I tell you one thing, when I ran the sweeps for the AR3K, I could CERTAINLY hear a difference even sitting in the captain's chair, compared to the XR3M. I'm talking "hold on... did I connect the wires right" kind of difference.
The impulse response was quite interesting as well...
Mark just sent me a preliminary copy of his PDF. My specs seem to match pretty reasonably. There's always going to be SOME difference in measured data because no two setups are exactly alike. As long as I'm consistent within my own tests, that's all I'm looking for.
I'll try to post up the Vas/Mms/Sensitivity T/S parameters tomorrow. I ran them but posting them up inside very easy...
CVjoint, you actually raised a good point on the other site, so I'm just going to carry over what you've said here as well as my reply, if that's fine with you.
Quote:
Originally Posted by cvjoint
I don't have to be technical do I? haha From what I can tell these drivers will certainly have a signature of their own.
There is a 4-5db peak in the most sensitive area of our hearing 2-3khz. The FR is generally dropping too which accentuates the audibility. Since there is no impedance kink here the problem may be from that big ole motor, or an incorectly sized phase plug or resonance.
On the bottom end the QTS is really high which might explain the low end peaking. Even in IB that Q is way to large for my taste.
The cone break-up seems to be at 12khz for one of the drivers which limits its use to about 6khz.
I hope the Klippel redeems these drivers.
I would advise to be careful of writing off a driver entirely based on the FR. HD results would also be beneficial but I'm not quite set up for that just yet. But, I hope to get the results on Friday.
Also, I would still urge people to listen to these drivers so you can really understand how what you see affects what you hear.
Personally, just going off the XR3M response alone, I would take away these things:
Lowered response from 200-1khz which might prove to be a smoothed response in this passband
The 2khz jump might help to give it some dynamicism
Dip @ 8khz should help from sibilance
HOWEVER, that doesn't mean that what I've said above isn't seen differently by someone else. Prime example: what you posted. We seem to see things differently.
The end user needs to realize that X characteristic may be driven by X change in response and they need to understand WHAT they like. SQ is subjective much more so than we care to admit.
The two drivers certainly seem they would have distinct sounds, though. There are stark differences in some of these passbands that, IME, are very crucial to the overall sound of the driver (not to imply that the entire 20-20k range isn't crucial).
That said, the AR3K response looks really scary on the top end. personally, though, I can't say I wouldn't like it. Though, it seems I wouldn't.
Quote:
Originally Posted by cvjoint
I can make up stories for every drawback present in my equipment, but why would I want to? No matter how you flip this thing its linear distortion is ...well high. There are 7 inch drivers that have a smoother response, and a QTS that is workable. I'm waiting for that Klippel, I expect no less than 7mm of xmax.
Certainly, the linear distortion seems high on an absolute scale. I'm not saying you should or shouldn't buy the drivers. I am saying, however, that unless you've really trained yourself on how certain characteristics affect the sound and just what you like to hear, it's a bit dangerous to presume too much from response graphs alone. Make sense?
You presented negative viewpoints and I presented to you EXACTLY what I thought when I saw the XR3M-LE performance.
As far as "flat FR" goes, I hesitate to really jump all over that. I understand why it's stated as preferred (and have the books/papers to cite if needed). The problem is, we all know that most people typically do not have a flat system FR - at least in car audio. Some of the things the XR3M-LE do with no manipulation (aka: EQ) are things I have going on in my system curve right now.
So, again, it's not a question of what's right or wrong. The data is there for people like you, who apparently know what they want. I just would hate for everyone, especially newcomers, to write off a driver based on the linear distortion alone because it doesn't tell the whole story.
Additionaly, note that the graph is in 3dB steps. What you see is a blown up representation of the data. What you see in the comparison for 0 and 30* is approximately a +3dB/-6dB response up to 10khz which, as far as I've seen, isn't horrible compared to other driver responses. Here's the 12m tech sheet for reference (chosen because it's widely touted as one of the best midrange drivers out). Make sure to compare apples to apples. http://www.scan-speak.dk/datasheet/pdf/12m-4631g00.pdf
Quote:
Originally Posted by cvjoint
Yeah, and tubes sound better because we LOVE 2nd order distortion. Tapes sound good because we simply can't stand a recording that can reproduce 20khz.
