Budget Duelund Three-way

[gregrueff]

Do you hear anything special about the sound that is noticeably different than your regular speakers?

Gosh that's kind of tough.
But no, not really. They just sound really clear to me and just right.

I'm not good at the hyperbole (although I am from Louisville, Kentucky and have been around Rick Pitino for most of my life) but I think they're just speakers without obvious flaws.
I personally don't subscribe to the notion that some magical crossover is going to be the acoustic equivalent of the Sistine chapel, but they do follow some very well thought out and published research regarding sound reproduction.

The FR is relatively flat, with no controlled directivity. The listening axis is not dictated by any one specific position. This is mostly a function of my lifestyle. I don't listen in one place for long.

The power response is smooth and steadily decreasing with frequency. It follows very closely to what Geddes describes as ideal. That is both a function of rolling off 2nd order, or more precisely adhering to the Duelund slopes but also the diameters of the drivers and their corresponding inherent directivity.

HD is kept relatively low, although I still consider these budget drivers. They work within their respective passbands. This results in a very neutral sound (the best I can describe is grey, like the color) that is noticeably absent of the warm 2nd order tones I'm used to from my RS125P (which I like).

The breakup on the DA215 is around 30 dB down, which is the target presented in Olive's work.

The cabinet construction is one of my best efforts yet to overbrace in an effort to push the cabinet modes and resonances as high as I can to be absorbed by the copious amounts of stuffing employed in both the midrange chamber and bass cabinet.

The midrange chamber is way overstuffed and awesome.

The crossover between the midrange and bass driver was optimized as best I could to minimize intermodulation distortion, both from the bass driver and midrange driver, given that they both have relatively budget motors. You don't want the bass driver playing too high and you don't want the midrange playing too low.

The bass alignment as first built was critically damped and the transient response was ideal. I've now deviated from that slightly to get better integration with the subwoofer.

The bass response rolls off around 100Hz, where it is more ideally reproduced by multiple subs at strategic locations within the room to help balance out room modes.

The aesthetic is something I can keep in my living room that my wife can live with.

And..... I'm sure I'm missing something but that's all I have for now.

Cheers!

[shudson105]

Do you hear anything special about the sound that is noticeably different than your regular speakers? Soundstaging/imaging? I haven't started on my project yet - will definitely be buying the woofers with the stimulus check lol. Not sure what direction I want to go though.....traditional three way, try my hand at this type of project....a MTW arrangement.....a foam waveguide around the mid and tweeter? Time-aligning the drivers just to get the most out of the inexpensive drivers? The possibilities ;-P

[gregrueff]

Listening to music through these, a couple things are clear.

These require a subwoofer for good tonal balance. They sound too thin played by themselves. But that was always the plan anyway.
If anyone were to build these and plan to not use a subwoofer, I would recommend they build it as a floor stander (if space permitted) and vent it to get good bottom end response.
This woofer requires a fairly large volume to go vented which is partly why I went sealed to begin with. On that same note, I think a maximally stuffed enclosure and resultant critical response rolls off too soon to properly integrate with a subwoofer. I think maximally flat allows for better integration.

I've spent a lot of time tweaking the resister values on the midrange and tweeter. I'm not sure that I can say which is best and it probably comes down to personal preference. If anyone were to build these, I would again recommend they pick up a couple extra values to tweak to their own taste. My first iteration had the midrange a little high and the vocals were unbelievably clear, but a little too forward. It'd be perfect as a center channel for a Christopher Nolan movie.

Also, at moderate listening levels I can't get any crossover component even warm. I'm using single electrolytic caps on the woofer shunt and midrange high-pass and single resistors everywhere. I take safety seriously and can remember when someone almost caught their Statements on fire eons ago, but I'm just not seeing it here. I'm trying to think of a way to stress these that doesn't involve running them at party listening levels for hours on end.

Overall I'm very pleased.
I still need to build the final crossover boards and decide how to finish the cabinets, and I'm not sure I'll spend the time anymore on the FR anomaly at 7.5 kHz because I'm not sure I hear anything off.

Cheers!

[rpb]

I've seen enough strange distortion measurements with my speakers, that I don't take them very seriously. It can be maddening trying to figure out why a peak in the distortion suddenly shows up during xo work.

[gregrueff]

Good stuff Dave.
Thanks for your work.

I spent some time this morning seeing if I could tilt the listening lobe the final 3 degrees by tweaking some values.
I was not able to do so, so the crossover between the midrange and tweeter may very well be complete.

I also explored raising the crossover point between the woofer and the midrange.
The system sweep shows that the HD3 exhibits a 10 dB rise from 150 Hz to 500 Hz, but otherwise stays about 60 dB down for most of the range.
I had wondered if this was a result of crossing the TG9 over too low where the shortcomings of its small diameter become apparent.

So here is the baseline (Rev11) measured on the tweeter's axis at 24 inches:


And here is (Rev12) under the same measurement conditions. The crossover is shifted up 100Hz:


Darn near identical.
I suppose I could experiment moving that crossover point even higher, but we're really deviating from the Duelund concept at that point (without recalculating all the slopes from scratch) and it kind of goes against one of the original system goals. I had always wanted to push the crossover between the woofer and the midrange fairly low as a means to try to eliminate as much IMD as possible.

