Jump to content



Sign up for a free account!

Signing up for an account is fast and free. As a member you can join in the conversation, enter contests to win things like this Logitech Harmony Ultimate Remote and you won't get the popup ads that guests get. Click here to create your free account.

Photo

My first DIY WMTMW Design. Help and feedback encouraged.


This topic has been archived. This means that you cannot reply to this topic.
11 replies to this topic

#1 of 12 Phil Olson

Phil Olson

    Stunt Coordinator

  • 54 posts
  • Join Date: Dec 31 1969

Posted October 12 2002 - 04:09 AM

Hello all,

I currently have a Snell HT setup with 5-LCR500's, 2-SUR500's, 2-SUB550's and and a dual Tempest sealed 0.7 sonosub. After reading an article on Dunlavy in Widescreen Review and his supsequent series of articles, I haven't been able to get him out of my head.

His beliefs are that first order, time aligned, phase correct line source systems are the way to go.

Has anyone cloned his SC-IV A? I think that the IV specs are what I would like to shoot for. Perhaps a litle smaller footprint if possible.

Barring a clone, I guess I would start with Speaker workshop but have been also looking at Soundeasy. Is anyone using either of these? My guess is that I will purchase Soundeasy as it's the only complete solution that I've found that will cover the WMTMW design.

I know I'm biting off a big chunk but the best way to learn, (at least for me), is to dive right in and solve what you can and ask questions on what you can't.

I already have an idea of what I want.

35-20K +/- 1.5dB
flat phase
time aligned
line source
95-100 dB
1st order XO's if possible.

Is there anything I am missing? Am I nuts?

If the goals look close to the Dunlavy SC-IV A's, your right!Posted Image

Here's the first candidate:

An all Vifa solution

XT25TG30-04 Tweeter
M10MD-39-08 Mid
P21W0-39-08 Woof

WMTMW First order XO at 300 and 3000. Boxes at 0.57 35-20k @ 100 dB looks possible.

Doable? Dumb? The parts individually are ranked high on the snipets page I was sent to. On first glance they seem to play nice.

The only challenge, (beside the XO of course), is that the tweet and mid will have to be fairly close and a stepped cabinet is a must. The question is how to optimize that.

By the way, I read in "Loud Speaker Design Cookbook" that stepped baffles are prefered over slanted baffels. Supposedly they are equivalent on center but NOT off axis. The stepped maintains better alignment.

I haven't done any simulation yet, (that's next weeks assignment), so this all may fall apart but the overlaps seem adequate and the XO's don't seem out of whack.

Thoughts anyone?

Any comments or suggestions would be greatly appreciated, (except building a kit..I'd rather learn by doing!).

TIA,
Phil

How do you eat an elephant? One bite at a time!

#2 of 12 Guest_Anthony_Gomez_*

Guest_Anthony_Gomez_*
  • Join Date: --

Posted October 12 2002 - 05:32 AM

Quote:
know I'm biting off a big chunk but the best way to learn, (at least for me),


...no....for just about everyone.

I will strongly advise you not to do this. Why,...because the likelyhood of your sucess is very very minumal. 3 ways alone are very hard to get RIGHT, and a 3-way "mtm" is an order f magnitude harder.

I know I am sounding discouraging, but you chose to tackle an exceedingly hard FIRST XO project. I wish you the best of luck on it though!

EDIT: I am not saying NOT to do the project, just don't expect it to sound the greatest....you will learn alot though!

#3 of 12 Isaac C

Isaac C

    Stunt Coordinator

  • 142 posts
  • Join Date: Aug 01 2000

Posted October 12 2002 - 06:51 AM

Have you had a chance to email JohnK? He frequents the MAD board, so you could post the same thread over there (if you haven't already).

The design is certainly possible, but it might take a while to get used to your new speakers Posted Image TP speakers are said to be a bit harsh on the treble side.

Isaac
Subwoofer/Crossover Simulator

Factor Conversion Tool

#4 of 12 Phil Olson

Phil Olson

    Stunt Coordinator

  • 54 posts
  • Join Date: Dec 31 1969

Posted October 12 2002 - 09:45 AM

Anthony,

I thougt I would start with them as a simple MTM first and get that right. I'd then proceed to add the W's. I already have LAUD and Soundeasy will do a complete model of the WMTMW and optimize an XO for me.

