• 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!

Ascend Sierra Luna Duo Center/Main Speaker Review

CDMC

Major Contributor
Forum Donor
Joined
Sep 4, 2019
Messages
1,172
Likes
2,321
Our testing has led to a number of companies buying Audio Precision analysers. Prior to that, they would bulk at $28K price of that but not now. They much rather optimize the design themselves than waiting for me to test and find them!

It is rather amazing to me that people who are supposedly engineers designing electronics wouldn't have the equipment necessary to test their designs. It speaks volumes about this hobby that most "hi end" manufactures do not provide detailed measurements of their products, but spend pages discussing how their product is superior and the measurements cannot capture what makes it superior. I have reached the point, if a manufacture doesn't provide comprehensive measurements, I pass. It is ironic, we are seeing the measurements posted more by the "entry level" manufactures, who have the least profit margin in their products.
 

GimeDsp

Senior Member
Joined
Oct 30, 2020
Messages
418
Likes
362
Location
Earth
Here's a visualization of what I mean when I say the quasi-anechoic method often results in a resolution not suitable to show resonances in the midrange (and below). This is the Buchardt S400 I measured. Amir measured the same model (not the exact same speaker) and his results also showed this high-Q peak/dip around 500Hz.

In red is the response when I set the IR time to filter out anything beyond 4ms, which gets me down to about 250Hz. This is pretty typical of most measurements not performed in an anechoic chamber.

In black is the full anechoic response generated using Klippel's ISC module paired with a ground-plane reference measurement and my indoors measurement.

As you will see, the red line does not come close to showing the high-Q anomalies in the midrange. They are there. The resolution is just too low for the points to fill in and you are instead left with a smoothed over response compared to the anechoic.


View attachment 91487
This is true but as CNTRL, FLoyed Toole, and many others have pointed out there are ways to get good data without complex systems.
https://www.audioholics.com/loudspeaker-design/audio-measurements/audio-measurement-techniques.
"One method of removing the room while actually measuring a loudspeaker in a room is to do a summed nearfield response. This is where each driver is measured separately within its nearfield (usually within an inch from the center of the cone) and then each response is summed. Care must be taken to ensure only the driver under test is playing while being measured and proper driver summation of all the driver responses in calculated and plotted. "
However, in the case of the AA luna I am not sure proper math and summation would even be needed to show the BR port issues.
 

hardisj

Major Contributor
Reviewer
Joined
Jul 18, 2019
Messages
2,907
Likes
13,908
Location
North Alabama
This is true but as CNTRL, FLoyed Toole, and many others have pointed out there are ways to get good data without complex systems.
https://www.audioholics.com/loudspeaker-design/audio-measurements/audio-measurement-techniques.
"One method of removing the room while actually measuring a loudspeaker in a room is to do a summed nearfield response. This is where each driver is measured separately within its nearfield (usually within an inch from the center of the cone) and then each response is summed. Care must be taken to ensure only the driver under test is playing while being measured and proper driver summation of all the driver responses in calculated and plotted. "
However, in the case of the AA luna I am not sure proper math and summation would even be needed to show the BR port issues.

Absolutely. And I am not giving any excuses for why the data doesn't reflect the real response. Rather, just highlighting something that many don't realize. If the measurer (manufacturer in this case) doesn't understand the drawbacks of the measurement method then it's simply ignorance. Not deceit. And I think that's where Ascend falls here. I don't believe they would have purposely left something out. Now, that naturally brings up the aspect of "well, they should be aware of how to measure or model or both correctly" and I can't argue with that.

That said, NF measurements are not without issues as well, such as upper frequency limits and lack of baffle step being properly accounted for.

More and more, I would encourage anyone building speakers to send them off to Warkwyn to be tested. That way you don't just have reliable data that you can post on your site, you also have a 3rd party evaluating performance and able to point out issues you may have not noticed due to blinders/bias. One can do crossover stuff with quasi-anechoic methods but when you've wrapped up the design and are ready to get the 'real' data, pay the money and use it as a marketing tool to your benefit.
 

GimeDsp

Senior Member
Joined
Oct 30, 2020
Messages
418
Likes
362
Location
Earth
Absolutely. And I am not giving any excuses for why the data doesn't reflect the real response. Rather, just highlighting something that many don't realize. If the measurer (manufacturer in this case) doesn't understand the drawbacks of the measurement method then it's simply ignorance. Not deceit. And I think that's where Ascend falls here. I don't believe they would have purposely left something out. Now, that naturally brings up the aspect of "well, they should be aware of how to measure or model or both correctly" and I can't argue with that.

