Mag cone experiment

jhenderson010759 · « **on:** 2 Aug 2005, 05:32 am »

I've read a lot of conjecture on the virtues of drivers such as the Seas in conjunction with extremely sharp XO filters. So, I just ordered some Seas 18-E001's to replace the ScanSpeak 8545s in my NC Rhythm satellites.

Though the DEQX already provides nearly perfect nearfield response with the SS drivers, I'm hoping that the Seas' will result in lower distortion and improved transparency. This should provide a little more latitude in my crossover point. Possibly I can cross over to my (LCY130) tweeters a little higher - perhaps at 3Khz (as opposed to the 1.8 to 2 Khz currently). Anyone else tried this? Your thoughts?

Also, I use a pair of NC Poseidon subs, which use the Peerless 10" XLS. Has anyone experimented with crossing over to satellites at a relatively higher frequency - say 100 to 120 Hz (instead of 60..80), in an effort to mitigate distortion in the mid-woofers?

TIA,

Mudjock · « **Reply #1 on:** 2 Aug 2005, 12:51 pm »

I think you will find the extra transparency you are looking for with the W18's. I've never heard of anyone crossing them as high as 3 kHz, but have heard them crossed close to 2.5 kHz. The big resonance that needs to be avoided is around 4.3 kHz. Dave Ellis has some comments on the W18e001 vs. the Scan 8545 on his website www.ellisaudio.com.

WerTicus · « **Reply #2 on:** 2 Aug 2005, 01:07 pm »

wouldnt the driver be slow at 3khz vs the tweeter? I bet it sounds worse if you do it that way.... Then again maybe thats well within that particular drivers limits im not familiar with it.

JoshK · « **Reply #3 on:** 2 Aug 2005, 03:17 pm »

3khz is likely way too high for this driver, DEQX or not. I think you have gone a ways beyond the pistonic region of the driver at 3khz, so you will engage some resonance regardless of slope. I think you may want to take a look at North Creek's Pegasus kit and read up on HTGuide.com for some hints on the best choice of xo point with this driver, using steep slopes or not.

Cheers,

jhenderson010759 · « **Reply #4 on:** 2 Aug 2005, 06:01 pm »

Thanks for the feedback. It seems that most of the other designs using this driver cross at 2.5 Khz or lower which is the Seas recommended upper passband. However, I suspect that this was done because of passive XO constraints, since the frequency-response of the driver looks smooth until about 3.3 KHz or so. If I follow this guideline and set the XO to 2.5 kHz @ 100 dB/octave, I'll be down ~65 dB with an XO at the driver resonance.

What about the low-end, with my subs? Since I use them in an array, centered on the front wall close to my satellites, do you believe that I can get away with crossing them at 100 to 120 Hz in order to mitigate the low frequency distortion from the Seas? Looking at this (http://www.madisound.com/pdf/seas/e018.pdf) distortion plot, it seems that the Seas should not be used below 120 Hz or so, if at all possible.

My subs integrate nicely now, with the SS drivers XO'd at 75 Hz. But after studying the distortion plots and decay plots for the SS and Seas drivers on the Linkwitz site, coupled with the fact that he crosses at 120 Hz to the mid-driver in the Orion, I am contemplating the trade-offs. Of course, the Orion bass drivers aren't in a box...

Problem is, I can't find any distortion figures for the Peerless 10" XLS drivers. Anyone know where those can be found? I guess I should measure them...

JoshK · « **Reply #5 on:** 2 Aug 2005, 06:27 pm »

I think 120hz should be fine on the low end.

Jon Marsh points out that the driver resonance stated on SEAS's site & measurements is the major resonance, but that there is a minor resonance at the first knee which is much lower down (1.8khz IIRC) and that should determine the xo point since this is when the driver become non-pistonic and distortion rises. Monte Kay (mfk-projects) makes a bunch of measurements and he has the w18 and the peerless 10" xls iirc.