I guess what I am really asking is "are these good, compatible speakers for first order XO purposes? I can get started for about $120 for the MTM and see how that goes. I just want to make sure that when I go to add the W's that the MTM and added W's will be compatible.

I know it will be a challenging project but if all else fails I can sell the 3 speakers on e-bay and will have learned a lot in the process for very little investment.

Isaac C,

I haven't contacted JohnK yet. I tried on his web site but the e-mail was invalid for some reason. I'll put this on the MAD board as you suggested.

Phil

P.S. I'm making this a stretch project on purpose. Even if I take a year and never get it right, I'll just be that much smarter.

#5 of 12 Scott Simonian

Scott Simonian

    Screenwriter

  • 1,282 posts
  • Join Date: Jun 20 2001

Posted October 12 2002 - 11:58 AM

Nothin' too wrong with starting with a challenge. Posted Image

When I first came to the HTF I wanted something in your direction but a little different. THe consensous is that 3ways are definitly hard to design. I would try and find someone else who made something that is a 3way and just modify it to your taste. Could take out a few steps out of designing yourself.

Just a thought.

Good luck.

Posted Image Posted Image
Another supporter of 1080p and uncompressed multi-channel sound!

My Twin 18's. 50cuft of box, tuned to 11hz and with 2k watts on tap.

#6 of 12 Greg Monfort

Greg Monfort

    Supporting Actor

  • 884 posts
  • Join Date: May 30 2000

Posted October 12 2002 - 05:17 PM

While I'm a 'dive right in where Eagles dare to tread' kind of guy, based on your post you don't seem to have a clue what you're getting yourself into. Might as well pack a 'chute using instructions written in Sanscrit and go skydiving. Posted Image

That said, I'm a big believer in time alignment for a variety of reasons, but I've never read Dunlavy or heard his speakers, though I've built a bunch of them over the decades.
====
>His beliefs are that first order, time aligned, phase correct line source systems are the way to go.
====
Agreed except that I don't agree that 1st order is necessarily best when the XO point is in the <250Hz BW. Below this point, folks have an increasingly hard time hearing phase shift, so prefer 4th order active for power handling reasons in most cases.
====
>35-20K +/- 1.5dB
flat phase
time aligned
line source
95-100 dB
1st order XO's if possible.
====
Hmm, looking at a pic of the Dunlavy SC-IV A's they are neither line source nor physically time aligned, so must have some pretty elaborate XOs to accomplish the necessary time delay to get flat acoustic phase.
====
>Is there anything I am missing? Am I nuts?
====
IMO, Yes. Dunno, I'm not your shrink. Posted Image
====
>An all Vifa solution

XT25TG30-04 Tweeter
M10MD-39-08 Mid
P21W0-39-08 Woof

>WMTMW First order XO at 300 and 3000. Boxes at 0.57 35-20k @ 100 dB looks possible.

>Doable? Dumb? The parts individually are ranked high on the snipets page I was sent to. On first glance they seem to play nice.
====
Never having heard a Vifa I liked, all I can say is that if the manufacturer considers these to be a tonally balanced set, then you could do worse. If you chose them based on specs and/or a FR plot, then you may wind up with a well mannered speaker that flip/flops tonally. Not the 'Hot Ticket' IMO.
====
>The only challenge, (beside the XO of course), is that the tweet and mid will have to be fairly close and a stepped cabinet is a must. The question is how to optimize that.
====
I got a cure for this, use either a HE single driver such as a Fostex or a horn for this entire BW. Posted Image
====
>By the way, I read in "Loud Speaker Design Cookbook" that stepped baffles are prefered over slanted baffels. Supposedly they are equivalent on center but NOT off axis. The stepped maintains better alignment.
====
Yep. Baffle offset can be estimated to a first approximation by aligning the VCs to get the initial offset, then add in the additional offset due to the XO phase shift, or (speed of sound/crossover frequency)/(360/required phase). Using your 300Hz XO point as an example:

A 1st order XO leads the LF 90deg, so (13,560/300Hz)/(360/90) = 11.3" of additional setback added to the initial setback value. At 3khz we're still looking at 1.13" of additional offset. This should get you within a 1/4WL, which is 'close enough' for most folks.