Do you have a SOURCE (AES paper or the like) that absolutely quantifies these statements? Serious question. Not found anything myself and no one seems to be able to produce anything that has merit. Typically the question winds up becoming a pissing match.
Last edited by bikinpunk; 04-13-2011 at 01:27 PM..
And to go ahead and comment on the apparent hypocrisy in my post above asking for sources correlating distortion to sound and my posting of what the FR tells me about the sound...
FR can easily be measured by nearly anyone with a calibrated mic and a computer. Make changes to the EQ and listen to the resultant. Easy enough. Distortion, however, is much harder to capture (how many distortion grahs have you seen and how many have you even produced yourself), therefore, much harder to correlate to X sound quality.
So, is there anything you can provide me/us that correlates various orders of distortion with sound? I see these things thrown around a lot but no one seems to have any data on it. Again, if you have sources that can correlate the data I'd LOVE to see it. If you can't provide anything, then just say it. This isn't a chest-puffing contest; we're here for the same goal and I'm here to help others learn.
To be honest, the real reason I'm doing HD plots is for myself; I want to listen to a speaker, write down what I hear, then test it. At some point my goal is to simply be able to look at a grpah and know right away what the speaker will sound like (all environment influences aside). I don't think many people can really do that because not many have the ability to do the testing themselves. So, I'm hoping I can help lighten the load within our community by posting my data and then the end user can listen to the driver in their own time and hopefully get a feel for how the data represents what they hear (though the two test/listening setups will always vary). It's really as close as one can get without buying the software/hardware themselves and doing the tests themselves. Which, I will warn you now is quite a money pit (I'm in $1k right now) and is very time consuming (been working on this since last Summer with a 4 month hiatus).
If you want to get into a pissing match over what I take away from a graph and what you want to take away, then I'm done. TomAto, TomAHto. If you want to discuss the potential for the drivers tested, then I'm game.
I don't have a bias or tie to the MFG here. He is a friend, but I'm a straight shooter, and furthermore can't lie about the data I obtain (well, maybe I could but it would be a TOTAL PITA to do it). What you see in the results is what I obtained and what I've said is what I see from the results. Much like what you see is different, but could still be valid. All future testing will be conducted in the same manner. At the end of the day the person reviewing the figures presented has to know and understand how what they see may affect the sound, but most importantly has to know what they want out of a driver. if they want flat response, then that's what they need to pay attention to. If they like a hot top end, that's what they need to pay attention to.
That's it. But, the key is that the viewer needs to have the ability to listen to the drivers, at least up to a certain point where they feel confident that looking at FR alone tells them what they want to know.
Still, though, as I pointed out previously, the LD of most drivers isn't truly flat. Only when you lower the resolution and make it appear that way, does it seem "flat". And, like I also mentioned, I've seen numerous people say that they do not tune for flat and leave it that way; they add their own touches via EQ.
Defending myself repeatedly will not happen. Everyone had their chance to speak when I posted the "driver testing, need your help" thread. what you get is what you get. I encourage others, if possible, to do their own testing. I'd be happy to help out in any way I can. Otherwise, take my results for what they are; results based on my criteria and test setup. Sometimes what I get will differ from mfg and other enthusiasts' results. But, that's the nature of:
a) not testing in an anechoic room
b) not having the EXACT same setup the other has
We deal with what we've got. I'm making a valiant effort here. If I feel it is not well received, I'll save myself the hassle of defending myself and keep the data to myself. Not that I don't like being asked questions, but discussion and defending are two different sides of a coin.
Oddly enough, had I simply said some audiophile terms all would have been well received. But, once the data is posted in graphical form, all hell breaks loose. But, hopefully we can ride out the bumps and get back to the place where this site was based on data and peer review of subjectivity rather than pure subjectivity (which I've been victim of myself, admittedly).
Hope that explains my position on the above. I'm not going to argue any further points because it's draining. But, I'll be happy to discuss the data.
- Erin
Last edited by bikinpunk; 04-13-2011 at 01:06 PM..
