DiyMobileAudio.com Car Stereo Forum banner

41 - 60 of 74 Posts

·
Moderator
Joined
·
11,394 Posts
Gary, let me ask you this about that drop...given the width of the pillars and where these speakers normally end up, could you take the rise from 2.5k and down as some sort of built in baffle step conpensation? I am asking because I am up in the air about BSC in the car.
 

·
Registered
Joined
·
2,298 Posts
Gary, let me ask you this about that drop...given the width of the pillars and where these speakers normally end up, could you take the rise from 2.5k and down as some sort of built in baffle step conpensation? I am asking because I am up in the air about BSC in the car.
Baffle step is really hard to predict in a car, and I don't even plan for it. The most you'll usually see are artifacts due to the "baffle" not being flat. You can measure the area you're working with and come up with a rough idea of where the artifacts will occur.

For example, if your a-pillar sits 2 inches above your door window, you can expect an artifact in the 7khz range. As the a-pillar rolls around to meet with the windshield you can expect artifacts from that as well.

So, regardless of what you do, the artifacts will exist. No way around it in a car, but not likely to be as predominant in the 2-2.5k ranges as it would in the upper octaves. That's why a ragged frequency response up high would only be combated purely by luck when trying to run a fullrange driver fullrange. A smoother response tweeter is much better for that application (again from a purist standpoint). A 2-3db shift in response from a tweeter that has ruler flat frequency response at 7 khz is not nearly as noticeable as a 2-3db shift in response from a fullrange that already has a 3db shift, or rather fluctuations in that area.

Now, in the real world there's really no way to tell how it will really sort itself out, but from the planning stage, it seems going with smoother frequency response from the beginning is the best bet for achieving a flatter response in the end application.

Now, as for this particular driver in an a-pillar install, I say, emphatically, YES!! I would seat the driver into the a-pillar at 60* off-axis. That would put the drivers flat into the pillars and not be nearly as obtrusive as your typical on-axis pillar install. Of course, contingent on using a tweeter and crossing at that magical 2.5khz.

If somebody wants to make a good argument of crossover point at 2.5khz and the inherent issues with vs accepting a more ragged response from 2.5khz and above (and placing the driver on-axis) I'll listen. It's a valid argument, and only answered based on preference.

I'll take the crossover.
 

·
Registered
Joined
·
736 Posts
If somebody wants to make a good argument of crossover point at 2.5khz and the inherent issues with vs accepting a more ragged response from 2.5khz and above (and placing the driver on-axis) I'll listen. It's a valid argument, and only answered based on preference.
I am about argued out (reference previous two days on this thread), so there's not much willpower in my brain or my two fingers to type out a good wide-bandwidth explanation and argument, but there is one, and I definetely have one. That being said, I am elated to see good discussion going on here. Would I cross this driver over at 2,500 Hz? No. But boy do I appreciate your views on the subject. Truly thankful for good, technical discussion.

Maybe later/tomorrow/Friday aftyer a good night's sleep, some marital activities, and a good dinner, not necessarily in that order.

Scott
 

·
Moderator
Joined
·
18,499 Posts
Discussion Starter #45
Now that I have power again and all the tornadoes have passed, I'll try again...

To me, it's not so much as the need or non-need, but the fact that even the most well-received drivers aren't "flat" or necessarily anywhere thereabouts. I'm halfway tempted to rip my pillars apart and measure the scan 12m's just so people can understand the difference between a manufacturer's 10dB scale and my 3dB scale becuase it seems a lot of folks may be glazing the difference over and worrying about something that isn't as extreme in terms of speaker design as one may think.



As far as the data on this sole driver, I was (insert choice word here) to see the distortion levels all stay within the same bounds, relatively. I am not saying it's good or bad by any means. I am just noting that it's something I don't recall seeing before with HD plots such as what Zaph provides. The overall levels are down pretty significantly, IMO, but I still wonder what the fact that each order is stacked closely within the other tells us about the driver sound.

You'll notice, even outside of other testing I've done, that most drivers seem to exhibit a pretty large delta between F1 and F2, then a small delta between F2 and F3, while F4 & F5 are somewhat intertwined. I'm not seeing that here and it's the first thing that made me same "hmph". I tested the driver 3 times because I thought the data was incorrect. I even tested it 8 hours apart (once before work, once after) and had nearly identical results each time (as identical as HD can be).

Scott - or even Jason - do either of you guys know what this can tell us about how the driver may sound? I still have yet to listen to them myself.