Linkwitz is limited on the upper end of the low pass by the cabinet resonance not the peerless driver. You will probably also want to avoid cabinet resonance but I suspect it is higher in a ported or sealed box than in an H-frame.

Rick Craig · « **Reply #6 on:** 2 Aug 2005, 07:29 pm »

Quote from: jhenderson010759

Thanks for the feedback. It seems that most of the other designs using this driver cross at 2.5 Khz or lower which is the Seas recommended upper passband. However, I suspect that this was done because of passive XO constraints, since the frequency-response of the driver looks smooth until about 3.3 KHz or so. If I follow this guideline and set the XO to 2.5 kHz @ 100 dB/octave, I'll be down ~65 dB with an XO at the driver resonance.

What about the low-end, with my subs? Since I use them in an array ...

While the W18 has received good reviews in speakers where it's crossed around 2.5K it will sound better if crossed lower. I've tried it both ways and always preferred the lower crossover points (2K and below).

jhenderson010759 · « **Reply #7 on:** 2 Aug 2005, 09:31 pm »

Previously, I have adjusted the XO between the SS and LCY drivers between 1.5 and 2.2 kHz, but have never found a particular setting to be leaps and bounds better than another. Guess I need to listen longer and/or more carefully. Or perhaps the SS drivers are more forgiving, and it just doesn't matter.

The impetus for raising the XO to 3KHz was to avoid a driver transition within the "critical" 1-3K region, a band in which human perception is reported most critical. However, the consensus view, with respect to the Seas W18, appears to be that it is more important to avoid the drivers first resonance than to avoid an XO in the 1-3K frequency range. Since the DEQX stiches drivers together very well within the XO region, this seems like the appropriate trade-off.

Should be fun - I just hope it makes some sort of positive, discernable difference. I read a summary written by John K, in which he compares the CSD of the Seas against the SS drivers, in a restricted frequency band. He illustrated that their spectral decay patterns were similar when appropriately band-limited. Since the DEQX is already aggresively band-limiting the signal sent to the ScanSpeak mid-woofer, I wonder why replacing it with the Seas will change the sonic signature. After all, if the SS is capable of ruler flat FR from 100 to 2KHz, and I aggresively band-limit its input within this range, what can the Seas do differently within this range?

I have a hard time accepting the "improved speed/transient response argument", since a transient (by definition) contains higher-frequency spectral elements, that are rejected by the DEQXs passband filter for the mid-bass. Consequently, for all intents and purposes, there simply isn't any frequency content outside of the 100-2k band.

Why should the Seas sound better? Or even different, for that matter? Aside from lower distortion, I don't know.

John Ashman · « **Reply #8 on:** 2 Aug 2005, 10:43 pm »

Jim, a few things:

1. NHT Xd is using 110dB/octave at 2kHz with their version of the W15 and 48dB/octave at 120Hz, so you can use that as a guideline, since they've been working this system for years to perfect it. They might have done higher order crossovers, except for the added time delay which could be unacceptable with video material.

2. Since magnesium is so light and rigid, you get lower inband distortion (at the expense of that ringing). The sense of "speed" or "transient" improvements would be because of the lowered distortion. Much of what gives this sense of speed/transient is in the midrange, especially for lower octave instruments. So, the tweeter is more adding "ambiance" and upper harmonics except on the upper instruments. Think of where the detail comes in a bass guitar, for instance. Neither from the woofer or the tweeter.

Rick Craig · « **Reply #9 on:** 2 Aug 2005, 11:19 pm »

Quote from: jhenderson010759

Previously, I have adjusted the XO between the SS and LCY drivers between 1.5 and 2.2 kHz, but have never found a particular setting to be leaps and bounds better than another. Guess I need to listen longer and/or more carefully. Or perhaps the SS drivers are more forgiving, and it just doesn't matter.

The impetus for raising the XO to 3KHz was to avoid a driver transition within the "critical" 1-3K region, a band in which human perception is reported most critical. However, the consensus view, with ...