That said, NF measurements are not without issues as well, such as upper frequency limits and lack of baffle step being properly accounted for.

More and more, I would encourage anyone building speakers to send them off to Warkwyn to be tested. That way you don't just have reliable data that you can post on your site, you also have a 3rd party evaluating performance and able to point out issues you may have not noticed due to blinders/bias. One can do crossover stuff with quasi-anechoic methods but when you've wrapped up the design and are ready to get the 'real' data, pay the money and use it as a marketing tool to your benefit.

Seems fair. If they sell 1000 speakers and it cost $1000 to test it 3rd party they would only need to raise the price $1!
I would like to think even if they sell 100 speakers, they could raise the price and to pay for it and people would still be fine, then again I don't sell stuff, I only buy!
 

tktran303

Addicted to Fun and Learning
Forum Donor
Joined
Mar 27, 2019
Messages
683
Likes
1,181
Here's a visualization of what I mean when I say the quasi-anechoic method often results in a resolution not suitable to show resonances in the midrange (and below). This is the Buchardt S400 I measured. Amir measured the same model (not the exact same speaker) and his results also showed this high-Q peak/dip around 500Hz.

In red is the response when I set the IR time to filter out anything beyond 4ms, which gets me down to about 250Hz. This is pretty typical of most measurements not performed in an anechoic chamber.

In black is the full anechoic response generated using Klippel's ISC module paired with a ground-plane reference measurement and my indoors measurement.

As you will see, the red line does not come close to showing the high-Q anomalies in the midrange. They are there. The resolution is just too low for the points to fill in and you are instead left with a smoothed over response compared to the anechoic.


View attachment 91487
How does this happen Erin?

Conventional teaching/my interpretation that it’s accurate down to 1/x ms. Eg 5ms time window —accuracy is good doto 200Hz, and Reduced accuracy below.

If I completely miss that resonance... then I’m ignorant to it, and wouldn’t design it out...

Edit:

with a 5ms window- the accuracy is in 200Hz intervals or steps?
Eg. 2000Hz, 1800Hz, 1600Hz, 1400Hz... down to ...200Hz?

NOT (as I previously thought) accurate, in a continuous manner, down to 200Hz?

Is this the key to unlocking my confusion?
 
Last edited:

Beave

Major Contributor
Joined
May 10, 2020
Messages
1,330
Likes
2,728
It is rather amazing to me that people who are supposedly engineers designing electronics wouldn't have the equipment necessary to test their designs. It speaks volumes about this hobby that most "hi end" manufactures do not provide detailed measurements of their products, but spend pages discussing how their product is superior and the measurements cannot capture what makes it superior. I have reached the point, if a manufacture doesn't provide comprehensive measurements, I pass. It is ironic, we are seeing the measurements posted more by the "entry level" manufactures, who have the least profit margin in their products.

About 15 years ago, a company I was working for bought a small Swedish company that was doing A/V electronics sold in Europe. When we asked for data (measurements) on some of the audio and video outputs on one of their products, we were given data derived from an analog oscilloscope. One of their engineers, I think who had gotten a degree from Stanford, was doing everything he could to make good measurements with the (very limited) equipment they had. The process he had come up with was exceptionally clever and well conceived, but his results were still only gross approximations of actual performance. He was handcuffed because they wouldn't (or couldn't) actually afford to buy proper test gear (an Audio Precision, a Tektronix VM700, or similar such devices).

Meanwhile, the company I worked for had a room literally filled with old Audio Precisions and VM700s, probably near a dozen each, that had previously been used in a factory we operated that had shut down when we moved operations to another location.
 

Beave

Major Contributor
Joined
May 10, 2020
Messages
1,330
Likes
2,728
Here's a visualization of what I mean when I say the quasi-anechoic method often results in a resolution not suitable to show resonances in the midrange (and below). This is the Buchardt S400 I measured. Amir measured the same model (not the exact same speaker) and his results also showed this high-Q peak/dip around 500Hz.

In red is the response when I set the IR time to filter out anything beyond 4ms, which gets me down to about 250Hz. This is pretty typical of most measurements not performed in an anechoic chamber.

In black is the full anechoic response generated using Klippel's ISC module paired with a ground-plane reference measurement and my indoors measurement.

As you will see, the red line does not come close to showing the high-Q anomalies in the midrange. They are there. The resolution is just too low for the points to fill in and you are instead left with a smoothed over response compared to the anechoic.


View attachment 91487

This is precisely the issue with these resonances not showing up in the measurements Ascend provides.
 

ctrl

Major Contributor
Forum Donor
Joined
Jan 24, 2020
Messages
1,616
Likes
6,087
Location
.de, DE, DEU
This is what I think I’m understanding, or misunderstanding.

