Seas W18E001 Excel

[npdang]

Efficiency was a bit lower than expected, but then again the bass extension was much better. The driver tested was well broken in, having been used in my car for over a year. This driver would work well free-air in the doors. The response is also pretty ruler flat, except for of course that nasty peak at 4.5khz.





No surprises here... nasty cone breakup at 4.5khz.



800hz decay. Perfect.



1200hz decay. Good, but getting worse.



1600hz decay.



2000hz decay. Significantly worse. Looks like ~1600hz is the upper limit. However for car audio use this doesn't really matter since your doors will contribute significantly more energy storage than the speaker.



2500hz decay.



3000hz decay.



4500hz decay. Ouch.



Distortion 50hz. Fair performance.



Distortion 60hz. Fair performance. I'd probably cut the driver off here for car use, although 50hz is certainly doable it is noticeably "dirtier" than 60hz.



Distortion 80hz. Good performance.



Distortion 200hz. Good performance, but I was expecting better.



Distortion 1000hz. Very good performance. Below 1%.



Distortion 2000hz. Again outstanding.



Distortion 3000hz. Not too good. I'd keep the driver below 2000hz, although again for car use this is acceptable.



Distortion at 4000hz.







Xmag = -7mm to +6mm
Xsus = -2.5mm to 7.5mm
Xmax = 2.5mm 1 way

Overall a good driver. A few things to note. Distortion performance and energy performance above 2khz gets bad, BUT still very useable for car audio. If this is going in a door, I'd say go ahead and use it full range without a lowpass or a notch filter. At all other frequencies between 80hz and 2khz, this driver is below 1% distortion and an excellent performer. The frequency response is also ruler flat minus that nasty breakup at 4.5khz.

Efficiency is a bit lower than expected at ~85db, but the bass response is much more ideal for free-air use in the doors. I wouldn't be surprised to see flat response to 20hz for most car doors (properly treated of course).

Xmax 2.5mm??! Well, yes. The suspension has a huge forward bias. The travel outward is much greater than rearward travel. This is to prevent the spider from smashing into the top plate. Keep in mind that the peak to peak travel is about 20mm... it just won't be "linear".

I was also surprised at the BL curve. Not quite as good as I'd imagined. Bl is fairly symmetrical and very good, but not as flat a plateau as others I've tested... in other words not the best. LE variance is outstanding however.

All in all, I'd say it's still a great driver. Low distortion, excellent bass response, and great linearity (except for the suspension).

[tf1216]

Thank you npdang!

It is probably too late to ask for distortion measurements of your nextel driver seeing how it is in your car already. Could you please make some comments between the midrange capabilities of the two drivers, W18E vs. W18NX?

Is the nextel cone as hard as the magnesium?

[npdang]

Nextel is fuzzy I have the data for the W18NX, just haven't posted it yet.

[bbfoto]

Hey Nguyen!

For door-mounted mids/mid-bass, which would driver would be best matched to the LCY110's overall, the W18NX or the W18E?

Just received my LCY110's and caps from you as well! THANKS! Can't wait to give them a listen. If they don't work out in the car (highly unlikely,) I'll just build a nice home setup. I'm actually itching to drop the W18/LCY combo into nOrh 9.0 solid marble enclosures for a listen, ha!

I'll shoot you an email as well. Thanks again!

[Nothingness]

Originally Posted by bbfoto

Hey Nguyen!

For door-mounted mids/mid-bass, which would driver would be best matched to the LCY110's overall, the W18NX or the W18E?

Just received my LCY110's and caps from you as well! THANKS! Can't wait to give them a listen. If they don't work out in the car (highly unlikely,) I'll just build a nice home setup. I'm actually itching to drop the W18/LCY combo into nOrh 9.0 solid marble enclosures for a listen, ha!

I'll shoot you an email as well. Thanks again!

Better yet...drop the W15CH001 in with the LCY for the home setup...

[capslock]

How's your German today? I found this on a www.audioavid.de in a thread that linked to these results here. The poster prefers to stay anonymous, but there are only a few Klippelusers in Germany, and few have the experience and eloquence of this one, so I have a hunch who it might be. Anyway, for those who don't want to rely on Bablefish, I have provided a translation (but it being late now, I will probably start looking for typos tomorrow):

... ein Klippel-System zu besitzen, bedeutet nicht zwangsläufig auch zu wissen, was es macht.

... owning a Klippel system does not mean knowing what it does.

Das Kippelsystem ist nicht nicht ein Meßgerät im klassischen Sinn, sondern ein Controller, der versucht ein Chassis durch lernen zu vestehen, damit es in ein festgelegtes mathematisches Model passt. Eigentlich ist das Gerät ein Controller, der das Chassis korrigieren soll/kann. Die Messungen fallen dabei halt so ab.

The Klippel is not a measurement instrument per se. It is more like a controller that tries to understand the driver by trial and error. So in a sense the system is a controller that should / will correct the driver's shortcomings. The measurements are more of a by-product.