I have only just now set them up to listen to music through them.
I anticipate maybe having to play with the woofer alignment, but then again maybe not.

With the stuffing they are currently pretty darn clos to critically damped (0.5). This may well suit them because I intend to place them in bookshelves. But I am curious to hear the difference with the alignment closer to flat (0.707).

[dlr]

I couldn't figure out a way to overlay the two measurements in the OmniMic program and show the phase of both, so I guess we just have to compare the two side-by-side like this?

I have an improved (UI changes) version of this, haven't uploaded yet, busy with other things, but you can use this version to see magnitude and phase of any number of measurements.

WinGraph

dlr

[rpb]

Have you played them much yet?

[gregrueff]

Just found this post by accident and it looks quite fascinating. I've always had an interest in the exotic and was looking around for something to inspire me on my latest project. I have Zaph's Bargain Mini that I was considering turning into a 3-way using this same DA215 woofer. With the Bargain Mini, it has a 4" woofer and the tiny Aura NT1 tweeter. I am really curious about the baffle/waveguide.....wondering if I should build it out of wood like you did or cut it out of felt or even foam. You did this really cool design, but I see companies like Wilson using felt with a star pattern specifically to breakup cancellations and help with off-axis behavior.

Hey cheers, and welcome to the party!
The Duelund concept looks exotic, and it really thrives for math geeks who like perfectly solved systems (and I'm one of those guys who couldn't for the life of me replicate or produce the polynomials, but that's a different issue..) but in practice just resembles a second-order-acoustic-slope speaker with a lot of overlap between drivers.

Zaph's work taught me a lot and the way he was willing to post to teach others was admirable (and appreciated), but he really trended towards 4th order slopes as in that Bargain Mini. The driver in that system has some decent break-up going on starting at 7 kHz, so you can see why. You might be hard pressed to mold that driver into this concept, but you can still add the bottom end of the DA215 to it if you wanted. Inevitably, someone will advise you that's a risky proposition without a decent plan though.

I will say though, I'm a bit conflicted on the ideal choice for the midrange driver in this type of system.
On the one hand, a 3" with a smaller diameter plays well with the upper integration with the tweeter and the vertical lobe.
But a 4" plays so much better down low and allows for a lower crossover point for approximately equal HD. The MCM 55-1853 in your Bargain Mini system likely has the leg up there.

[shudson105]

Just found this post by accident and it looks quite fascinating. I've always had an interest in the exotic and was looking around for something to inspire me on my latest project. I have Zaph's Bargain Mini that I was considering turning into a 3-way using this same DA215 woofer. With the Bargain Mini, it has a 4" woofer and the tiny Aura NT1 tweeter. I am really curious about the baffle/waveguide.....wondering if I should build it out of wood like you did or cut it out of felt or even foam. You did this really cool design, but I see companies like Wilson using felt with a star pattern specifically to breakup cancellations and help with off-axis behavior.

[gregrueff]

If you are still set up to measure, I'd check off axis to see if anything odd shows up. Also, you can turn the speaker sideways for measurements, and measure above and below (ie vertical stuff), and you should see some reverse nulls there too at xo frequencies.

Excellent suggestion on the measurement technique!

It's way easier to reposition the cabinet on the stand than move the mic.



This is a comparison of 45 deg off-axis down versus 45 deg. off-axis up.
These off-axis angles were chosen because a pretty deep null manifests in both directions that is easy to identify and track.
The fact that these nulls are approximately at the same magnitude, but also shifted in frequency by 500 Hz is consistent with a model that has the acoustic origin of the drivers off by about 3.5mm. (i.e. I can replicate this exact behavior in PCD by playing with the Z Offset)
This sounds like a lot, but that only equates to a 3.16 deg. downward tilt on the vertical lobe (which is what is actually happening).

This is very interesting!
Because while I still consider the phase tracking excellent, it can be improved.
If I modify the tweeter cap from 3.9 uF to 3.3 uF, both the reverse null deepens and the +45 deg & - 45 deg nulls converge in frequency.

I feel like I'm close to splitting hairs at this point though. I can almost shift an entire 1.0 degree just by playing with component tolerance.

Good stuff.

[Wolf]

I meant as in #2. Thanks.
Wolf

[rpb]

If you are still set up to measure, I'd check off axis to see if anything odd shows up. Also, you can turn the speaker sideways for measurements, and measure above and below (ie vertical stuff), and you should see some reverse nulls there too at xo frequencies. They usually show up around 45 degrees or so. No need to post them, but it's a good confirmation that things are as expected. Then, if you hear something you question, you will have more measurements to look back at for possible causes.

[gregrueff]

Oh okay.
This is a new analysis technique to me, so sorry if I'm not understanding.




I couldn't figure out a way to overlay the two measurements in the OmniMic program and show the phase of both, so I guess we just have to compare the two side-by-side like this?

Am I understanding you guys correctly?

[Chris Roemer]

I think Ben means phase on the drivers.

[gregrueff]

You aren't showing the phase plots, Greg. Reverse Null is good, but phase plots would be better.
Wolf

I'm being dense, but do you mean this one:


Or this one: