• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Most smoothed frequency response ever? Ex Machina Quasar MKII

Elkios

Active Member
Joined
Nov 7, 2020
Messages
114
Likes
61
Location
Australia
To me It's pretty obvious Exmachina is using FIR tuning, to a spot.
And as YSC and voodooless point out, the graphs are the same for different models.
I also noticed the low freq cutoff is claimed -6dB @ 23 Hz. I'd make it more like 28Hz going by their graph.

So it seems to me they are using FIR tuning to a spot make it look extra good, and then marketing bias to make it look extra, extra, good. :p

All that said, doesn't mean it's not a good speaker imo.
Just means we need better info, model specific haha, polars, and smoothing disclosure.
And hey, how about measurements when the speaker is not on FIR...the ones with 3ms latency.

Anyway, switching gears...I thought i'd back up my earlier statements that their frequency response and phase graphs don't necessarily mean heavy smoothing.
And that the impulse response could be real.

Got a major version upgrade to Smaart yesterday, and just now tried out it's improved capability with a quick "FIR tune to a spot", on a 4-way main speaker.
Done indoors at about 1m on a synergy, made for 120Hz up.

All graphs below have zero smoothing.

The first shows each of the 4 sections measured acoustic response. I need to say acoustic response because the responses almost look electrical, given how well formed they are. The power of FIR (to a spot) :)
120Hz high pass on low setion, with xovers then at 300Hz, 700Hz, and 6300Hz. All 16th order linear phase.
View attachment 234314



Second graph is with all sections playing.

View attachment 234315


Third is the impulse response.
View attachment 234318



So i hope that helps show how good 'FIR to a spot' can look....
The question always is, how well does a tune to a spot hold up over polars.

That's the info Exmachina owes customers imho.
I got onto Dev turns out there was a stuff up with the plots all for the one model .
 
OP
Recluse-Animator

Recluse-Animator

Senior Member
Joined
Jul 27, 2022
Messages
331
Likes
316
To me It's pretty obvious Exmachina is using FIR tuning, to a spot.
And as YSC and voodooless point out, the graphs are the same for different models.
I also noticed the low freq cutoff is claimed -6dB @ 23 Hz. I'd make it more like 28Hz going by their graph.

So it seems to me they are using FIR tuning to a spot make it look extra good, and then marketing bias to make it look extra, extra, good. :p

All that said, doesn't mean it's not a good speaker imo.
Just means we need better info, model specific haha, polars, and smoothing disclosure.
And hey, how about measurements when the speaker is not on FIR...the ones with 3ms latency.

Anyway, switching gears...I thought i'd back up my earlier statements that their frequency response and phase graphs don't necessarily mean heavy smoothing.
And that the impulse response could be real.

Got a major version upgrade to Smaart yesterday, and just now tried out it's improved capability with a quick "FIR tune to a spot", on a 4-way main speaker.
Done indoors at about 1m on a synergy, made for 120Hz up.

All graphs below have zero smoothing.

The first shows each of the 4 sections measured acoustic response. I need to say acoustic response because the responses almost look electrical, given how well formed they are. The power of FIR (to a spot) :)
120Hz high pass on low setion, with xovers then at 300Hz, 700Hz, and 6300Hz. All 16th order linear phase.
View attachment 234314



Second graph is with all sections playing.

View attachment 234315


Third is the impulse response.
View attachment 234318



So i hope that helps show how good 'FIR to a spot' can look....
The question always is, how well does a tune to a spot hold up over polars.

That's the info Exmachina owes customers imho.
Well I guess it's fair to say you couldn't do that with analog crossovers.
 

ernestcarl

Major Contributor
Joined
Sep 4, 2019
Messages
3,106
Likes
2,313
Location
Canada
To me It's pretty obvious Exmachina is using FIR tuning, to a spot.
And as YSC and voodooless point out, the graphs are the same for different models.
I also noticed the low freq cutoff is claimed -6dB @ 23 Hz. I'd make it more like 28Hz going by their graph.

So it seems to me they are using FIR tuning to a spot make it look extra good, and then marketing bias to make it look extra, extra, good. :p

All that said, doesn't mean it's not a good speaker imo.
Just means we need better info, model specific haha, polars, and smoothing disclosure.
And hey, how about measurements when the speaker is not on FIR...the ones with 3ms latency.

Anyway, switching gears...I thought i'd back up my earlier statements that their frequency response and phase graphs don't necessarily mean heavy smoothing.
And that the impulse response could be real.

Got a major version upgrade to Smaart yesterday, and just now tried out it's improved capability with a quick "FIR tune to a spot", on a 4-way main speaker.
Done indoors at about 1m on a synergy, made for 120Hz up.

All graphs below have zero smoothing.

The first shows each of the 4 sections measured acoustic response. I need to say acoustic response because the responses almost look electrical, given how well formed they are. The power of FIR (to a spot) :)
120Hz high pass on low setion, with xovers then at 300Hz, 700Hz, and 6300Hz. All 16th order linear phase.
View attachment 234314



Second graph is with all sections playing.