The W18 will be fine crossed at 60-80hz. Keep in mind that most of these non-linear distortion tests are running the drivers full-range. Once you place a high-pass filter on it the W18 will have no problem handling the 60-80hz range.

jhenderson010759 · « **Reply #10 on:** 3 Aug 2005, 03:08 am »

Rick, how would the portion of the distortion curve in the Seas data sheet plot which increases linearly at frequencies below 130 Hz affected by a high-pass filter? It seems that you're implying that there is some modal coupling between the upper-frequency breakup modes and this distortion increase at the low frequencies.

Nothing in the distortion plot seems indicative that this linear distortion increase is an intermodulation effect from any of the numerous modes that appear at frequencies beyond 1.8 kHz.

Rick Craig · « **Reply #11 on:** 3 Aug 2005, 04:20 am »

Quote from: jhenderson010759

Rick, how would the portion of the distortion curve in the Seas data sheet plot which increases linearly at frequencies below 130 Hz affected by a high-pass filter? It seems that you're implying that there is some modal coupling between the upper-frequency breakup modes and this distortion increase at the low frequencies.

Nothing in the distortion plot seems indicative that this linear distortion increase is an intermodulation effect from any of the numerous modes that appear at frequencies beyond 1.8 kHz.

You misunderstood me. What I meant to say was that increasing the crossover point to the sub from 80-120hz really isn't needed. The non-linear performance will be very good with the right size enclosure / tuning crossed @ 80hz to the subwoofer. I also think large excursion subwoofers operate better crossed at 80hz or below. The Orion uses 120hz because of the dipole operation contraints placed on the 8" driver.

The 2nd/3rd harmonic curves as posted on the Seas site will be affected by drive level and the size of the enclosure that was used. They use the Klippel analyzer so if there are any distortion problems with a driver they would know it.

Since you have the DEQX you may be interested in a dipole array that's coming in a few months

jhenderson010759 · « **Reply #12 on:** 3 Aug 2005, 05:16 am »

I am definitely going to try a dipole very soon. I have "almost" ordered the Orion kit numerous times. I think that after this experiment with my Rhythms, I'll be ready to try something else.

Many thanks to all who responded within this thread. You've been very informative and helpful.

JoshK · « **Reply #13 on:** 4 Aug 2005, 03:38 pm »

Here we go...the post with re: to the W18 and implementation that sums it up succintly and better than anywhere else I have come across.

Jon Marsh's comments on various implementations of the Excel W18 driver

Short answer: 1400hz = best xo point, 1800hz = bang for buck xo point, 2400hz = popular but suboptimal point.

jhenderson010759 · « **Reply #14 on:** 4 Aug 2005, 05:38 pm »

One of Jon's questions is a perfect segue into the use of the DEQX for this driver:

Quote

In principle, to fully correct these in the amplitude and time domain will work IF the driver is truly minimum phase by itself, and IF we compensate each one of these issues individually. How many tuned notch circuits are you willing to implment in your passive or active crossover?

The answer - up to 4096 "circuits" (since that is the frequency bin resolution of the DEQX FIR filter implementation).

One need look no further than the Seas distortion plots to see why an XO of 1400Hz is considered optimal, since that's the frequency corresponding to the first significant breakup mode - reaching nearly 1% at that frequency. But, if we're using distortion as a primary criteria through which we select the XO frequency (and I am), then the distortion characteristics of the tweeter must also be examined, since all drivers exhibit tremendously increased distortion at the lower bound of their operating range, particularly approaching their natural resonance frequency.

In my first batch of experiments, I'll attempt to configure the XO to yield minimal systemic distortion. I'll implement a series of filters with driver transitions in steps of .2 kHz, ranging from ~1.4 to 2.2 Khz, and use SoundEasy to take distortion measurements. I'll live with whichever configuration yields the lowest distortion at the system level.

Sound reasonable?