Since you have LAUD, a more accurate way is to build your 1st order XOs with impedance compensation, etc., then find the total offset by physically moving the HF driver backwards until their acoustic phase matches up.

This way, with well behaved drivers, textbook XOs normally suffice, saving mass quantities of time and limiting the amount of components in the signal's path to something approaching acceptable. Really, take a hard look at series XOs. Since you'll need the drivers to be within ~1/3WL (pi), this equates to a c to c spacing of ~14.38" max/300Hz. At 3kHz, it's 'not gonna happen', so you'll have to give up some of the gains of time alignment if you don't use a single wide BW driver or horn. Posted Image
====
>I haven't done any simulation yet, (that's next weeks assignment), so this all may fall apart but the overlaps seem adequate and the XO's don't seem out of whack.
====
So the drivers in a Q = 0.5 (1st order roll off to Fb) sealed chamber have sufficient BW and a smooth enough HF roll off that they can be EQ'd to have a 1st order roll off? There aren't many off the shelf drivers out there that can meet these goals AFAIK, and the only one's I know of are prosound drivers.

Anyway, along with my 'negativity', hopefully I also gave some useful info in case you pursue this (which I hope you do).

GM

Loud Is Beautiful, If It's Clean

#7 of 12 Dennis XYZ

Dennis XYZ

    Stunt Coordinator

  • 115 posts
  • Join Date: Jun 01 2002

Posted October 13 2002 - 06:11 AM

Quote:
A 1st order XO leads the LF 90deg, so (13,560/300Hz)/(360/90) = 11.3" of additional setback added to the initial setback value. At 3khz we're still looking at 1.13" of additional offset. This should get you within a 1/4WL, which is 'close enough' for most folks.


Oops. First order crossovers require that the acoustic centers be aligned, not offset by the phase difference. The phase quadrature (90 degrees out of phase) is an intrinsic part of the the design that ensures transient-perfect, flat, on-axis amplitude response, as well as flat in-room power response.

Phil, you might want to check out JohnK's B&O filler design. It's a similar 3-way to what you are trying to do but he does a novel crossover to get transient-perfect response. Basically, the woofer and tweeter use 2nd order filters IN PHASE so there is a big notch at the crossover frequency. The mid fills in the notch with 1st order filters an octave above and below the woof-tweet crossover.

#8 of 12 Phil Olson

Phil Olson

    Stunt Coordinator

  • 54 posts
  • Join Date: Dec 31 1969

Posted October 13 2002 - 12:00 PM

Dennis,

I did get a hold of John and he filled me in. It sounds like a great simplification. I think it's definitely doable now, particularly with the optimization tools available these days.

Thanks,
Phil

#9 of 12 Greg Monfort

Greg Monfort

    Supporting Actor

  • 884 posts
  • Join Date: May 30 2000

Posted October 14 2002 - 05:20 AM

>Oops. First order crossovers require that the acoustic centers be aligned, not offset by the phase difference. The phase quadrature (90 degrees out of phase) is an intrinsic part of the the design that ensures transient-perfect, flat, on-axis amplitude response, as well as flat in-room power response.
====
??? I assume you mean WRT JK's hybrid 'filler' design, but for ~phase coherent alignments this is only the initial setback and applies to all XO orders. The XO's phase rotation must be accounted for. With 2nd/4th/8th orders you can just swap the leads of the HF once their acoustic centers are aligned though. With 1st/3rd, etc., orders you can align the acoustic centers and use electronic delay to add in the XO's delay as is common in prosound. Also, there's other considerations that are lumped into the 1/4WL approximation, and why for best performance they should be measured to get the necessary offset, from the listening position if possible.
====
>Phil, you might want to check out JohnK's B&O filler design. It's a similar 3-way to what you are trying to do but he does a novel crossover to get transient-perfect response. Basically, the woofer and tweeter use 2nd order filters IN PHASE so there is a big notch at the crossover frequency. The mid fills in the notch with 1st order filters an octave above and below the woof-tweet crossover.
====
Hmm, my dial up connection is too slow for me to wait for everything to load, but at a glance, a 2nd/1st, 1st/2nd, plus no offset for XO phase won't be time, and therefore, phase coherent. IOW the point in space where a frequency appears to originate should progressively move away from the mike/listening position with increasing frequency just like it does with a single driver in its passband. Now if his speakers do this, then I've learned something new, but I'm highly skeptical.