Well now, if you decided to have a full on dispute you certainly cannot make an exit at will. For one I'd like to hear your view on:
-how the 2-3khz sounds to the average listener?
-do we want a 2-3khz peak?
-what does the equal loudness curve you linked to say about how much we like this region?
-do you think the manufacturer wanted this particular frequency response?
My view is simple. We are not singers, or mastering engineers. We simply reproduce sound. We should reproduce it the way it is, which means our equipment should be fingerprintless: 20-20khz flat, no distortion. In particular a 3 inch driver should have a smooth frequency response over the entire vocal range, or at least 200hz-6000hz. It's hard enough to do this in a car when you start with the perfect driver, let alone one with ragged frequency response.
What about all the people that have my view, reference sound? Should they all have to EQ away all the frequency response nonlinearities? What if they already have a peak in their car, 10db at 2khz, and together with the speaker peak it becomes non equable, what then? What if I like a 10db dip at 2khz? With this driver there are very few processors that would even allow you to cut 15db or more. See where I'm going?
Well now, if you decided to have a full on dispute you certainly cannot make an exit at will. For one I'd like to hear your view on:
-how the 2-3khz sounds to the average listener?
-do we want a 2-3khz peak?
-what does the equal loudness curve you linked to say about how much we like this region?
-do you think the manufacturer wanted this particular frequency response?
I'm good with this, man. But, over on the other site, you called my results "simple" and seemed a bit argumentative, rather than wanting to discuss things. My answer for the above is pretty simple, though:
Quote:
Originally Posted by bikinpunk
At the end of the day the person reviewing the figures presented has to know and understand how what they see may affect the sound, but most importantly has to know what they want out of a driver. if they want flat response, then that's what they need to pay attention to. If they like a hot top end, that's what they need to pay attention to.
That's it. But, the key is that the viewer needs to have the ability to listen to the drivers, at least up to a certain point where they feel confident that looking at FR alone tells them what they want to know.
What I stated about those bands were things I took away from my experiences and tastes. But, I absolutely encourage others to listen to as many drivers as they can with readily available data and use that to draw conclusions for themselves. I can't speak for everyone and I won't.
My point wasn't to say that's how it will sound. My point was to show how preference rules what one really wants to buy.
But, since that wasn't received in the manner I expected, I've corrected that portion of the thread. Thank you for calling that to my attention.
Quote:
Originally Posted by cvjoint
Well now, if you decided to have a full on dispute you certainly cannot make an exit at will. For one I'd like to hear your view on:
-how the 2-3khz sounds to the average listener?
-do we want a 2-3khz peak?
-what does the equal loudness curve you linked to say about how much we like this region?
-do you think the manufacturer wanted this particular frequency response?
My view is simple. We are not singers, or mastering engineers. We simply reproduce sound.
Totally agree, dude. What you said above was what YOU prefer and I countered it with things that I've found I prefer in a tune. Again, see bolded above. You didn't see me say ANYTHING subjective about these drivers. I haven't even listened to them yet. Neither one.
However...
Quote:
Originally Posted by cvjoint
We should reproduce it the way it is, which means our equipment should be fingerprintless: 20-20khz flat, no distortion.
Above you mentioned distortion/tubes. So, while it should be the goal to be true to the recording, the simple fact is that, whether we like it or not, sound quality is SUBJECTIVE. I've tried to fight this battle and finally realized that the only real "goal" should be the listener's connection to the music. The enthusiasts that we are, we want to eek out every bit of performance and hopefully that's because we want to understand sound but realistically want to gain the best connection we can. That's what a true audiophile is, IMO, not a person with an endless budget and desire to be the best/most impressive.
My purpose here is to present the data so that the end user can take that and hopefully correlate it with what he or she hears with the goal of further making their listening experience more enjoyable. I'm a total optimist in this regard. I love music. Truly. I think most everyone here does.
Quote:
Originally Posted by cvjoint
In particular a 3 inch driver should have a smooth frequency response over the entire vocal range, or at least 200hz-6000hz. It's hard enough to do this in a car when you start with the perfect driver, let alone one with ragged frequency response.