Like I said, it's just not something I recall seeing and am curious what benefit/affect it may have on the sound.

- Erin

This may simply be what great distortion measurements looks like.
I just don't recall seeing all order distortions so closely grouped in any other testing such as Zaph's.
 

·
Registered
Joined
·
736 Posts
Scott - or even Jason - do either of you guys know what this can tell us about how the driver may sound? I still have yet to listen to them myself.
The close grouping of low distortion data is indicative of a very well-mannered driver. When viewed as a whole unit, a speaker that exhibits this type of HD data is extremely well damped across the usable bandwidth of the driver, edge modes are being tamed well, spider modes are small or non-existent, BL centering is exceptional, and of course the speaker is being tested within its mechanical limits. How will it sound? It will predominantly sound “uncolored” and perhaps even “neutral.”

A comment about the 92 dB testing. I think people might get slightly confused by precisely how loud 92 dB is from one driver. A driver that has a diaphragm that is well less than three inches in diameter, tested FULL range, no bandwidth limitation. In the company-owned G35, the L3SE’s are playing from 160 Hz with no lowpass, at about 30 degrees off axis, and at volume 52 out of 60 on the DEX-P99RS and a bridged pair of channels from a SE4200, a single L3SE is playing at not much more than 92 dB. Combined with near-field “liveliness” being close to glass and hard accoutrements, step response, and undoubtedly a potential for at least some modal response, a single mid playing at 92-93 dB is REALLY loud. They EASILY keep up with a pair of 18” subs firing into a common chamber, with a bridged SE2300 on each sub (~1,200 watts each). So I’d caution the reader to fully understand precisely how loud a bandwidth limited (aka, a driver with highpass crossover) is at 92 dB, no less a driver tested full-range at 92 dB.
 

·
Moderator
Joined
·
18,499 Posts
Discussion Starter #47 (Edited)
Edit:
Regarding HD test levels, I agree Scott. But, I don't really feel like rehashing this over and over again. I've had waaaay too much personal stuff happen the last few days (see: tornadoes killing hundreds in my area) to argument semantics any longer.
I ask anyone who wants my input on this to see the test input discussion stickied somewhere in this sub forum.
 

·
Moderator
Joined
·
18,499 Posts
Discussion Starter #48
Erin's excellent measurements (pretty much showing worts and all) probably show more than we are used to seeing, and as a result people are going to interpt the data in the wrong
Im glad the purpose of me measuring and posting results in the detail I have (3dB scale, 1/24 smoothing) is appreciated. My goal was to show pretty much everything that most mfgs don't show... Or make it easier to see than their plots do.

I'll take it as a compliment. :)

Oh, as far as my excellent measurements... 31/32. Yeah, baby!


OT, Gary did you decide to hang on to the praxis setup?
 

·
Registered
Joined
·
2,298 Posts
The close grouping of low distortion data is indicative of a very well-mannered driver. When viewed as a whole unit, a speaker that exhibits this type of HD data is extremely well damped across the usable bandwidth of the driver, edge modes are being tamed well, spider modes are small or non-existent, BL centering is exceptional, and of course the speaker is being tested within its mechanical limits. How will it sound? It will predominantly sound “uncolored” and perhaps even “neutral.”
Which is why I advocate NOT running this driver full range. Use what's great about this driver and leave where it starts to degrade to a driver more capable of that extended bandwidth (i.e. tweeters and midbasses).

A comment about the 92 dB testing. I think people might get slightly confused by precisely how loud 92 dB is from one driver. A driver that has a diaphragm that is well less than three inches in diameter, tested FULL range, no bandwidth limitation. In the company-owned G35, the L3SE’s are playing from 160 Hz with no lowpass, at about 30 degrees off axis, and at volume 52 out of 60 on the DEX-P99RS and a bridged pair of channels from a SE4200, a single L3SE is playing at not much more than 92 dB. Combined with near-field “liveliness” being close to glass and hard accoutrements, step response, and undoubtedly a potential for at least some modal response, a single mid playing at 92-93 dB is REALLY loud. They EASILY keep up with a pair of 18” subs firing into a common chamber, with a bridged SE2300 on each sub (~1,200 watts each). So I’d caution the reader to fully understand precisely how loud a bandwidth limited (aka, a driver with highpass crossover) is at 92 dB, no less a driver tested full-range at 92 dB.
Agree with all you said here. Except, not sure if the "amp" setup is necessary. Requiring a ton of power to get to 92 db shows a serious lack of efficiency. Not avoidable though in a driver like this. In fact typical.