Is measuring a rear firing port or PR, and getting its “approximate FR up to 1-2KHz”, at a resolution that’s not high enough, to determine how it contributes, or affects. with it’s out of phase behaviour, to the frontal hemisphere response of the woofer?

I hope I understand you correctly, but I think there is a problem of communication and understanding.

The near-field measurements (directly at the BR ports and as close as possible to the woofer chassis), which have been described more often, have a very high resolution over the whole measuring range.

But if you measure with the microphone 0.2'' away from the dust cap of the woofer chassis, the measurement to high frequencies is less and less in accordance with reality. For example, a surround resonance is only insufficiently imaged in such a near field measurement - I have described further effects in the previous post.

The curves consisting only of the near field measurements have a high resolution and show the effects of the BR-Port resonances on the overall frequency response of the loudspeaker at 1kHz (with the mentioned limitations, therefore "approximated").
1604536394520.png



After having determined the total frequency response of the near-field measurements of the loudspeaker (red curve), which is shown in the upper diagram, this is combined with a gated frequency response measurement (1m micronphone distance) - and this at the point where both measurement curves match each other as closely as possible.

This gives the quasi anechoic free field frequency response of the loudspeaker. Since the joining of the two curves usually happens in the range around 300Hz, from 300Hz on you only have the resolution of the gated measurement.
This is where the problems begin that @hardisj has also described, where the effects of the BR-port resonances may not be recognized correctly due to the low resolution, because the FR is too strongly smoothed - especially in the lower frequency range of the gated measurement (300-1000Hz).
1604536996161.png


But the above near-field measurements have been determined anyway and these show with high resolution approximately the effects of the BR-Port resonances. Thus, a loudspeaker developer should always notice the possible problems.

Hope I could make it a bit more obvious.
 
Last edited:

sam_adams

Addicted to Fun and Learning
Joined
Dec 24, 2019
Messages
976
Likes
2,368
Hi,

Here is the EQ, really challenging to say the least.
Score no EQ: 1.41 poor! 4.26 with sub
Spinorama no EQ
...

Like for its little brother, many, many speakers, at much lower prices, measure much better...

I just saw your Matlab/Octave tool for Spinorama data. I don't know how I missed it, but with everything else going on... Very Nice!
 

sam_adams

Addicted to Fun and Learning
Joined
Dec 24, 2019
Messages
976
Likes
2,368
This is a speaker designed to be used with its ports open so that is how I tested it. It is not like it comes with any kind of plugs (which would be hard to do with rectangular ports). Bass response will also suffer. The near-field measurements have already shown the problem anyway. We don't need to spend an entire day re-measuring a speaker like this that in grand scheme of things, is not common.

Let's remember that external services charge $1000 to $1,500 for a measurement like I do. That is the opportunity cost when you ask me to keep measuring the same speaker rather than moving to another one.

Might I suggest a sock or two or a pair of Guinness bar towels, perhaps.

I know it would hold you up on further measurements, but in a case like this it might be enlightening to see what the interior construction of the speaker and its port are like. I wouldn't be suprised to find a flimsy plastic port tube acting like a kazoo inside. Additionally, as a D'Appolito design, we should expect to see at least a 3rd or 4th order XOver, but it looks like we have a 2nd order from the slopes.
 

GimeDsp

Senior Member
Joined
Oct 30, 2020
Messages
418
Likes
362
Location
Earth
I think Amir did enough. He tested the speaker and gave it a fair listen. It's really up to AA and other owners to get to the bottom of the issues with the Lunas. I know I have never been a fan of ribbon, amt, etc in 2 way designs. I just don't like the stark contrast in sound between the woofer and the tweeter. Ribbon with a dome mid in a 3 way is pretty good for me though!
 

Beave

Major Contributor
Joined
May 10, 2020
Messages
1,330
Likes
2,728
I think Amir did enough. He tested the speaker and gave it a fair listen. It's really up to AA and other owners to get to the bottom of the issues with the Lunas. I know I have never been a fan of ribbon, amt, etc in 2 way designs. I just don't like the stark contrast in sound between the woofer and the tweeter. Ribbon with a dome mid in a 3 way is pretty good for me though!

Err, uhm, don't ALL woofers and tweeters have a 'stark contrast in sound?' Can you clarify what you mean?
 