Well, I agree with this to an extent. The 'extent' being that it harkens back to how sound really is subjective. However, objective data can be used to help further one's goals if they understand it. Pretty simple in concept, but I'm finding it much harder to pull of in reality. Prime Example: I'm still awaiting the subjective words and how they correlate with HD, which you mentioned above.
Furthermore, the response looks very jagged, but again, you need to consider the resolution it's given in. Just like an impedance curve with a 100ohm ceiling, you can't see the little stuff. That's why I blow up the data: I want it all to be seen. I'm not saying this is the best performing driver in terms of linearity. I'm saying to be cautious when doing comparisons so that you are taking into account the dB scale the data is given in. This is HUGELY important and can be easily overlooked.
Quote:
Originally Posted by cvjoint
What about all the people that have my view, reference sound? Should they all have to EQ away all the frequency response nonlinearities? What if they already have a peak in their car, 10db at 2khz, and together with the speaker peak it becomes non equable, what then? What if I like a 10db dip at 2khz? With this driver there are very few processors that would even allow you to cut 15db or more. See where I'm going?
"Your reference sound"... is that a flat FR (which is what you say above is how the music should be)? Or is this your preference? I imagine it's the latter which again just goes back to my discussion about SQ truly being subjective, as much as we like to think it isn't.
The bottom line is that there is no perfect loudspeaker; all will have varying degrees of distortoin. Same thing for music. Preference is KING. I may listen to speaker A with great linear and non-linear results and not like the sound of it over Speaker B with worse L/N-L performance. The viewers of the data posted hopefully have or will obtain an understanding of how the data shapes what they hear. That's really a primary goal and is fundamental to future posts I give.
I appreciate the discourse, though.
Last edited by bikinpunk; 04-13-2011 at 01:34 PM..
So far the graphs Erins has presented correspond to what I heard in my own listening sessions with both speakers.
On axis the AR3k had some harshness to them which is typically in the 2-4k region. it was reduced quite a bit when moving them even slightly off axis.
top end sounded "better" on axis but the harshness was hard to overcome.
I also heard Bartholemy's BMW with the XR and we had a similar discussion and I had him cut around 1.6 and 2k and and 6.3 and 8k.
He PM'd right after Erin posted the results that what i heard was pretty much dead on, with what the response showed.
Bottom line for anyone---look at the graphs--if you understand it, great. if not, read some more, ask questions bc they can be valuable.
But dont put all your stock in the graphs
and certainly dont shoot the messenger.
Erin isnt being paid by anyone to post the info. Hes doin it bc hes a techy-geek guy who loves information and sharing information.
So take the results to mean whatever they mean to you--but dont get personal with Erin.
2010 MECA Extreme World champion,
2005 IASCA World Champion Pro 600+
2010 MECA Mike Baylor Sportsmanship Award
2011 MECA SQL Spirit Award
2012 MECA SQL Judge of The Year
2012 IASCA FINALS SQ/INSTALL Judge
2012 USACi Japan Finals Judge
cvjoint, I understand where the confusion really started and that was completely my fault. I've edited that particular post to make it clear that my point wasn't that this driver WILL sound the ways I posted; rather that the driver shows different characteristics to me than it does you and that goes to show that SQ really is subjective.
I think I'm going to split our conversation off to the 'test build' thread where it seems more fitting. I don't really want to have our conversation stand in the way of the data, and since what we've posted seems to be in regards to general data/correlation, I think it best to be in a separate thread. Thoughts?
The 2khz jump might help to give it some dynamicism
Quote:
Originally Posted by Mic10is
On axis the AR3k had some harshness to them which is typically in the 2-4k region.
One of you sees the peak as introducing dynamicism and one as harshness. From a scientific point of view this is linear distortion and you want none of it. From a subjective point of view, whether you like this type of distortion is not that important. BECAUSE, there are guys on both sides of the camp you would still desire no peak. In fact, given the equal loudness curve this peak will be offensive more often than not.
