now auditioning... "3-for-Me"

[charlielaub]

I've put this 3-way project on the back burner for a couple of months while I worked on finishing up and releasing the Active Crossover Designer. Finally put one speaker together today, took some measurements, threw together a crossover design in ACD, and then implemented it in my MiniDSP box. Now it's time to sit back, open a beer, and let the ears check for balance, etc.

So far I'm liking it!

Here are some preliminaries:
DRIVERS:
CSS ERT26
Focal 6VE3251B
NHT npt-11-075-2

BOX:
IKEA "BaffleXchange" concept, 44"H, 12"W, 12"D
subdivided in to upper and lower compartments at about 28" H

Attachments:
crossover version 1 frequency response from ACD
cabinet/speaker

-Charlie

.

[ReissM]

Wow that's a really nice FR measurement. From 25Hz to 20kHz roughly +/- 3dB.

And I'm totally digging the low crossover point for your midrange.

Normally it's not a good idea to separate the woofer and tweeter by this large of a distance. However, you clearly know acoustics and it shows. The woofer and mid should integrate nicely despite the physical separation... because we're talking about 11 foot wavelengths at 100Hz.

Now add in the fact that the floor bounce problems will be minimized in this design due to the physical location of the woofer. Bravo. A low woofer to mid crossover frequency is the key to pulling all this together.

You've even mounted the drivers off-center to address baffle diffraction. And flush mounted the drivers into the baffle, nice touch.

Honestly it's tough to even come up with anything to offer for constructive criticism. This is one heck of a well thought out design. It's like trying to find flaws in a Ferrari F430 Spider. Ok... maybe I could suggest a slightly lower crossover point for your tweeter. So that your 6" diameter mid doesn't have to play up to 3200Hz.

3200Hz is a 4.25" wavelength, so there might be a tiny bit of beaming starting to take place when trying to reproduce a 4.25" wavelength with a 6" diameter driver. Any chance you could try XO at 2200Hz? Compare FR of 3200 Hz XO with 2200Hz XO at 45 degrees off axis and see if there's any measureable difference. As I said earlier... it's like trying to find flaws in a Ferrari.

Nice design, nice measurements... what are you thinking for the finish?

[charlielaub]

Thanks for your post focusing on the positives!

Actually, the M and T drivers are not flush mounted. The M has a truncated flange, and that is a PITA to do for me, so my plan is to build up around the tweeter with a rectangular "plate" added on to the baffle, about the same thickness as both M and T flanges. I haven't gotten around to it at this point. As it is, the diffraction problem is surprisingly mild, so for now I will leave it as is.

The tweeter has a 1.5 kHz resonance, and I'd like to cross it over closer to 2k Hz (currently it's at 3k Hz) but I need to measure the 2k-3k Hz distortion before I feel comfortable moving the crossover point lower. I just whipped up the crossover in about 15 minutes, so this is all very preliminary at this point. I just wanted to get something working with more or less flat response so that I could give it a listen.

As far as finishing goes, I will probably just chamfer the front baffle slightly (e.g. 1/16" around edges) and then spray on black and then a sealer coat. The cabs are made to have baffles taken on and off repeatedly and I will use the same cabs for several different speakers. They already have a 'birch look" finish on the top, bottom, and sides, and the black should match that nicely. It will look a little boring and utilitarian, but I don't have the tools or a workshop space, nor the $$$ to do a real nice proper cabinet.

The crossover point between the sub and midwoofer works pretty well. Unfortunately this particular midwoofer has pretty high VC inductance, and the response rolls off quickly. This is why I want to cross pretty low, and am using 100Hz in this case. The midwoofer's natural closed box response is also rolling off around that point, so I can use it as part of the acoustic crossover. The downside is that the midwoofer has a 6+ dB suckout from floor bounce around 120 Hz due to its height above the floor. I have just EQ'd this out for now. It would be better if I could push the crossover point up to around 200 Hz or slightly higher, but with this particular woofer I don't think that will work out.

I have lots of other drivers to play with, so I can always make another front baffle, drop in some other drivers, and then see what happens. This is the first speaker I have built in this particular size of the baffleXchange cabinets, so I am still getting a feel for what works and what doesn't.

-Charlie

[charlielaub]

OK, I just took distortion measurements using STEPS for each of the three drivers. The results, plotted, are attached in case other people are using these drivers.

So, it looks like I could decrease the M-T crossover point into the low 2000's. I might also increase the W-M crossover point to 150-200 Hz... not sure.

Luckily I can easily try out all of these changes and have a listen.

-Charlie

[charlielaub]

Got a little distracted since my last post... have been finishing up the FRD blender.

I finally put together the second speaker and hooked them up together for a listen. I did move the W-M crossover point to 180 Hz and the M-T crossover point to 2250 Hz. The speakers are close together, but the sound stage is very wide. Interesting. I will have to go back and model this in ACD - I just made the changes on the fly in the MiniDSP.

