Ascend from a DIY'er's standpoint...

[Jay_WJ]

It's been over a year since I wrote a post here. Now I became a speaker DIY'er, and all commercial speakers I purchased are gone including Ascend CBM-170SE and Onix x-ls. As a DIY'er, I now have a different perspective and standards for evaluating speakers. At this point, if someone asks me to recommend most cost-effective and reliable commercial loudspeakers, I won't hesitate to pick Ascend Acoustics. I once preferred the Onix x-ls to the 170SE, but now I can re-evaluate them. A pair of CBM-170SE's was the most accurate and cleanest speakers among many I auditioned under $500 with respect to linear and nonlinear distortions. Its crossover is nicely tuned---for example, its baffle step compensation is just a right amount that can fit most people's listening environment.

Dave, keep up the good work! You certainly know what you're doing.

-jAy

[curtis]

Hey Jay....welcome back.

Can you tell us how your preference evolved over the past year?

[Jay_WJ]

It's not just about the change of my preference. From the experience of using different drivers and tweaking crossovers, my ears has become able to better discern what kind of sound has low distortion, better tonal balance across different frequency ranges, smoother power response, and so on. The CBM-170SE I experienced is a well balanced design in various aspects so that it can satisfy as wide a range of people in different listening environments as possible. An extremely well engineered product that shows the designer's effort put into its development.

-jAy

[DougMac]

Jay,

Did you come to your preferences solely from listening or did you do some measurements? I was thinking that as a DIY'er you may use REW or some other evaluation tools.

Doug

[curtis]

It's not just about the change of my preference. From the experience of using different drivers and tweaking crossovers, my ears has become able to better discern what kind of sound has low distortion, better tonal balance across different frequency ranges, smoother power response, and so on. The CBM-170SE I experienced is a well balanced design in various aspects so that it can satisfy as wide a range of people in different listening environments as possible. An extremely well engineered product that shows the designer's effort put into its development.

-jAy
Thanks Jay....that is the type of explanation I was looking for. It really seems you have taken the time to understand these different elements.

[curtis]

Jay,

Did you come to your preferences solely from listening or did you do some measurements? I was thinking that as a DIY'er you may use REW or some other evaluation tools.

Doug
REW is really just for what the name implies....Room EQ Wizard. I don't think it is really a good tool for evaluating speakers. You could use it compare in room FR of different speakers, but it can't tell you stuff that Jay has mentioned.

[Jay_WJ]

Jay,

Did you come to your preferences solely from listening or did you do some measurements?
Doug

Measurements are a useful tool for designing a preliminary xover and diagnosing certain anomalies. But no speaker designers finalize/evaluate their projects based on measurements. Various kinds of balance, such as the amount of baffle step compensation, port tuning, tweeter-woofer level balance, etc. should be adjusted by listening feedback.

For a longer explanation on this issue and what I do to obtain a satisfactory result from tweaking crossovers, see the last section titled "Voicing your speaker" of my following web page:

http://www.geocities.com/woove99/Spkrbldg/DesigningXO.htm

-jAy

[davef]

Welcome back Jay!


Measurements are a useful tool for designing a preliminary xover and diagnosing certain anomalies. But no speaker designers finalize/evaluate their projects based on measurements. Various kinds of balance, such as the amount of baffle step compensation, port tuning, tweeter-woofer level balance, etc. should be adjusted by listening feedback.


Sorry -- not true. While we might make very minor changes to a crossover after listening, the only way to get accurate results for the amount of baffle step compensation, tweeter attenuation and the correct port tune frequency is by precise measurements... Listening is far to environment dependent to make such critical adjustments. It is the major difference between DIY (designing for yourself) or designing for mass appeal.

Of course, one has to know how to take the precise measurements, have the right tools to do so, know which measurements to take and how to interpret the results. It is a rather complex process, all things considered.