I like the results and discussion in the same thread so I don't have to open two tabs. My .02
One of you sees the peak as introducing dynamicism and one as harshness. From a scientific point of view this is linear distortion and you want none of it. From a subjective point of view, whether you like this type of distortion is not that important. BECAUSE, there are guys on both sides of the camp you would still desire no peak. In fact, given the equal loudness curve this peak will be offensive more often than not.
I like the results and discussion in the same thread so I don't have to open two tabs. My .02
I agree. Personally, I tune for relatively flat from 200 and up. some minor adjustments for added effect on certain instruments like a snare or something. my ears are particularly sensitive in the 3-5k region. and Having used horns for so long a peak between 1-4k is easy to detect
but again--lets take the info and use if for our own personal knowledge and application. some may like certain things, some may not. lets avoid broad generalizations which could somehow be detrimental to Mark and his company.
I would say this about any company as well.
If its not for you, its not for you. but it may be for others
2010 MECA Extreme World champion,
2005 IASCA World Champion Pro 600+
2010 MECA Mike Baylor Sportsmanship Award
2011 MECA SQL Spirit Award
2012 MECA SQL Judge of The Year
2012 IASCA FINALS SQ/INSTALL Judge
2012 USACi Japan Finals Judge
Mic heard my car and IIRC he didn't care for the 2-4khz range. I heard his car and thought the 2-4khz was lacking. So, what he said above was correct: it's all in how you view the data that counts. I like a bump in that small region (though a cut at 4khz) while he apparently does not.
Only goes toward the point that the data should be taken relatively with consideration of what you like (and that SQ is subjective ).
Ultimately, I hope to ultimately get some feedback from MFG's (not just Mark; future tested drivers' manufacturers, too) regarding why it might be we're seeing what we're seeing.
Someone elsewhere brought up impedance vs. FR testing...
Keep in mind that the two tests were performed differently. Why? Because small signal alalysis is typically done in a free-air (thus the nomenclature: free-air resonance), non-mounted manner, with the driver placed vertically. Of course, I'm early enough in the game to change that regimine (ie: before I start down the road of further testing). I can easily do t/s and impedance sweeps with the driver on a baffle if that is a group decision. Realistically, as long as the methods are consistent then the relative data is good. I certaintly wouldn't say that my data is Gospel (aka: absolute). The work and effort was put into this task to make it as rooted in application as necessary, taking aspects from others' test setups and implementing some of my own. My goal was to stay on the path of what we are used to seeing given that, if someone does slip and try to compare mine to another, at least the methods will be somewhat the same.
but again, I've been seeking input/advice on this topic for a month now. I implore everyone to give their two cents now. It's not too late, but the deadline is QUICKLY approaching.
See the Test Plan thread linked in the OP for more discussion. Heck, here's the link again: http://www.diymobileaudio.com/forum/...-new-post.html
Last edited by bikinpunk; 04-13-2011 at 02:15 PM..
but again--lets take the info and use if for our own personal knowledge and application. some may like certain things, some may not. lets avoid broad generalizations which could somehow be detrimental to Mark and his company.
I would say this about any company as well.
If its not for you, its not for you. but it may be for others
I wasn't the guy saying this particular test result will sound like this (fill in positive or negative characteristic). I didn't even know my words would end up here, I was just quoted on DIYMA. For the most part my take was objective and yes negative. I'm not a heartless Chupacabra, but when it comes to objective test reviews I tell it how I see it. Whoever created this driver didn't make it our of his foreskin. How do they say, business is business, so even businessmen don't care about feelings.
Why should I, a hobbyist, on a test thread, where the scientific approach should be embraced not frowned upon, be deterred from answering to responses to my own largely objective quotes? For the company, for the company's market, because users can't read a frequency response graph?
You asked why we had different inputs on the driver's FR. Mic and I both gave you a reply; rooted in how we perceive sound. When I view data, I'm looking for a number of things but primarily if it has certain characteristics I like. An EQ can do a lot on either end. I'm just of the mindset that if it has X before and I want that X, then that means the driver likely has some sound qualities that I'm looking for. Of course, there's more to the data than that, but that's a top level approach. Nothing to argue there, so I don't know what the last reply is in regards to.
Audible Physics XR3M-LE and AR3K Drivers
Linear Mode