View attachment 234315


Third is the impulse response.
View attachment 234318



So i hope that helps show how good 'FIR to a spot' can look....
The question always is, how well does a tune to a spot hold up over polars.

That's the info Exmachina owes customers imho.

As amazing as it looks, no speaker manufacturer I am aware of "tunes" their speakers in such an extreme manner -- i.e. very "high-res" (brute-forced) magnitude and phase correction focused onto a tiny, singular, pin-point spot -- not even bothering (seemingly) to take into account overall directivity when it comes to their FIR EQ. This DSP approach comes off as naively myopic in my view. Now, to be fair, i'll presume ex machina's currently posted incomplete, low-res plots are in there for nothing more than marketing purposes.
 

alex-z

Addicted to Fun and Learning
Joined
Feb 19, 2021
Messages
910
Likes
1,685
Location
Canada
Their polar plots are notably lacking in resolution compared to the on-axis response. Every 15 degrees and 1 octave...

I would wager there are some moderate anomalies in the treble region + upper mid-range. There are certain problems that crop up, like the tweeter mounting gap causing a resonance, or the non-smooth edge transition causing diffraction. Things which you can EQ for on-axis, but can be revealed with sufficient measurement resolution or listening time. That is what sets apart KEF/Genelec coaxial units from the Seas coaxial units.
 
OP
Recluse-Animator

Recluse-Animator

Senior Member
Joined
Jul 27, 2022
Messages
331
Likes
316
Their polar plots are notably lacking in resolution compared to the on-axis response. Every 15 degrees and 1 octave...

I would wager there are some moderate anomalies in the treble region + upper mid-range. There are certain problems that crop up, like the tweeter mounting gap causing a resonance, or the non-smooth edge transition causing diffraction. Things which you can EQ for on-axis, but can be revealed with sufficient measurement resolution or listening time. That is what sets apart KEF/Genelec coaxial units from the Seas coaxial units.
So the Kef and Genelec designs are better than the Seas ones no matter how much money you throw at them to upgrade them as Ex Machina has done?
 

alex-z

Addicted to Fun and Learning
Joined
Feb 19, 2021
Messages
910
Likes
1,685
Location
Canada
So the Kef and Genelec designs are better than the Seas ones no matter how much money you throw at them to upgrade them as Ex Machina has done?

Pretty much. There are certain innovations like a radial phase plug, tweeter gap damper, and cone/voice coil decoupler which set KEF apart. Fancy cone materials don't compensate for that.

It puts the DIY and low volume manufacturers in a bit of a bind. Coaxial drivers are a low volume product, so tooling a production line to make better ones is expensive. Plus they have to avoid patent infringement.

I am not saying the Seas units are bad, you can still build a great speaker with them. I am just saying that KEF and Genelec have an inherent advantage.
 
Last edited:

Elkios

Active Member
Joined
Nov 7, 2020
Messages
114
Likes
61
Location
Australia
To me It's pretty obvious Exmachina is using FIR tuning, to a spot.
And as YSC and voodooless point out, the graphs are the same for different models.
I also noticed the low freq cutoff is claimed -6dB @ 23 Hz. I'd make it more like 28Hz going by their graph.

So it seems to me they are using FIR tuning to a spot make it look extra good, and then marketing bias to make it look extra, extra, good. :p

All that said, doesn't mean it's not a good speaker imo.
Just means we need better info, model specific haha, polars, and smoothing disclosure.
And hey, how about measurements when the speaker is not on FIR...the ones with 3ms latency.

Anyway, switching gears...I thought i'd back up my earlier statements that their frequency response and phase graphs don't necessarily mean heavy smoothing.
And that the impulse response could be real.

Got a major version upgrade to Smaart yesterday, and just now tried out it's improved capability with a quick "FIR tune to a spot", on a 4-way main speaker.
Done indoors at about 1m on a synergy, made for 120Hz up.

All graphs below have zero smoothing.

The first shows each of the 4 sections measured acoustic response. I need to say acoustic response because the responses almost look electrical, given how well formed they are. The power of FIR (to a spot) :)
120Hz high pass on low setion, with xovers then at 300Hz, 700Hz, and 6300Hz. All 16th order linear phase.
View attachment 234314



Second graph is with all sections playing.

View attachment 234315


Third is the impulse response.
View attachment 234318



So i hope that helps show how good 'FIR to a spot' can look....
The question always is, how well does a tune to a spot hold up over polars.

That's the info Exmachina owes customers imho.
Off axis response is now on the Exmachina site.
 

ernestcarl

Major Contributor
Joined
Sep 4, 2019
Messages
3,106
Likes
2,313
Location
Canada
Off axis response is now on the Exmachina site.

What’s posted looks like horizontal off-axis (up to 70 degrees) curves only, but much more sensible. The “on-axis” response now looks more realistic for this kind of coax where some linearity clearly has been sacrificed to get a better overall off-axis spatial response.
 