GimeDsp

Senior Member
Joined
Oct 30, 2020
Messages
418
Likes
362
Location
Earth
Err, uhm, don't ALL woofers and tweeters have a 'stark contrast in sound?' Can you clarify what you mean?
One subjective term subjective audiphiles use is "texture" when talking about different driver materials.
What we see in measurments that correlates to "texture" would be

1. Frequency extension
2. Distortion profiles
3. Break up modes
3. Dispersion
4. Mass
5. Transient response/Resolution
6. resonance

A microphone is much like a loudspeaker in a certan ways. Condenser mics are very low mass and can pick up all the quick transient nature of sound they record, dynamic mics are much higher mass and smear the fine details. Even if you EQ a high mass dynamic mic to mach a low mass condenser mic, it will never capture the same detail and will not sound "the same". This is because of all the things I did mention and many others I didn't.
 
Last edited:

Dennis Murphy

Major Contributor
Technical Expert
Joined
Mar 17, 2020
Messages
1,071
Likes
4,535
That said, NF measurements are not without issues as well, such as upper frequency limits and lack of baffle step being properly accounted for.
Could you expand on this a little? A nearfield measurement clearly won't show any baffle step effects if the crossover is not in line. If the crossover is in line and has BSC built in, the plot will show a roll-off in the response starting in the 300 - 400 Hz region on up that doesn't really exist at a normal listening distance. If you didn't know how to interpret that, someone might think the bass was bumped up. But beyond that, I'm not sure what the issue is, assuming you understand the nature of the baffle step and baffle step compensation. Isn't the real issue whether even a nearfield measurement has enough resolution to pick up what the Klippel machine picks up? There's also the issue of whether the human ear is as discriminating as the Klippel monster, but I'll leave that to a different discussion.
 

Beave

Major Contributor
Joined
May 10, 2020
Messages
1,330
Likes
2,728
One subjective term subjective audiphiles use is "texture" when talking about different driver materials.
What we see in measurments that correlates to "texture" would be

1. Frequency extension
2. Distortion profiles
3. Break up modes
3. Dispersion
4. Mass
5. Transient response/Resolution
6. resonance

A microphone is much like a loudspeaker in a certan ways. Condenser mics are very low mass and can pick up all the quick transient nature of sound they record, dynamic mics are much higher mass and smear the fine details. Even if you EQ a high mass dynamic mic to mach a low mass condenser mic, it will never capture the same detail and will not sound "the same". This is because of all the things I did mention and many others I didn't.

No offense, but 'texture' sounds a lot like audiophile silliness to me. How does one 'match' the 'texture' when selecting woofers and tweeters?
 

Francis Vaughan

Addicted to Fun and Learning
Forum Donor
Joined
Dec 6, 2018
Messages
933
Likes
4,697
Location
Adelaide Australia
How does one 'match' the 'texture' when selecting woofers and tweeters?
One approach that some advocate is to look at the relative harmonic distortion levels, particularly in the crossover region and choose drivers where there is a smooth transition for each harmonic between the drivers. There does seem to be some merit in the idea. How much it really matters is another matter.
It is pretty easy to choose drivers where there are quite significant differences in the relative levels.
 

Helicopter

Major Contributor
Forum Donor
Joined
Aug 13, 2020
Messages
2,693
Likes
3,944
Location
Michigan
Don't you know? Flax has a different texture than aluminum if you put your finger on it :)
The linen is a natural material, so it looks nice, in the same way as wood or stone. I had thought it might provide some dampening between the layers of glass, but Focal says it is just cosmetic. I suppose that makes sense too, because it shouldn't do much for stiffening with the glass there and the resin will probably do more to break up the glass. Since they are using prepregs, it doesn't change the mass of the resin, so the only change to sound would be the mass of the thin layer of linen, this is almost like a natural fiber gauze or cheesecloth. I'm not sure how much it changes the feel to the touch since it has glass on both sides.

16045789486975645796502075293459.jpg
 
Last edited:

Todd74

Active Member
Joined
Jun 11, 2020
Messages
165
Likes
91
If you used this speaker horizontally then where the above becomes your horizontal response, you better sit at the axis of tweeter and not move one bit! Never mind about anything else sitting to your left and right.
I brought up something pertaining to this in a thread discussing center channel speakers. For those of us only concerned about our own on-axis listening position, what would make for a better center?

If we use a traditional horizontal MTM center positioned under the screen, will it deprive us of our vertical response and thus make it feel more localized below the TV? .... and if so, would the answer to that be to use a single bookshelf on its side to essentially do the opposite of what we’re seeing in the graphs below, turning the stronger horizontal response into better vertical performance, just like how turning this speaker on its side did?

1604614210631.png

1604614235864.png
 
Top Bottom