[Jay_WJ]

Sorry -- not true. While we might make very minor changes to a crossover after listening, the only way to get accurate results for the amount of baffle step compensation, tweeter attenuation and the correct port tune frequency is by precise measurements... Listening is far to environment dependent to make such critical adjustments. It is the major difference between DIY (designing for yourself) or designing for mass appeal.

Of course, one has to know how to take the precise measurements, have the right tools to do so, know which measurements to take and how to interpret the results. It is a rather complex process, all things considered.

Hi Dave!

It seems that I may want to disagree with you. BUT I think we're on the same page. Apparently, you misunderstood my statement. I DO know the importance of objective measurements in speaker design. Any, particulary commercial, speaker designs should be backed up by measurements.

But, whether the designer is a professional or a hobbist, he does not finalize a design simply by interpreting a set of measurements. Some DIY'ers are able to obtain precise measurements like John Krutke, Mark Krawiec, John Marsh, John Krevosky, etc. But they know tonal balance of a speaker is not a simple, linear sum of on-axis, off-axis linearity (i.e., linear distortions) and nonliear distortions. Tonal balance is a nonlinear sum of them, which can be properly judged by listening.

We also need to consider tonal balance in actual listening environments. But how do we do that? In an anechoic room? A final step of tuning or voicing by listening is a must in any loudspeaker designing process, whether the designer is a professional or a hobbist. Of course, how much weight they place in objective measurement and tweaking via listening can be different from a designer to another.

-jAy

[davef]

Hi Dave!

We also need to consider tonal balance in actual listening environments. But how do we do that? In an anechoic room? A final step of tuning or voicing by listening is a must in any loudspeaker designing process, whether the designer is a professional or a hobbist. Of course, how much weight they place in objective measurement and tweaking via listening can be different from a designer to another.

-jAy

Still, I disagree with your statements. You would be extremely surprised by just how many hugely succesful loudspeakers have been approved / finalized for production by interpreting various measurements. I can name at least a dozen.


We also need to consider tonal balance in actual listening environments. But how do we do that? In an anechoic room?

Tonal balance is one of the easier aspects to get right by use of measurements. With the proper equipment, one can look at the power response with no reflections (anechoic), first reflections only, second reflections, 3rd etc. This data is critical to determine how the speaker will perform in various rooms

Since you have a newfound respect for our CBM-170, you should know that every aspect of this speaker was designed and finalized by measurements only :p Of course, I did extensive listening, but -- in my experience and from the highly regarded engineers I have worked with side-by-side over the years - we have learned to never fully trust our ears. Too many variables involved and too much risk to make critical adjustments using your ears. Your own hearing abilities can vary day to day...

Most professional designers I know use measurements from start to finish. If something does not sound quite right to them, they go back and try to identify what they are hearing using measurements, then make comparative adjustments using measurements, go back and listen -- then the process repeats.

--- I once submitted a final crossover design to the powers-that-be for listening and final approval. It was initially rejected. As an experiment, I re-submitted it a few days later to the same listener with documented changes and it was immediately approved. Funny thing, I changed absolutely nothing and the listening was done by someone with vast experience. This speaker was a huge commercial success....

Jay -- it is OK to disagree. Everyone has their techniques but you should not speak in absolutes.

To say:


But no speaker designers finalize/evaluate their projects based on measurements.

is just not the case.

[Jay_WJ]

Dave, of course, I cannot write an article about designing speakers. But will you be surprised if I mostly agree with what you're saying above? You need to know that I'm also an objectivist and totally agree with what John Krutke states here:

http://www.zaphaudio.com/evaluation.html

And I also agree that objective measurements should be major decision factors in speaker design. But there is no absolutely right, single choice in various design choices (BSC, for example). They should depend on the speaker's intended use. But if you believe that EVERY single aspect in speaker design can be determined by seeing measurements, I have to disagree with you. For example, midrange (or bass-mid) tonal balance is largely affected by a proper amount of BSC. How do you determine a right amount of BSC? Using an anechoic response only? Or using your quasi-anechoic (i.e., nearfield and farfield merged response) only? Then, you miss one important thing. Every midwoofer has a different nonlinear, harmonic distortion profiles across frequencies. Some woofers have relatively high bass distortions than midrange ones, and others exhibit opposite behavior. This SHould be considered in choosing a right amount of BSC. How? Simply looking at its FR and harmonic distortion sweep separately? No, that doesn't work. You have to listen to the design in different, intended listening environments to determine BSC amount that gives desirable tonal balance.

