ship them back. Pushed them with some hip hop. It's a ridiculous amount of sub-bass for something this size; even pushing the speaker to uncomfortable listening levels in my room (I listen about 10 feet from the speakers), the bass doesn't seem to be giving up. Turning the Devialet app's volume knob to 65/100 is already much louder than I ever to listen to/borderline unbearable, but I still couldn't hear anything letting up in the bass.

Everytime I test a competently design active speaker it's hard to go back to the passive world.
 
Last edited:

Soniclife

Major Contributor
Forum Donor
Joined
Apr 13, 2017
Messages
3,359
Likes
3,414
Location
UK
#15
Turning the Devialet app's volume knob to 65/100 is already much louder than I ever to listen to/borderline unbearable, but I still couldn't hear anything letting up in the bass.
How loud approximately would that be, in room at the LP?
 

Purité Audio

Major Contributor
Industry Insider
Barrowmaster
Forum Donor
Joined
Feb 29, 2016
Messages
4,104
Likes
2,350
Location
London
#16
This gradient is a very unusual design. I really like it. Do you have a thread on the gradient-clone build?
They are super sounding speakers, cardioid response, co-axial mid/treble very neutral,
Keith
 

Juhazi

Major Contributor
Joined
Sep 15, 2018
Messages
1,202
Likes
1,663
Location
Finland
#17
Keep in mind I'm talking about the new, smaller Phantom Reactor. The Reactor doesn't have a tweeter - it uses two 4-inch woofers and then just a single 1-inch-ish driver handling the rest. Not sure what kind of driver that is honestly. Here's the step response:

View attachment 39247
---
Oh, then I guess there is a 32mm dome tweeter or 2" cone with large dustcap (they call it fullrange) and xo seems to be LR4 with one cycle delay for the woofer.

Distortion through passband?
 

Frank Dernie

Major Contributor
Forum Donor
Joined
Mar 24, 2016
Messages
3,891
Likes
8,001
Location
Oxfordshire
#19
Keep in mind I'm talking about the new, smaller Phantom Reactor. The Reactor doesn't have proper a tweeter - it uses two 4-inch woofers and then just a single 1-inch-ish unit they call a 'full range' driver handling the rest. Not sure what kind of driver that is honestly. Here's the step response:

View attachment 39247



While it's always somewhat risky to compare measurements from different sources, I don't see why the Reactor's off-axis measurements look worse than the golds; if anything, they look better to me.

As for the dip, if you EQ it away, then you introduce a peak in the off-axis response. Some would probably help, but you wouldn't fully fix the issue.

Side note: I'm not nearly as smart/experienced as the folks over at the NRC, but I did notice when taking my measurements that the Phantom Reactor is particularly sensitive to its placement on the measuring platform, likely because of the quasi-spherical shape and low-sitting highs driver. I saw a smaller but similar dip in the Reactor's response when there was just an inch or two of a 'lip' in front of the speaker's base. Adjusting my rig so that tweeter saw less of the platform eliminated the issue. If you buy these, it's important to make sure the base is pushed all the way to the edge of their platform.

Again, I don't claim to know more than the NRC/soundstage, but it just seems like such an egregious mistake that Devialet managed to let a ~6dB dip into its flagship speaker's frequency response.

Anyway, I've been spending some more time listening today before I run a final suite of measurements and ship them back. Pushed them with some hip hop. It's a ridiculous amount of sub-bass for something this size; even pushing the speaker to uncomfortable listening levels in my room (I listen about 10 feet from the speakers), the bass doesn't seem to be giving up. Turning the Devialet app's volume knob to 65/100 is already much louder than I ever to listen to/borderline unbearable, but I still couldn't hear anything letting up in the bass.

Everytime I test a competently design active speaker it's hard to go back to the passive world.
Why are you sending them back then?
 
OP
N

napilopez

Major Contributor
Joined
Oct 17, 2018
Messages
1,385
Likes
4,447
Location
NYC
Thread Starter #20
How loud approximately would that be, in room at the LP?
Here are the in-room, stereo measurements from the listening position 9 feet away in a large listening room. The speakers are 1 and 2 feet away from the rear wall (my front wall is uneven).

Keep in my I didn't take any care for the speaker's positioning when making these measurements, so you can effectively ignore the response above the bass; this is just to show you how the bass response changes as you push the speakers to their max.

Reactor Compression.png


The yellow line is the '50' volume level in Devialet's app. The purple line is maxed out volume.

There's some compression happening below 60Hz, but frankly less than I expected. I also appreciate that the compression is limted to the very lowest portion of the speaker's response. Even at max volume you're still getting more thump than you'd expect for a speaker this size.
 
Last edited:

Similar threads

Top Bottom