Extremis 6.8 MMT design help (1 Viewer)

[Seth_L]

Which of these two designs is likely to be sonically better? And yes, I know I'm going to have to do a completely custom crossover for either.

Design 1)


Design 2)


They are both rear ported with a 3" flared port. Box will be 52L tuned to 30Hz.

I would think Design 1 would be the prefered sonic layout because the tweeter is close to both midwoofers. My understanding is that this means that in design 1 both m/w's will begin to comb with the tweeter at the same time around 2290Hz. Design 2 has a one m/w close and one m/w far away, which means they will comb at 2290Hz and 1000Hz respectively.

Am I right, or not understanding things correctly?

 

[Seth_L]

Or, should I use a D'Appolito vertical MTM design with the distance between the MTM picked by the crossover frequency?

 

[Seth_L]

Boy, it's like a ghost town around here...

I downloaded some new software tools. A Baffle Diffraction Simulator ( http://www.pvconsultants.com/audio/d...ownloadbds.htm ), and their Asymmetrical Response Pattern Estimator ( http://www.pvconsultants.com/audio/radiation/arpe.htm ).

I ran some of my design ideas through both programs to see how the baffle size and shape as well as driver placement affected the frequency response and dispersion of the speaker.

Design 2 was just a mess with it's vertical dispertion around the crossover point presumably due to the difference distance between the woofers and the tweeter.

A D'Appolito MTM was pretty smooth vertically around the crossover frequency, but the vertical dispersion got real ugly where the two woofer started to play frequencies that correlated to the distance between them.

Design 1 has a pretty good vertical dispersion with no real problem from what I could see. This design seems to have a dispersion problem on the horizontal axis around 20 degrees off perpendicular on the tweeter side around the crossover frequency though. However a few degrees off the normal to the other side of the speakers with the midwoofer the response was nice a smooth at all frequencies. It looks like if I put the tweeters to the outside with respect to the listener Design 1 is the way to go.

I'll post a few graphs later.

 

[Seth_L]

Here's design 1) with a 3rd order crossover at 2400Hz.

1400Hz:


2800Hz:


10kHz:

 

[DanWiggins]

Seth,

We went with layout 2 for the KIT261 designs. Properly implemented, it's not that bad, really. If you pay attention to lobing, you can point the null you get from the MMT arrangement at the floor, and get quite a smooth vertical response from on-axis to approximately +45 degrees, which covers just about all seated-to-standing positions.

However, I would strongly recommend actually using the phase (time delay) measurements of real drivers in a real cabinet - an error of just an inch in terms of effective acoustic centers is quite a bit when crossing over in the 2+ kHz range.

Dan Wiggins
Adire Audio

 

[Seth_L]

Dan,

I'm not sure I follow you here. I should put the drivers in my baffle and measure the delay of the woofers vs the tweeter from an impulse (aka step response)? Or does it have to be in a box not just a baffle? What is the connection between this and the location of the acoustic centers?

I will be using a digital crossover and bi-amping, so I can add delay to either the woofers (together) or the tweeter independently. I can also change the crossover slopes and centers at will. I'm not sure if this matters or not in context of your statement.

 

[DanWiggins]

Seth,

Yes, you should design around the actual delay as measured. A baffle is fine, as long as you can discriminate the leading edge of each driver cleanly (sometimes not too easy to do just on a baffle).

The acoustic centers are the effective "source" for the driver. It is not an easily located spot! You can't just look at the front of the driver and say "it's here". Note that a 1 inch error at 2 kHz translates to ~50 degrees of phase - getting the acoustic center wrong can significantly affect the crossover integration around the XO point.

Since you have a digital XO, you can dial in the delay - that is a good thing! However, you still need to know it, or you won't get the crossover correct. Also, do you have a good EQ as well? Most of the time baffle step, FR smoothing, voicing, etc. is done in the crossover as well, so using just a straight "4th order LR" crossover only will result in poor performance. To replace a passive XO you'd need the active XO and an equalizer.

Dan Wiggins
Adire Audio

 

[Rob Kramer]

Over on Madisound, Dennis Murphy (of MBOW1 fame) was looking for Extremis/OWII (crossed at 2K) input, and was pretty much told by the Extremis gurus (try HTGuide) that the Extremis would not cross that high. Thier recommendation, based on their measurements and designs, is for a 1.2K crossover - which will not fly with the smaller OWxs.

 

[Seth_L]

Dan,

Ok, now how do I go about determining the acoustic center by measuring the delay of each driver? I have a measurement mic, preamp, and TrueRTA, but I'm not 100% sure what the connection between the delay and the acoustic centers is.