Die Ergebnisse hängen in großem Maße davon ab, wie lange man dem System gibt, die Identifikation hin zu bekommen. Default sind 5 Minuten Lernphase, aber das ist zu kurz.
Bei schwierigen Chassis sind es auch gerne mal 2 Stunden.

The results dep3end largely on how long the Klippel can try to learn about the driver. The default setting is a learning time of five minutes but that is too short. A difficult driver can easily take two hours.

Die Messungen zeigen immer 2 Kurven - schwarz und bunt. Die schwarze zeigt, welcher Bereich sicher identifiziert worden ist, die bunte ist eine Vorhersage. Vorhersage deshalb, weil das Chassis nicht oft genug weit genug ausgelenkt wurde um einwandfrei zu identifizieren. Erst bei längeren Messungen bekommt man mehr Sicherheit.

The plots are always subdivided into black and colored regions. The black region is the region that has been fitted reliably, the colored is a prediction or extrapolation. It is not as reliable because either insufficient excursion was allowed or the system was not allowed enough time to learn and adapt the parameters.

Außerdem verstehen die meisten Leute nicht, was die Steifigkeitsmessungen sagen. Sie gelten nur für die Konditionen, die bei den Messungen festgelegt wurden. Begrenzt man die Messung auf ein Limit von 35% der ursprünglichen Steifikeit, dann zeigt das System halt den Steifigkeitsverlauf bei diesen Bedingungen.

Also many people fail to interpret the compliance results correctly. They are only valid for the conditions that were used in the setup of the test. If the limit was set at 35% of original stiffness, then the system will show compliance vs. excursion for this kind of a drive level.

Die Steifigkeit im Bereich der Nullposition ist dabei nur gültig bei diesen Bedingungen. Das heißt, dass bei den hohen Auslenkungen die Zentrierung gedehnt wird und für eine bestimmte Zeit weich bleibt. Kleine Unsymmetrien sind völlig harmlos und teilweise auf das nicht perfekte mathematische Modell zurückzuführen.
Würde man die Messung bei 60% begrenzen, sähe die Kurve viel besser aus. Der Unterschied zwischen Steifigkeit am Maximum der Auslenkung und in der Nulllage wäre wesentlich geringer.

The stiffness around zero is only valid for the conditions of the measurement because high excursions will stretch the suspension, and it takes a certain time to back to original conditions (suspension creep). Also, small asymmetries can be attributed to imperfections of the mathematical model. If stiffness decrease is limited to 60% or original, the curves will look a lot better.

Begrenzt man die Messung auf wenige Watt, denn ist die Steifigkeitskurve wesentlich glatter und natürlich liegt bei kleinen Leistungen die weichste Stelle immer in der Nähe der Nulllage.

If the drive level is limited to a few Watts, the stiffness curve becomes a lot flatter and is centered close to zero excursion.

Um das Verhalten der Zentrierung genau zu bestimmen, muss man halt 3-5 Messungen bei unterschiedlichen Randbedingungen machen.

To really characterize the behavior of the suspension, 3 to 5 measurements with different drive levels are needed.

Das SEAS-Chassis zeigt im Bereich der BL-Kurve eine leichte Unsymmetrie, die man leicht mit einer anderen Polkerngeometrie hin bekommen würde, wobei ich es erst mal länger messen würde, um sicher zu gehen, dass die Messung stimmt. Da das Chassis neuer ist, glaube ich eher an zu kurze Messdauer.

The Seas driver (he's probably referring to the W18E) exhibits a slight BL asymmetry which could probably be improved with a different pole piece geometry, but personally I would first try a longer measurement to go sure. Actually, this driver being a pretty recent development, I suspect the apparent asymmetry is really due to too short a measurement time.

Bei der Steifigkeit sieht man nur, dass dieses Chassis unsymmetrisch klippt. Das sagt nix über das Verhalten bei kleineren Auslenkungen aus , wobei ich auch hier erst mal die Messung in Frage stelle, weil die Lerndauer unbekannt ist. Sollte die Messung stimmen, müsste man das näher analysieren und auch mal die Sicke weg nehmen, um zu sehen, wer denn da unsymmetrisch klippt. Auch eine Messung des mechanischen Offsets bei hohen Spannungen würde hier helfen, denn dabei sieht man in Sekunden, ob das Chassis stabil ist oder nicht.

From the stiffness curve shown here the only thing one can learn is that the mechanical clipping of the driver is assymetrical, and even this is questionable until a longer learning / measurement time has been tried. If the result remains the same, one could remove large parts of the surround to see if the spider is asymmetrical (translator's note: knowing it is a flat spider I doubt that). Also, a measurement of the mechanical offset at high drive voltage could be revealing.

Leider sind ein paar hingeworfene Kurven in einem Forum spektaklulär, aber sie sagen nix aus, wenn man nicht weiß, wieviel der Nutzer des Systems weiß oder nicht.