Elkios

Active Member
Joined
Nov 7, 2020
Messages
114
Likes
61
Location
Australia
What’s posted looks like horizontal off-axis (up to 70 degrees) curves only, but much more sensible. The “on-axis” response now looks more realistic for this kind of coax where some linearity clearly has been sacrificed to get a better overall off-axis spatial response.
If you want to know more I'd call Exmachina as Dev would answer your questions and enjoys a chat .
 

mSpot

Senior Member
Joined
Nov 15, 2018
Messages
404
Likes
519
25mm / 1 ”GrapheneQ ™ tweeter, coaxial alignment with midrange.
This is the first I've seen of a speaker manufacturer using a GrapheneQ graphene driver from ORA (https://www.oragraphene.com/), and in fact Ex Machina says that they were the first: https://exmachinasound.com/2020/01/10/a-primer-on-our-materials-science/

ORA had some publicity several years back when they did a Kickstarter for a headphone using their graphene drivers. They got conductor Gustavo Dudamel and a physics Nobel laureate to endorse it. As it happens so often on Kickstarter, the release kept getting delayed, and in the end the reviews on Head-Fi were mixed. Many things contribute to the sound quality of a headphone, and it wasn't clear what part of the sound was due to the graphene driver. They make a persuasive argument for making speaker cones from graphene, but I hadn't seen any further news about the company until now.
 
Last edited:

ernestcarl

Major Contributor
Joined
Sep 4, 2019
Messages
3,106
Likes
2,313
Location
Canada
If you want to know more I'd call Exmachina as Dev would answer your questions and enjoys a chat .

Thanks, Elkios! I don't think I'm the target group for their particular coax speakers -- and it's not as if I have that amount of cash to spend at my disposal! LOL

But the graphs for the off-axis curves posted for the Pulsar & Quasar seem to be identical on the site (?)

Dunno if that's an error or something they've yet to update...
 

jae

Major Contributor
Joined
Dec 2, 2019
Messages
1,208
Likes
1,508
Latest Patreon update from Erin:
Hey, folks. I *hope* to start providing content again by late December. I have the following speakers already tested and will provide a review of each over the next few months. Which would you prefer to see first?

Buchardt A700
Ex Machina Pulsar II

Monitor Audio 50 7G
Mesanovic CDM65 (Controlled Directivity Monitor) (Not yet released)
SVS Ultra Bookshelf
So if you are eager for those spins, easiest way is to let him know or support him so he prioritises this review before others.
 

Elkios

Active Member
Joined
Nov 7, 2020
Messages
114
Likes
61
Location
Australia
Thanks, Elkios! I don't think I'm the target group for their particular coax speakers -- and it's not as if I have that amount of cash to spend at my disposal! LOL

But the graphs for the off-axis curves posted for the Pulsar & Quasar seem to be identical on the site (?)

Dunno if that's an error or something they've yet to update...
Send him an email and see what he has to say you may be surprised
 

Elkios

Active Member
Joined
Nov 7, 2020
Messages
114
Likes
61
Location
Australia
But. How do they sound?
You could call me a Genelec and ATC fanboy but I've shifted . My ATC's are about to go the Gens went a while ago just a small shipping hurdle with the ATC right now . The only thing I can say is you should seek a set out and see if you react as I have .
 

tifune

Major Contributor
Forum Donor
Joined
Jan 18, 2020
Messages
1,085
Likes
767
Is there something to this checkered pattern Ex Machina and Perlisten are using, or is it just what's hot in the streets right now?
 

fpitas

Master Contributor
Forum Donor
Joined
Jul 7, 2022
Messages
9,885
Likes
14,191
Location
Northern Virginia, USA
As for the price...we've seen speakers come by here near $100k that don't measure as well. I think the $10k price is fair for an actual boutique product that performs as intended.
 

ernestcarl

Major Contributor
Joined
Sep 4, 2019
Messages
3,106
Likes
2,313
Location
Canada
Send him an email and see what he has to say you may be surprised

If I have questions that crop up in my mind in the future, maybe I will.

Is there something to this checkered pattern Ex Machina and Perlisten are using, or is it just what's hot in the streets right now?

This quote from the technology page pretty much summarizes the most interesting part:

By varying the properties of the TPCD in different areas and directions, the symmetric break up modes are eliminated and replaced with smaller and local distributed break up modes, enabling a more linear frequency response.

While smoothed to an unknown amount, I guess you can already see some of that behavior in the on- and off-axis response curves posted by Ex Machina:

1667614472223.png




Personally, I'm more interested in a coax design that manages to create a narrower pattern reaching down to the bass range even if it doesn't have the most linear response.

Example:

CCX1265
Fucrum Acoustic CCX1265.png

Eh, a different breed of speaker altogether... meant for PA installations with difficult acoustics in mind. I know it's kind of insane, but this coax speaker design piques my interest quite a bit more. ;)
 
Last edited:
Top Bottom