Now 92 db in a car is one thing where we're in the nearfield, and where artifacts will present themselves, but remember you're competing with diy home audio drivers as well. So, I look at it as, put it on a front baffle for a home tower and you lose another 4-5 db. Not necessarily the application that these were designed for, but I refuse to make the distinction between home audio and car audio driver. A speaker is a speaker.

So, if combating baffle step, it's perfectly reasonable to test at 96 db to show how the driver will behave on a limited width baffle at 91-92db.

This is a weakness of ALL small fullrange drivers. ALL OF THEM!!! I'm a huge proponent of fullrange drivers in car usage but for the reason that you gain a few extra db over using them in a home audio tower. But, buyer beware. The lack of efficiency is a HUGE drawback, and why I'm sticking to my stance that these are amazing if run limited bandwidth, and that if Scott has been able to get that kind of distortion profile from 300-2500hz, then he could make an industry coup by limiting the overall bandwidth of the driver and design a pure midrange.

Scott's afraid of crossovers though. ;):p:eek:
 

·
Registered
Joined
·
736 Posts
Edit:
Regarding HD test levels, I agree Scott. But, I don't really feel like rehashing this over and over again. I've had waaaay too much personal stuff happen the last few days (see: tornadoes killing hundreds in my area) to argument semantics any longer.
I ask anyone who wants my input on this to see the test input discussion stickied somewhere in this sub forum.
Whoa bro, did I miss something? Nothing I wrote was a personal attack. <confused> Or have there been some thread edits I didn't see?
 

·
Registered
Joined
·
736 Posts
Requiring a ton of power to get to 92 db shows a serious lack of efficiency.
Well, I guess I should temper that comment with the fact that the 4200's gain are "off." As in, not even cracked a pube hair.

If Scott has been able to get that kind of distortion profile from 300-2500hz, then he could make an industry coup by limiting the overall bandwidth of the driver and design a pure midrange.
I can do it. The formula is actually not that different than the existing L3SE formula. And in fact, the L4SE might just be what you're looking for WRT power compression, distortion, and sensitivity.

Scott
 

·
Moderator
Joined
·
18,499 Posts
Discussion Starter #53
Whoa bro, did I miss something? Nothing I wrote was a personal attack. <confused> Or have there been some thread edits I didn't see?
Not at all.

I just posted something about the testing and realized that I was breaking my own rule about keeping the testing out of the thread. I've kept your comment here because it still pertained to the driver itself.

I've just been trying to keep it a point to keep things solely about my testing methods in a separate thread so things don't go OT.

Trying to moderate fairly and need to stay within my own rules.

It's all good.
 

·
Premium Member
Joined
·
6,214 Posts
* Thanks Erin for taking the time to test all those drivers - I now know why many wants to see distorsion plots, waterfall diagram and specs :D
* Thanks to all posters in the thread - made my life easier and now I can interprete and understand data better
* Thanks to Scott for letting Erin test the driver - I know it can be difficult to explain what is seen and how it translate to what is heard. It can be really difficult to deal with DIYers always looking for perfection (or the other way around : imperfections).

Nice driver BTW, now send Erin some L4SE and bring the L4Pro out already ;)

Kelvin
 

·
Moderator
Joined
·
11,394 Posts
Since Kelvin mentioned waterfall plots...you can usually tell how a waterfall plot will look by looking at the FR of the driver. All of the bumps will show as stored energy ridges on a waterfall plot. High Q sharp peaks will show as little mountains of stored energy ridges. Waterfall plots show linear distortion. The flatter the FR, the better the watterfall plot, and the better the linear distortion will be. A waterfall will show more stored energy than a FR graph, but you can start to get ideas as to what a waterfall wil look like by looking at the FR graph of a speaker.

This might be trival to some, but it took me a long time of looking at CSD (waterfall plots) before making the connections. However CSD plots aren't very useful for looking at the lowend stored energy of a speaker. A Wigner-Ville plot would show that data better. But all of these are derived from the FR data of a speaker.

Please think of my "stored energy" as resonances...which is more correct.

Oh and be very leary of looking at published FR data that does not have any labeled smoothing or resolution...the more smoothing or lower the resolution the smoother the FR looks. Tang Band is notorious for having FR data that is too smooth to be believed.
 
41 - 60 of 74 Posts
Top