John Ashman · « **Reply #15 on:** 4 Aug 2005, 06:49 pm »

You're dead on Jim.

Even with DEQX, it's still a balancing act. It would be nice to get a 16Hz-20+kHz speaker with near perfect dispersion and distortion characteristics, but even with DEQX, you really can't get there. I think you could do it with a 4-way though. I've spoken with Jack at NHT at length and I can tell that they labored over every option and combination til they finally had to make the best compromise possible for a 3-way, 3-driver system. You can overcome some stuff with added drivers, but at the expense of lobing.

They obviously felt that the W15 was the best overall choice for the best blend with the tweeter while balancing dispersion, distortion, other things. But then that pretty well forced them into a 10" woofers and accepting a little lower dynamic range that is possible from a bigger system. So, they pretty well maxed out the midrange/treble dispersion/distortion elements at a slight penalty in deep bass and/or dynamic range. Going for more midrange output means accepting a lower crossover for the treble or letting resonances creep in or dispersion issues. So, it was a balancing act. Too bad, it would be *really* nice to have those kinds of qualities with unlimited bass and dynamics. Because it's needed? Nah, just because

But you could go to multiple drivers, sacrifice a little in dispersion/lobing and wind up with your own "maxell commercial" system

BTW, if I were going to design an optimized DEQX system, but I mean an ass-kicking one, I think I'd actually break down and do dual W15s as mids with the tweeter jammed to the side, between them and then dual 12" woofers opposing each other, front to back or side to side with ~2000Hz and ~100Hz crossovers. OR, I'd do a single W15 with with 4 W22s firing front/rear and just EQ the bass. And then use ~2000Hz and ~200Hz crossovers. On the other hand, there's a part of me that just wants to build a huge ass bass tower with 8 or more 10" or 12" woofers per side

Rick Craig · « **Reply #16 on:** 4 Aug 2005, 08:03 pm »

Quote from: JoshK

Here we go...the post with re: to the W18 and implementation that sums it up succintly and better than anywhere else I have come across.

Jon Marsh's comments on various implementations of the Excel W18 driver

Short answer: 1400hz = best xo point, 1800hz = bang for buck xo point, 2400hz = popular but suboptimal point.

It's nice to see someone else post the same thing that I've been saying for the last few years

The tradeoff is pushing a 1" dome down that far in frequency (small ribbons and 3/4" domes need not apply).

JoshK · « **Reply #17 on:** 4 Aug 2005, 08:07 pm »

I think that is why JM & SL both write off using ribbon tweeters and use some of the more robust 1 1/8" tweeters. You have done a nice solution with the Carnelian, et al by using the morel dome in between to avoid problems with compromises of top end of the midbass and bottom end of the ribbon or Hiquophon.

Rick Craig · « **Reply #18 on:** 4 Aug 2005, 08:09 pm »

Quote from: John Ashman

You're dead on Jim.

Even with DEQX, it's still a balancing act. It would be nice to get a 16Hz-20+kHz speaker with near perfect dispersion and distortion characteristics, but even with DEQX, you really can't get there. I think you could do it with a 4-way though. I've spoken with Jack at NHT at length and I can tell that they labored over every option and combination til they finally had to make the best compromise possible for a 3-way, 3-driver system. You can overcome some stuff with added driver ...

I think NHT made a good choice for a system of that type. Hopefully my DEQX will arrive soon

Rick Craig · « **Reply #19 on:** 4 Aug 2005, 08:13 pm »

Quote from: JoshK

I think that is why JM & SL both write off using ribbon tweeters and use some of the more robust 1 1/8" tweeters. You have done a nice solution with the Carnelian, et al by using the morel dome in between to avoid problems with compromises of top end of the midbass and bottom end of the ribbon or Hiquophon.

The larger ribbons (Aurum Cantus G1 / Fountek Pro5) will cross pretty low provided you don't mind the narrow vertical sweet spot. The G3 and CD2.0 also work well but cross a little higher.