Is it as simple as measuring the phase difference between the drivers at various frequencies and turning that phase difference into a distance (by using wavelength)? Is this a distance in the Z-axis (ie: apparent depth behind the baffle the drive appears to play from because of phase delay)?

I'm using a Behringer DCX2496 http://www.behringer.com/DCX2496/index.cfm?lang=ENG for my crossover, so I can add Equalization as needed to the woofer and tweeter to compensate for the 6dB baffle drop as well as anything else that pops up.

 

[Seth_L]

I would say that Dan is the real Extremis guru. After all, it's his driver... His KIT6.1 design crosses the Extremis at 2.8kHz.

So, What was their explanation why it won't cross above 1.2kHz?

 

[DanWiggins]

Rob,

Had no problem with a higher crossover point - I'm not sure using nearfield measurements is really valid as a way to determine the highest usable frequency response for the speaker anyway. We've had no problems with XO points from 600 Hz to 3 kHz (about all the higher I go, based on dispersion), and the UniSound speaker (which garnered a lot of positive praise at THE Show) was crossed over a LOT higher than 1.2 kHz (different designer of that crossover, Ken Goerres did it).

In general, when someone makes a conclusion based on atypical conditions, it's best not to stretch that conclusion to cover all conditions.

Note too that Dennis concluded it was a definitely worth the price, and definitely usable as a midwoofer in the vein of the SS8545, Seas W18s, Peerless HDSs, etc.

Seth,

You can use the phase difference, as long as it's absolute phase. Do you have the ability to measure an impulse response, or any time-of-flight? What you want to do is find out how much time elapses between a stimulus to the driver and the arrival of the wave at a point in space (typically 3 meters away - about the standard listening distance). If phase was absolutely measured, then you could use it to determine the acoustic center distances.

For the Behringer XO, it looks like a good option! Lots of flexibility - I'd recommend starting with a 36 dB/octave roll-off around 2.4 kHz, and using some EQ to compensate for baffle step (I like 2-3 dB of compensation for most rooms), and the bump on the top end of the OW-IIs, plus the inevitable bumps around the XO point.

Dan Wiggins
Adire Audio

 

[Rob Kramer]

http://www.madisound.com/cgi-bin/dis...gi?read=336641

(Dan is in this discussion also)

 

[DanWiggins]

I'd point out my post:

http://www.madisound.com/cgi-bin/dis...gi?read=336899

Note that Dennis' measurements and our measurements are VERY close. The divergence down below 1 kHz is really from baffle step - our measurement in an infinite baffle, his in a baffle. And note that the scale is an exaggerated 6 dB/division, with a total of 30 dB shown - VERY zoomed in.

Once a person stands back and actually LOOKS at what's measured and HOW it's measured, you'll find we really don't fudge anything. Those who show FRs otherwise are typically looking at something different. Which may or may not be valid (depending upon the application).

I'd recommend paying attention to the actual people who have listened to these things - or a person like Roman Badnarek who has actually completed a design AND listened to them - say.

http://home.twcny.rr.com/cnydga/extremis.html

"The polypropylene Extremis 6.8 woofer definitely has a different character than the wood-pulp Vifa XT18 woofer used in the Asterion design but I wouldn't say that the Extremis is any less detailed."

"The XT18 seems to have more energy in the 200-500Hz region that makes it sound a bit fuller in the midrange than the Extremis but the Extremis has a crispness that gives them a cleaner presentation of vocals. Due to this midrange crispness I chose to go with a little bit extra baffle step compensation to give them a fuller sound without being too strong in the bass."

"Overall the response is pretty balanced as indicated by the response plot and they have a natural sound where there aren't any frequencies that stick out."

"They are inefficient at around 80dB but they can handle a lot of power and the nice thing about them is as you crank up the volume the woofers don't compress like most do so the sound doesn't get shrill but remains balanced and warm."

"As I mentioned before these woofers have remarkable bass extension especially for a 24 liter enclosure. With a sine sweep bass notes can be heard clearly down to 25Hz in my small listening room."

Many people seem to want the Extremis to fail, or are upset that it doesn't measure as smooth as the best midranges out there (of course, all the THD/IMD measurements are down in the 0.1% range - basically inaudible). Of course, they overlook that this is NOT a midrange - it is a MIDWOOFER, designed for very good midrange and true bass extension. Looking at the package as a whole - bass, midbass, midrange - I don't think you'll find another driver that matches up.

Dan Wiggins
Adire Audio

 

[Seth_L]

Dan,

Yes, I can measure phase with ETF 5 and get something like this (this is a quick and dirty measurement I took of one of my current speakers).