As you said before, designing speakers is a complex process. I do agree with you.

-jAy

[Jay_WJ]

Dave, I think we're simply looking at different sides of the same coin. Let me ask this. Do you think your skill of interpreting various measurements are totally unrelated to your listening experience? If so, that's untrue (I can say this in absolutes :)). How do you think you have formed your skill of interpreting measurements? How do you know what a certain result of measurement will sound like? Your skill's definitely developed via countless (since you're a professional speaker designer) occasions of associating visual measurements with actual listening experience. Don't you agree?

I already noted this aspect in my web page:

"... Second, even if you have very accurate, measured system FRs from on-axis and off-axis, usually you don't have ability to translate the measurements into actual listening experience unless you are a professional or an experienced DIYer who has voiced speakers many, many times via listening and measurement feedback."

I'm a relatively inexperienced DIY'er. Now you may be able to understand why I said such things. But I still believe even professionals and experienced DIY'ers still have things to learn, in every new project they undertake seriously, by matching what they see in measurements and what they hear in listening.

-jAy

[davef]

But if you believe that EVERY single aspect in speaker design can be determined by seeing measurements, I have to disagree with you. For example, midrange (or bass-mid) tonal balance is largely affected by a proper amount of BSC. How do you determine a right amount of BSC? Using an anechoic response only? Or using your quasi-anechoic (i.e., nearfield and farfield merged response) only? Then, you miss one important thing. Every midwoofer has a different nonlinear, harmonic distortion profiles across frequencies. Some woofers have relatively high bass distortions than midrange ones, and others exhibit opposite behavior. This SHould be considered in choosing a right amount of BSC. How? Simply looking at its FR and harmonic distortion sweep separately? No, that doesn't work. You have to listen to the design in different, intended listening environments to determine BSC amount that gives desirable tonal balance.
-jAy

As I am sure you are aware, we all hear differently, Jay. The shape of our ears, heads, our age, the condition of our sinuses. When designing a speaker for yourself (DIY) you know what sounds "right" to you. Even in the same room, you can not determine if the balance or tonal qualities that sound correct to you will sound that way to someone else. Provided you have the right equipment at your disposal, precise measurements are the only method to accurately determine the critical aspects of loudspeaker performance.

For example, you can not -- by ear, determine the power response, or the polar response, electrical crossover point, phase, filter slope, Q, impedance and so many more critical aspects


Every midwoofer has a different nonlinear, harmonic distortion profiles across frequencies. Some woofers have relatively high bass distortions than midrange ones, and others exhibit opposite behavior.

And how do you determine these distortions? By ear? I hope not :p


How do you determine a right amount of BSC? Using an anechoic response only? Or using your quasi-anechoic (i.e., nearfield and farfield merged response) only?

I gave you some insight in my previous post... I am certainly not going to publicly post my techniques, but I will say this --- Ascend loudspeakers have been highly regarded for their tonal balance and midrange performance... No Ascend loudspeaker has ever had the amount or character of baffle step compensation tweaked by ear. And none ever will :cool:

You can't really compare DIY to mass market designs, it is an entirely different philosophy. What I look for in measurements and how I use these in design is based on feedback gathered from tens of thousands of people (possibly hundreds of thousands) -- consumers who own product that I have designed. Take tweeter attenuation for example, if the majority of customers said too bright, I know I am doing something wrong. If they said too dull, same thing. Some say too bright, others say not bright enough, the majority say great balance (regardless of how I hear it). While it takes many years of experience, many different models and feedback from many thousands of people, I am able to relate the majority of loudspeaker characteristics directly to various measurements.

Don't be fooled by frequency response measurements --- I take hundreds of various measurements, both acoustically and electrically. If I don't find what I am looking for with an acoustic measurement, I can usually find it through an electrical measurement or by use of accelerometers.

Our ears are inherently flawed devices and our hearing changes dramatically in the course of our lives. I leave the determination of performance to our equipment, never to my ears. Certainly, I consider myself to have an extremely critical ear (been critically listening for 1/4 century now). I can pick out flaws almost instantly but to specifically isolate those flaws and compensate for them, I identify them by measurement and proceed from there.

While I wish it were true – I am most certainly not the “only” loudspeaker designer to design in this manner.

Jay, my advice: read whatever you can get your hands on, design, design and then design some more – get as much feedback as is possible about your designs and then start to develop your own techniques. This is science and art combined and I know a successful commercial designer who is partially deaf and never listens, I know designers who never measure and I know those who use various percentages of the two. I have said this before many times, you need to know just what you are designing for before you head into a project. For my products, it has always been accuracy or lack of color (correct balance) and in order to achieve this, measurements are the only thing I truly trust... and yes, I have an exceptional ear – when the kids haven’t de-sensitized my hearing that day from yelling and screaming ;)

[Jay_WJ]

Again, I can say that I'm more on your side than the opposite about the way of designing loudspeakers. Yet, I still don't believe that you developed such an excellent product as CBM-170 using ONly what you see in measurements. In what sense? As I said above, your skill of properly interpreting measurements is closely related to your past experience of hearing.


If something does not sound quite right to them, they go back and try to identify what they are hearing using measurements, then make comparative adjustments using measurements, go back and listen -- then the process repeats.

As you noted above, listening is an important element in speaker design. You cannot obtain a successful result by using ONly measurements. And be careful in understanding what I mean by "listening." I never meant we need to use our ears without any external constraints. As you pointed out in your above post, to use a certain driver's HD profiles and understand how they affect a speaker's tonal balance, precise distortion measurements should be available. Otherwise, it'd be just like groping in the dark. I'm 100% sure that competent speaker designers, including you, use listening feedback to find a good balance in various aspects of speaker's tonality in well constrained conditions made possible by measurements. When used in this rigorous manner, our hearing is not used just subjectively. Of course, you're going to confirm your design choices with measurements. But note that these decisions are not made entirely by measurements in the above sense. If you thought I meant that we can build a loudspeaker merely by tweaking by ear, you misunderstood me.

Ultimately, we HEAR speakers, not SEE them. Measurements are simply an objective, graphical view of what we hear. We can't judge which is better or which is more important for decision making in speaker design without translating the measurements into actual listening experience. Professional designers or experienced DIY'ers have sort of a detailed map for this translation which they've built up by experience. Equal loudness contours are a good example that links physical measurements to human loudness perception. As a speaker desinger, you should know a hump in a certain frequency range is more obtrusive to ear than one in another range. This kind of knowledge is used when you interpret your measurements. Without listening experience, you can't make this kind of judgment.

-jAy

[Mike^S]

Hi Dave, since this is somewhat a discussion on measurements and speaker design, I have a question for you.

Someone I have chatted with thinks the Sierra lacks baffle step compensation and points to the NRC measurements as evidence. I have no idea if the Sierra does or not, but I would find I find it hard to believe that you would not include this in the design if it needed it.

I remember you discussed the NRC measurements once before. They measure in an anechoic chamber, which doesn't accurately show the effect of the rear mounted port, correct? Also they measure with the mic placed in line with the tweeter instead of halfway between the tweeter and the woofer, is that right?