I hope to post more on this soon. For now, I've attached another pic.

-Charlie

[charlielaub]

I now have a "version 2" crossover in place. I designed it using my Active Crossover Designer and implemented it with a pair of MiniDSP 2x4 boxes. See the attached frequency response plot.

The WMT system crossover includes:
W-M: LR6 @ 180 Hz
M-T: LR4 @ 2250 Hz
first order shelving network for baffle step compensation
Linkwitz Transform of the woofer to Fc=30Hz, Q=0.9
EQ: 6dB boost @ 130Hz, Q=1

The LR6 and LR4 happen to work out pretty nicely for this particular system. I've attached a plot of the relative phase angle between the midwoofer and tweeter. Just by luck, there is nice alignment of phase between 2000 and 3000 Hz, rising more steeply as frequency increases, so the 2250 Hz crossover point works great. Also attached is a FR plot at the "listening position", about 3m away and 40" above the ground.

I also added some delay to the mid so that the "sweet spot" for the system is moved upwards, above the speaker to about 5' high or so. On axis the response is still not bad and I can say that the system sounds pretty darn good no matter where I am in the room!

I really like this driver arrangement, and this confirms the results of diffraction modeling that I have done.

Great to have all these tools to use!

-Charlie


.

[charlielaub]

Now that I seem to have the crossover pretty much finalized, I can think about amplification to plop into the rear. One reason that I chose these particular drivers (apart from their affordability!) was the fact that they were 8 ohm nominal loads. Because of this, I can use dual chipamp modules that will supply about 35 WPC to the M, and T drivers. This is plenty for home use, especially since the listening space where these are located is not large and I don't typically crank up the volume. The chipamps use a Sanyo STK IC that actually has pretty good performance (some of the Sanyo's were pretty crappy) and the module comes with a nice small heatsink, power supply, etc. After a few mods, I should end up with a nice tidy little package.

The NHT sub driver will need a separate amp, and its Re is about 9 ohms IIRC. It's also not super sensitive, so I do need well over 100W of power. I have a pair of Ebay sourced BTL TDA7293 (2 parallel pairs, bridged) amp board that I have wanted to try out. Hopefully these will work as advertised - I didn't test them when I bought them!

Anyway, those are the next steps...

-Charlie

[ReissM]

Very interesting design. Nice write-up too.

Thanks for taking us through the design, and your recent changes.

[charlielaub]

OK, now at "version 3" of the crossover. This time I have implemented 5th and 6th order elliptic filters that I have been tweaking for crossover purposes so that the phase and amplitude sum is more acceptable compared to "standard" elliptic filters. These have relatively steep slopes in the transition band, while having low passband ripple (0.5 dB or less) and only modest group delay peaking. These filters have allowed me to reduce the system response ripples.

The system has an even more expanded soundstage compared to the LR4/LR6 crossover that I was using previously. I will definitely be devoting more time to investigating this type of filter for crossovers.

A plot of the system response from ACD is attached:


-Charlie

[NWLanding]

I like the analytic approach you've brought to this topic. I keep wanting to learn about active crossovers but haven't invested the time. Can you suggest any comprehensive description of designing and implementing an active crossover? I'm fuzzy on how each of the components are located in the chain.

[charlielaub]

One more thing I wanted to add here is that the relative phase angle between the woofer and mid, and mid and tweeter, are nicely close to zero (less than about 30 degrees) in the "crossover region" where the output from both drivers is significant. One nice benefit of the elliptic filters is that the steep slope reduces the range of the crossover region. This helps with off axis responses as well as on axis.

Here's a plot of the relative phase angles for the drivers involved:

In the upper plot, the crossover point is 2250 Hz and you can see that the relative phase angle has a broad minimum near zero degrees near this point. This means that the driver will stay relatively in phase throughout the M-T crosssover region.
In the lower plot, the crossover point is 180 Hz. The phase reaches a maximum of 30 degrees, which is still more or less "in phase", in this region. Again, the drivers will be is phase throughout the W-M crossover region.

The plot illustrates that the phase angles are well controlled while the filters are producing steep attenuation. This is a great result, and may have something to do with the great soundstage that I am experiencing.

-Charlie

[NWLanding]

ICan you suggest any comprehensive description of designing and implementing an active crossover? I'm fuzzy on how each of the components are located in the chain.

I'm answering my own query. Here's a FAQ

If there are other learning resources I'd appreciate hearing about them.

[charlielaub]

I'm answering my own query. Here's a FAQ

If there are other learning resources I'd appreciate hearing about them.

You might want to post this question in the TechTalk forum instead of here in my build thread...

If you start a thread there, I have a couple links that I can post in my reply. That way it gets more visibility based on the thread title and more people will chime in with info.

Thanks,

-Charlie