Phantom Gold measurements: a stunning web first!

[srima]

The Devialet Phantom Gold is a relatively new speaker, and while I've seen reviews of it in various places, I've never seen actual measurements of it. Devialet themselves have never released any measurements of this speaker to my knowledge. 

I've had my Golds for a week now, and after about 50 or so hours of playing them, I figured it was high time to do some measurements. There are of course many ways to measure loudspeakers, and many measurement systems. I am going to report on measurements done using short sine wave sweeps, of the kind that are typically used in room correction systems. This is different from pink noise measurements as well as from impulse response measurements. 

In measuring speakers, many other factors need to be considered, such as the type of smoothing one chooses. I chose 1/3 octave smoothing, which produces a reasonably smooth curve while revealing gross anomalies. One can use 1/6th octave smoothing etc. 

My setup has been shown before in my review of the Phantom Gold's (which you can find in the forum on D400 and mono block units). It's been very well viewed, so I am going to assume you've seen that already. 

I will show below frequency responses and impulse responses for the left and right channel separately. This is done to get a more accurate picture of each individual speaker. Overall, the responses are simply STUNNING for a small speaker. The bass is very extended, in fact a bit too much since my Golds are close to the back wall, and extends to below 20 Hz. The response is remarkably devoid of large irregularities in the midrange and treble, although a few bumps and dips are always expected. The treble is slightly rolled off as it should be.  

The measured response fully vindicates my glowing review of the Gold's (glittering review?). Devialet is asking for a paltry $3K for a speaker that should cost upwards of $10K. It's a bargain in every sense of the word. Snap it up before they wise up, and bump up the prices. You won't find a deal like this anytime soon. This is a full high fidelity system, not a speaker, and one that will put many a more expensive speaker/hifi system to shame. It's compact, and elegant to look at (although one of my neighbors commented on how it looked like the kind of headphones that Apple likes to use -- remarkably apt description). 

I will provide more measurements in the days ahead, including distortion, and many others. I am using the OmniMic measurement system, which is very well thought of in the professional world. See the below links. 

http://www.daytonaudio.com/index.php/omn...ystem.html

$300 from Parts Express. Read the glowing reviews yourself: 

http://www.parts-express.com/dayton-audi...m--390-792

Dear me, I'm forgetting something. Ah yes, the measurements themselves! Here you go... :-) 

   

   

[Gremlin]

Whilst I am grateful to you for the time and effort that you have taken, none of your measurements can have any actual meaning unless they were done in at least a pseudo anechoic chamber. Otherwise, we are just seeing the responses of the speaker overlaid by those of the room in which the measurements were made. And the room is likely the dominant factor.

Unless there is something that you haven't told us about. Maybe.

[ogs]

The Devialet Phantom Gold is a relatively new speaker, and while I've seen reviews of it in various places, I've never seen actual measurements of it. Devialet themselves have never released any measurements of this speaker to my knowledge. 

I've had my Golds for a week now, and after about 50 or so hours of playing them, I figured it was high time to do some measurements. There are of course many ways to measure loudspeakers, and many measurement systems. I am going to report on measurements done using short sine wave sweeps, of the kind that are typically used in room correction systems. This is different from pink noise measurements as well as from impulse response measurements. 

In measuring speakers, many other factors need to be considered, such as the type of smoothing one chooses. I chose 1/3 octave smoothing, which produces a reasonably smooth curve while revealing gross anomalies. One can use 1/6th octave smoothing etc. 

My setup has been shown before in my review of the Phantom Gold's (which you can find in the forum on D400 and mono block units). It's been very well viewed, so I am going to assume you've seen that already. 

I will show below frequency responses and impulse responses for the left and right channel separately. This is done to get a more accurate picture of each individual speaker. Overall, the responses are simply STUNNING for a small speaker. The bass is very extended, in fact a bit too much since my Golds are close to the back wall, and extends to below 20 Hz. The response is remarkably devoid of large irregularities in the midrange and treble, although a few bumps and dips are always expected. The treble is slightly rolled off as it should be.  

The measured response fully vindicates my glowing review of the Gold's (glittering review?). Devialet is asking for a paltry $3K for a speaker that should cost upwards of $10K. It's a bargain in every sense of the word. Snap it up before they wise up, and bump up the prices. You won't find a deal like this anytime soon. This is a full high fidelity system, not a speaker, and one that will put many a more expensive speaker/hifi system to shame. It's compact, and elegant to look at (although one of my neighbors commented on how it looked like the kind of headphones that Apple likes to use -- remarkably apt description). 

I will provide more measurements in the days ahead, including distortion, and many others. I am using the OmniMic measurement system, which is very well thought of in the professional world. See the below links. 

http://www.daytonaudio.com/index.php/omn...ystem.html

$300 from Parts Express. Read the glowing reviews yourself: 

http://www.parts-express.com/dayton-audi...m--390-792

Dear me, I'm forgetting something. Ah yes, the measurements themselves! Here you go... :-) 

The overall frequency balance is quite similar to measurements I got from the White Phantoms.
I agree with Gremlin though; your measurements does show Phantom AND room. With some experimentation on placement  you'd be able to sort out the bass issues shown in the measurements.
Is there an option to extract a step response in the OmniMic software?

[srima]

Whilst I am grateful to you for the time and effort that you have taken, none of your measurements can have any actual meaning unless they were done in at least a pseudo anechoic chamber. Otherwise, we are just seeing the responses of the speaker overlaid by those of the room in which the measurements were made. And the room is likely the dominant factor.

Unless there is something that you haven't told us about. Maybe.