Anyway, I'm all for transient perfect designs (which traditionally means it's critically damped), but it's just one piece of the puzzle, and without phase coherence there's no hope of even approximately reproducing a complex waveform. Not that physically time aligned ones do a great job except on axis, just that they do it better than ones that aren't, and every little bit helps IMO. Many designers feel the other problems it can cause outweighs their benefits, so as always YMMV.

GM
Loud Is Beautiful, If It's Clean

#10 of 12 Dennis XYZ

Dennis XYZ

    Stunt Coordinator

  • 115 posts
  • Join Date: Jun 01 2002

Posted October 14 2002 - 05:48 AM

Quote:
??? I assume you mean WRT JK's hybrid 'filler' design,
No, I mean traditional first order crossovers. Each filter is only down 3dB at Fc which would sum to a 3dB amplitude hump on axis if they were in phase (-6dB = half the SPL.) Phase quadrature allows them to sum to a flat amplitude. As well, the power sums flat because they are only down 3dB at Fc (-3dB = half the power.)
Quote:
Anyway, I'm all for transient perfect designs (which traditionally means it's critically damped),
It means much more than that the way JohnK uses the term. Short version, it will pass a square wave that looks like a square wave. Traditionally, only the first order Butterworth would do that. The higher order filters look like garbage on a scope. With EQ, it's possible to do it with higher order crossovers and that's where JohnK's experimentation has been focused. The idea is to get a speaker system that will play a square wave with real world drivers. First order filters require drivers with huge passbands.

#11 of 12 Bill Fagal

Bill Fagal

    Stunt Coordinator

  • 166 posts
  • Join Date: Jul 08 2002

Posted October 14 2002 - 11:23 PM

For me at least, a picture is worth a thousand words. Notice the phase relationships of the LP/BP/HP.

I'd like to try these ideas with a coaxial driver--either a W(coaxial)W using the coaxial dynamic as the filler driver or the coaxial by itself using John's 2nd-order + EQ crossover. Seems to me like the best compromise to mimic a FR point source short of trying to pull off a flat 1st-order butterworth.

Bill

#12 of 12 Greg Monfort

Greg Monfort

    Supporting Actor

  • 884 posts
  • Join Date: May 30 2000

Posted October 15 2002 - 08:40 AM

>No, I mean traditional first order crossovers. Each filter is only down 3dB at Fc which would sum to a 3dB amplitude hump on axis if they were in phase (-6dB = half the SPL.) Phase quadrature allows them to sum to a flat amplitude. As well, the power sums flat because they are only down 3dB at Fc (-3dB = half the power.)
====
Right, because they are 90deg out of phase, but time aligned they're not. Now you can argue which is more audible, out of phase flat amplitude or a 3dB/octave slope on either side of the XO point, but due to our shifting hearing acuity it's frequency and amplitude dependent as to how audible it will be. In a monophonic system, the flat amplitude might be preferred, but in a stereo soundfield that is generated by shifting phasing we're much more sensitive to timing since even < millisecond in delay can shift the soundstage enough to be audible.

Regardless, there's no 'oops' here since we're both technically correct, so it's up to the gentle reader to choose their tradeoffs and it seems you/JK and I are 90deg off. Posted Image

Note that for any others who may be reading this and on a learning curve, all textbook filters assume time aligned and resistive in nature, so impedance compensation/Zobels on one or both drivers may be needed.
====
>It means much more than that the way JohnK uses the term.
====
OK, but IMO this isn't a good idea since it can lead to unnecessary confusion.
====
>Short version, it will pass a square wave that looks like a square wave. Traditionally, only the first order Butterworth would do that. The higher order filters look like garbage on a scope. With EQ, it's possible to do it with higher order crossovers and that's where JohnK's experimentation has been focused. The idea is to get a speaker system that will play a square wave with real world drivers.
====
Well, without accounting for all of the time delays in the drivers/XOs/listening position it's going to be an exercise in futility. Still, any inroads he makes is a step closer to accurate reproduction, so more power to him. Posted Image
====
>First order filters require drivers with huge passbands.
====
Or elaborate EQ. Who was it, Theil?, that bragged about having >60 components in their three way, 1st order XO.

GM
Loud Is Beautiful, If It's Clean


Back to Archived Threads 2001-2004



Forum Nav Content I Follow