So, if I have the phase response of each driver at the same frequency, 2.4kHz for example, and determine there is 104 degrees for the one driver and 62 degrees for the other, then I will know there is a 42 degree phase lag for the woofer at 2.4kHz. 42 degrees at 2.4kHz is .65"

Is this the right idea?

If so, how do I factor in that .65" into the driver placement and the crossover? (or do I dial in a .65" delay with the Behringer unit and forget about it?)

 

[GrahamT]

I can only comment myself on the low frequency performance of the Extremis which has exceeded my expectations.

There may be people out there who want the Extremis to fail, but Jon certainly isn't one of them, and I'm not implying anyone said he was. Just clarifying.

The analyzing of this driver is the same extensive process Jon does for any driver. For example the tweeter we were using did not fare well in the tests and so we might be looking for a more suitable tweeter.

Jon has the ability to crossover VERY low with his unique designs and did the same thing for the HiVi woofers and the Dayton RS midwoofers. He likes low crossover points for a variety of reasons.

Of course speaker design is an art and everyone has their own opinions. This project we are doing is not just to end up with a good pair of speakers, but to have fun with the design process.

 

[DanWiggins]

Graham,

Oh, I wasn't blaming Jon! There's a few others out there, though, that seem to want to read more into the data than what's being shown.

As I've said many times before, making the perfect anything is art; making the perfect set of compromises is engineering. ALL drivers have tradeoffs; in this case, I am more than willing to accept THD levels in the 0.2-0.3% range - which are inaudible with complex signals (numerous studies on the subject), and gain a tremendous amount of LOWER distortion down deep as well as overall usable extension.

I think that, with the advent of low cost and easy-to-use test gear there's a lot of measurements being made in a vacuum of understanding. Just how the measured data relates to actual subjective performance isn't discussed; in fact, much of what is measured bears little if any relevance to actual subjective performance.

For example, in study after study, once the THD or IMD drops below ~1% - or -40 dB relative to the excitation signal - it's not audible with complex tones (music). So why the worry, and comparison of drivers down in the region where it makes no difference?

And if it is dominated by low order harmonics, then you can actually increase it by 10 fold (to 10%, or -20 dB) and it's still not objectionable. Audible, yes, but usually not objectionable. Much like a slightly overdriven tube amp.

What should be considered are dynamic effects, like dynamic offset, instantaneous compression, self-noise, and the like. Those tend to track a lot better with subjectiveness in terms of preference. Even clarity! PA speakers can be woefully inaccurate when looking at on and off axis response, as well as power response, but dynamically they typically do VERY well, and that is why they tend to also rate high in terms of perceived clarity.

Anyway, just wanted to clarify some of the data being posted - near-field measurements are interesting from a "what is the driver doing close-in" type look, but don't really relate to what you hear in usual applications. Comparing THD/IMD levels down in the sub-0.1% range may be an interesting engineering exercise, but should be of essentially zero consideration when looking at the actual real-world usefulness of the driver.

Dan Wiggins
Adire Audio

 

[GrahamT]

I appreciate that Dan. I do not recall seeing people bashing your products and apologize if ANY of my own comments have come across that way. I am a DW crew member and have learned a lot from you.

I just wanted to let people reading this to know that this is Jon's design process for any driver. Nothing malicious towards any company on his part. These are his design opinions. Again, not implying that you were talking about Jon specifically but his impressions were discussed.

When Jon and I started this project months ago, before the drivers were measured, the targeted crossover frequency was 1.6 kHz anyway. Even if the driver behaved perfectly up to 4k, we would still crossover low for a variety of other reasons.

Not to put words in Jon's mouth, he can speak for himself, but I dont think he visits this forum.

Edit: Sorry for the de-rail Seth. I am interested in seeing your design.

 

[DanWiggins]

Seth,

Dial in 0.65" into the crossover - there's ~13560 inches per second, so that would be ~50 microseconds.

Dan Wiggins
Adire Audio

 

[Seth_L]

Ok, thanks for the help Dan.

I am curious though. If I didn't have a digital crossover with the ability to delay the tweeter's signal to align the phase of the two speakers, I would either have to build the front baffle so the woofers are .65" closer to the listener's ear (which could cause an odd baffle response), or find a way to eat that extra, hypothetical, 42 degrees up in the crossover?

 

[DanWiggins]

Eat the phase in the crossover. Crossovers have a delay inherent inside them - when adjusting the crossover for a flat summation, you are essentially adjusting the phase - the delay.

It's not uncommon to have 2" of difference at 2.5 kHz between the woofer and the tweeter...

Dan Wiggins
Adire Audio