Precisely. One hears loudspeakers in typical rooms, not anechoic chambers. I have no interest in seeing such measurements since they do not tell me how a speaker will behave in a typical room. Room effects are largely significant in the bass region, below 200 Hz. Beyond that, one usually gets a pretty accurate picture of the speaker's sound.

What I'm showing is how 99.9% of the professional world of acousticians and sound engineers measure speakers prior to room correction. A sound engineer wants to see how the speaker measures in his studio, and unless he's hugely rich, does not have access to an anechoic room. He typically has no interest in anechoic measurements.

[srima]

The overall frequency balance is quite similar to measurements I got from the White Phantoms.
I agree with Gremlin though; your measurements does show Phantom AND room. With some experimentation on placement  you'd be able to sort out the bass issues shown in the measurements.
Is there an option to extract a step response in the OmniMic software?

Again, that's what I want: room + speaker is what I hear, and the input to room correction. I have no interest in some sterile anechoic response, which tells me nothing about how the speaker actually sounds in my listening room. 

Moving speakers by trial and error in a room to fix response anomalies is a 19th century approach: painful, tedious, mostly ineffective in many cases, and largely unnecessary in the modern DSP world of digital bits. What I plan to do is simply insert a digital equalizer (even a relatively cheap one like a Behringer is transparent in the digital domain, and there are much fancier ones like Z-systems), and reduce the large bass hump, and trim the treble a bit. The Behringer 2496, for example, has a mode that automates this. Tools like this are now used in every recording studio in the world. Studio engineers don't move heavy studio speakers around -- they just fix the response in the digital domain. Of course, some room treatment is useful, especially in the treble range.

[srima]

The overall frequency balance is quite similar to measurements I got from the White Phantoms.
I agree with Gremlin though; your measurements does show Phantom AND room. With some experimentation on placement  you'd be able to sort out the bass issues shown in the measurements.
Is there an option to extract a step response in the OmniMic software?

Again, that's what I want: room + speaker is what I hear, and the input to room correction. I have no interest in some sterile anechoic response, which tells me nothing about how the speaker actually sounds in my listening room. 

Moving speakers by trial and error in a room to fix response anomalies is a 19th century approach: painful, tedious, mostly ineffective in many cases, and largely unnecessary in the modern DSP world of digital bits. What I plan to do is simply insert a digital equalizer (even a relatively cheap one like a Behringer is transparent in the digital domain, and there are much fancier ones like Z-systems), and reduce the large bass hump, and trim the treble a bit. The Behringer 2496, for example, has a mode that automates this. Tools like this are now used in every recording studio in the world. Studio engineers don't move heavy studio speakers around -- they just fix the response in the digital domain. Of course, some room treatment is useful, especially in the treble range.

I forgot to mention that the responses I am showing are with the "blended" option in OmniMic, which means that room effects in the high frequency region are eliminated (by trimming the impulse response and looking only at the primary response, and not the later response). So, what you are seeing above the bass region is indeed the true sound of the speaker, and to the extent possible, room effects are eliminated by the software. The software has also the option of eliminating  room effects altogether, which I can also show.

[ogs]

The overall frequency balance is quite similar to measurements I got from the White Phantoms.
I agree with Gremlin though; your measurements does show Phantom AND room. With some experimentation on placement  you'd be able to sort out the bass issues shown in the measurements.
Is there an option to extract a step response in the OmniMic software?

Again, that's what I want: room + speaker is what I hear, and the input to room correction. I have no interest in some sterile anechoic response, which tells me nothing about how the speaker actually sounds in my listening room. 

Moving speakers by trial and error in a room to fix response anomalies is a 19th century approach: painful, tedious, mostly ineffective in many cases, and largely unnecessary in the modern DSP world of digital bits. What I plan to do is simply insert a digital equalizer (even a relatively cheap one like a Behringer is transparent in the digital domain, and there are much fancier ones like Z-systems), and reduce the large bass hump, and trim the treble a bit. The Behringer 2496, for example, has a mode that automates this. Tools like this are now used in every recording studio in the world. Studio engineers don't move heavy studio speakers around -- they just fix the response in the digital domain. Of course, some room treatment is useful, especially in the treble range.

I've used room correction for many years so I am quite familiar with the effect it has. I could not do without. What I (and probably Gremlin) actually commented was the fact that the measurements you present is Phantoms in your room with your speaker placement and your microphone position.
Acousticians and other sound pros would also use waterfall plots and time domain measurements to decide what needs to be done, not just the amplitude response.

The OmniMic software can apparently not show the step response. REW has this and also my Audiolense XO. It is the impulse response (which the OmniMic sw does show) displayed differently and easier to interpret.

[SwissBear]

Hi Srima,

Thanks for sharing. This is interesting.
Having extensively used OmniMic and REW, I would abound that REW is easier to use in the sense that, once you have made your measurements, you can exploit them in many different ways. The mic you have included with OmniMic is supported by REW and you can reuse its calibration file.

Having said that, it would be useful to have more detailed views (say with a smoothing of 1/12th octave or 1/24th octave instead of 1/3rd which is very favorable as you know). Also, it would be helpful to have a look at the time domain, say for instance the phase representation, just to be able to check if the speakers of the Phantom are correctly time-aligned. The step response would also give us an indication about this correct time alignment of speakers.

Could you share this with us ?
Thanks in advance.

[MountainGuy]

I too am interested to see if Phantoms are time aligned. Different people have different sensitivities to time alignment so for some, it is important.

[ogs]

I too am interested to see if Phantoms are time aligned.  Different people have different sensitivities to time alignment so for some, it is important.

White (and I assume Silvers) are not time aligned between the three (four) drivers. I have a hope this is one of the changes that make Golds sound better.