[Duke]

Feels to me like we have wandered pretty far off the original intent of the thread, any more and I might look to get some pruning done, to get us back on track.

I am not going to involve myself in the discussion, but I am going to ask that folks take it elsewhere, should they feel compelled to continue.

Thanks!

Fair enough, thanks for letting us duke it out some.

Okay, here's my on-topic $.02:  A horn or waveguide often has on-axis anomalies that disappear off-axis.  Let's look at a round waveguide as an example:  There will be a strong reflection around the perimeter or "lip" of the mouth that will interact with the rest of the sound, like this:  The path length from the throat to the microphone will be different than the path length from the throat to the lip to the microphone, so that lip reflection will arrive slightly delayed.  This delay results in cancellation where the path length difference is equal to one-half wavelength, and reinforcement where it's equal to one wavelength.  With a round horn or waveguide, the effect is most pronounced on-axis because that's where the lip reflection is coherent; off-axis the reflection gets spread out so that its effect is pretty negligible.   This effect is reduced with a rectangular or oval horn or waveguide because even on-axis the reflection isn't totally coherent. 

So if you equalize for the on-axis measurement, your EQ will introduce large off-axis anomalies as it compensates for the on-axis ones.   Perceptually, the off-axis energy is what matters most, unless you listen on-axis in the nearfield or in an anechoic chamber.  So one way to deal with this is to run your EQ with the microphone about 15 degrees off-axis.  This will leave the on-axis-only anomalies out of your EQ, which is what you want.  Then set the speakers up so that your listening position is somewhat off-axis to begin with.

[Tyson]

I tend to make a series of measurements - 0 degrees, 15, 30, 45, 60, 75, 90.  I do this for the tweeter, then I do it again for the midrange.  This allows you to see both power response, and dipole behavior for an OB.  The drop off you will see for the midrange between 45 and 90 degrees tells you where you might want to consider doing a crossover point.  And if you look at the measurements from 0 to 45, you can see where your drivers start to beam and you can attempt to match up their directivity, at the same time you are keeping an eye on total power response.  These are all tricky things and require a good mic and good measurement software.  MTM's have some unique issues to resolve (ie, lobing in the vertical plane), which is why I tend to not use them.  Plus, physically disparate drivers are very audible to me, an MT configuration is bad enough, but an MTM compounds it ever worse!  If you sit directly in the sweet spot, then MTM's are not bad.  Anywhere else, forget it!  Just my $.02.  This type of information should allow you to do your own measurements and create your own design, which is ultimately more rewarding than just copying somebody else...

[JohnR]

Excellent

[Tyson]

In case anyone wants to see what Duke is talking about, here's a set of measurements of my tweeter with a waveguide measured on axis and then 30 degrees off axis.  Look at the differences between 2khz and 4khz in particular:

[poseidonsvoice]

In case anyone wants to see what Duke is talking about, here's a set of measurements of my tweeter with a waveguide measured on axis and then 30 degrees off axis.  Look at the differences between 2khz and 4khz in particular:

Even more excellent, is the measurement to illustrate the point.

Anand.

[Duke]

Tyson's measurement technique is definitely superior to the one I described.  And if a picture is worth 1.0k words, his measurements are worth twice that.

Thank you sir!

[Tyson]

Aw, you guys are making me blush....

To follow up, here's my midrange measured from 0, 30, and 60 degrees. 





As you can see, the 8 inch driver has excellent off axis response, and it also has a nasty set of peaks at 3.5 and 4.2khz.  Given the downward slope after 800hz, I think a crossover point at 1khz with a steep slope is the best option.  BUT, when actually listening, I find that a 1.2khz slope works better because of the shape of the tweeter response.  IMO this tweeter is not happy at 1khz, but quite happy at 1.2khz with a steep slope.  And for the midrange, using a steep slope (like 48db Butterworth), allows the 3.5 and 4.5khz resonances to be avoided.  Other than that, I dial in a small dip at 700hz, and a small boost at 880hz to give me a smooth, flat midrange to tweeter transition.

I also have some bass woofers I bring in at 150hz so I just need a bit of a boost on the midrange between 150hz and 300hz to take care of the baffle step.  The bass woofers sound good at 150hz, but I've heard a rumor that Danny is developing 8 inch servo subs.  If these could be run up to 200hz with even better sound quality, I think it would take the V2's a major step up. 

[Tyson]

Oh, and obviously my speakers are not using Alpha drivers, but I just posted it to illustrate the types of measurements that are useful in deciding where to choose crossover slopes, and when to apply EQ.

Finally - one rule of thumb I've learned - 3db of EQ is the most you should use.  If you have to use more than that, you've screwed up somewhere and need to address it in the acoustic realm.

[JohnR]

I can't say I agree with that.... I think there are lots of reasons to use more than 3 dB, it's no different from a passive crossover design in that regard. Looking at your curve you have around 10 dB lift from 150 to 500, how are you compensating for that with only 3dB?

[Tyson]

Yes, I thought that way too, really I did.  But I find that EQ just sounds "off" period.  It's a balancing act in my experience - the worse your acoustic solutions, and the more you have to use EQ, the more artificial the end sound.  The better the acoustical solution, and the less EQ needed, the more real the system sounds.

Also, the frequency of the EQ matters.  More EQ is tolerated better in the bass than in the mids.  Hell, more EQ in the highs (above 7k) is tolerated better than the mids.

I guess the "rule of thumb" is actually this - in the MIDRANGE, it's better to only use 3db of EQ, otherwise it sounds off.  How's that?

[Tyson]

Ah, one other thing - I've recently treated my room with acoustic panels and it made a larger improvement than all the EQ's and crossovers I've done in the past 6 months.  It's not just the crossover/amp/speaker interaction, it's also the room interaction too.  And the room is probably even more important than all the rest of it.

[JohnR]

Oh OK, we may be largely talking about different things. There are no acoustic solutions to dipole rolloff or driver resonance.