Unfortunately, a few curves posted randomly in a forum like that seem instructive but are not unless one knows how much experienced the user of the system is.
Gruß

Klippeluser

[npdang]

Good info! Will have to keep this in mind for the next driver. A typical LSI test with Klippel usually takes from 30 mins to an hour... I would never have guessed that was too short of a measurement time!

The comment about suspension creep explains alot for me. I would love to hear more from "Klippeluser" regarding the Klippel testing and what he feels would be more useful.

[capslock]

I am glad you find his comments useful. I translated verbatim, but in going along I was afraid he was maybe a bit harsh. The default of 5 min seems to be definitely too short. Thirty to sixty minutes seems to be closer to the two hours he says a difficult driver may require.

I am not sure I understand his comments about the suspension creep completely. This issue is also adressed in some of Klippel's papers, but those models were quite complicated. Anyway, it might be worth testing one driver at a few different drive levels or stiffness degradation limits.



I will post at audioavid that you would like to hear more.

[capslock]

Cleaned up a few typos:


... owning a Klippel system does not mean knowing what it does.

The Klippel is not a measurement instrument per se. It is more like a controller that tries to understand the driver by trial and error. So in a sense the system is a controller that should / will correct the driver's shortcomings. The measurements are more of a by-product.

The results depend largely on how long the Klippel can try to learn about the driver. The default setting is a learning time of five minutes but that is too short. A difficult driver can easily take two hours.

The plots are always subdivided into black and colored regions. The black region is the region that has been fitted reliably, the colored is a prediction or extrapolation. It is not as reliable because either insufficient excursion was allowed or the system was not allowed enough time to learn and adapt the parameters.

Also many people fail to interpret the compliance results correctly. They are only valid for the conditions that were used in the setup of the test. If the limit was set at 35% of original stiffness, then the system will show compliance vs. excursion for this kind of a drive level.


The stiffness around zero is only valid for the conditions of the measurement because high excursions will stretch the suspension, and it takes a certain time to back to original conditions (suspension creep). Also, small asymmetries can be attributed to imperfections of the mathematical model. If stiffness decrease is limited to 60% or original, the curves will look a lot better.

If the drive level is limited to a few Watts, the stiffness curve becomes a lot flatter and is centered close to zero excursion.


To really characterize the behavior of the suspension, 3 to 5 measurements with different drive levels are needed.

The Seas driver (he's probably referring to the W18E) exhibits a slight BL asymmetry which could probably be improved with a different pole piece geometry, but personally I would first try a longer measurement to go sure. Actually, this driver being a pretty recent development, I suspect the apparent asymmetry is really due to too short a measurement time.


From the stiffness curve shown here the only thing one can learn is that the mechanical clipping of the driver is assymetrical, and even this is questionable until a longer learning / measurement time has been tried. If the result remains the same, one could remove large parts of the surround to see if the spider is asymmetrical (translator's note: knowing it is a flat spider I doubt that). Also, a measurement of the mechanical offset at high drive voltage could be revealing.


Unfortunately, a few curves posted randomly in a forum like that seem instructive but are not unless one knows how experienced the user of the system is.


Regards

Klippeluser

[Jim85IROC]

At first I thought maybe this commentary was coming from the person who developed the Klippel, but I went through my latest issue of Voice Coil and realized that the developer's name is Wolfgang Klippel, now Klippeluser. Apparently, "Klippeluser" might be just that... a Klippel User, not Wolfgang Klippel.

But maybe you guys already knew that.

[tf1216]

Haha, now worries. It took me a few minutes to catch that too!!

[Davey]

npdang,

I don't own Praxix, but I'm wondering why your 1/10 modulated non-linear distortion plots (on this and other drivers) don't show much more resolution and dynamic range. It seems the distortion products above the fundamental are quickly disappearing into the "noise" of the measurment. Would this be the way the measurement is windowed or some other aspect of your testing methodology?

As a comparison here's some measurements I made using Spectraplus on a Seas CA18RLY driver using a carrier modulated 100% with 1/10 the frequency:

http://home.comcast.net/~dreite/Temp/ca18dist.jpg

http://home.comcast.net/~dreite/Temp/ca18100.jpg

Cheers,

Davey.

[npdang]

It's really my own laziness... and the fact that there aren't too many complaints about it so I just leave well enough alone. I will endeavor to do better in the future

The spl's for the distortion plot may also be off a bit above 1khz because the 1m spl's are extrapolated from the nearfield.

[illnastyimpreza]

ok I find myself feeling less smarter and less smarter every time I come here...

How do you read those graphs???

I know that you basicly what a "flat responce" or something correct???

what do they mean?

[jay]

Originally Posted by illnastyimpreza

ok I find myself feeling less smarter and less smarter every time I come here...

How do you read those graphs???

I know that you basicly what a "flat responce" or something correct???

what do they mean?

How to interpret the data *READ ME FIRST*

there's a reason it's a sticky that says *READ ME FIRST*