[davef]

Hi Dave, since this is somewhat a discussion on measurements and speaker design, I have a question for you.

Someone I have chatted with thinks the Sierra lacks baffle step compensation and points to the NRC measurements as evidence. I have no idea if the Sierra does or not, but I would find I find it hard to believe that you would not include this in the design if it needed it.

I remember you discussed the NRC measurements once before. They measure in an anechoic chamber, which doesn't accurately show the effect of the rear mounted port, correct? Also they measure with the mic placed in line with the tweeter instead of halfway between the tweeter and the woofer, is that right?

Hi Mike,

Hope all is well!


Someone I have chatted with thinks the Sierra lacks baffle step compensation and points to the NRC measurements as evidence. I have no idea if the Sierra does or not, but I would find I find it hard to believe that you would not include this in the design if it needed it.

The Sierra-1 has plenty of baffle step compensation, about the same level of compensation as our 340 SE (both use 7.5" wide front baffles) I would be most curious how someone can make that determination by looking at the NRC graphs? Please PM me and let me know who this was.


I remember you discussed the NRC measurements once before. They measure in an anechoic chamber, which doesn't accurately show the effect of the rear mounted port, correct?

100% correct... here is the post: http://forum.ascendacoustics.com/showthread.php?t=3006

In summary, the port output of a rear-ported loudspeaker is almost completely nullified since the mic is only picking up energy that is traveling forward. All other energy is absorbed in the wedges. This is why, when examining the off-axis graphs, the further off-axis (this is done by rotating the speaker on a turntable) the greater the bass output, because the energy produced by the port becomes more in-line with the mic. At on-axis, the port is 180 degrees off-axis, with a 60 degree of-axis measurement, the port is now 120 degrees off-axis. The SoundStage should really work to change this and I have discussed this with Doug Schneider. Problem is, it would not be fair for the speakers that have already been tested. Tough call either way I suppose...

The converse is also true, a front ported speaker will show less bass output at off-axis angles. Look at the Paradigm Studio 20 NRC measurements...


Also they measure with the mic placed in line with the tweeter instead of halfway between the tweeter and the woofer, is that right?

Also correct, unless they are given instructions to do otherwise. If I remember, they did ask me what was the appropriate mic position when the 170 was measured but not with the Sierra-1. Just an oversight on my end.

Here is some info: http://www.soundstagenetwork.com/measurements/test_loudspeakers.htm

[DougMac]

--- I once submitted a final crossover design to the powers-that-be for listening and final approval. It was initially rejected. As an experiment, I re-submitted it a few days later to the same listener with documented changes and it was immediately approved. Funny thing, I changed absolutely nothing and the listening was done by someone with vast experience. This speaker was a huge commercial success....


That story reminds me of my days as a photography student at Art Center College of Design. For the first several semesters, you had to get your print approved by the one of the lab instructors. We'd work hard to get the print just like we wanted it and would then take it to the critique area. Because of human nature, the lab instructor would rarely if ever say: "Looks good, I'll OK it". They'd offer some advice for making it better. We'd take our print back in the darkroom and make another applying the opposite of the suggestions made. We'd then take both prints out, put the latest one up as the first and the original as the "improved" version. The lab instructor would proclaim: "Now, that's more like it! See how much better it is now?"

Since I'm the one who helped spark the discussion, I'll offer some clarification. I mentioned REW because I was working under the assumption that if you go the DIY route, you could tweak your design to fit a particular environment. That's a luxury that speaker manufacturers aren't given. My assumption speaks of my habits. I tend to buy a set of speakers and leave them in the same room for decades!

I'm also reminded of the original large advents. A leading consumer magazine consistantly gave them poor marks because the tested them in an anechoic chamber and pointed out their frequency response was not flat. Henry Kloss pointed out that it was true and it was by design. They were designed around what was considered the typical listening room size of the time.
It was two different philosophies in collision.

Doug