Interesting DEQX discovery

[Branwell]

In the past, I noticed that I prefer a well implemented passive crossover to the DEQX ( mid to tweeter ).

Never gave any thought as to why, just accepted that in an apples to apples comparison, the passive sounded better.

Had an interesting discovery this evening.

In general, the DEQX will correct the mic’ed FR within predefined limits that are system and user adjustable. While the system will adjust the lower limit depending on various variables, it doesn’t adjust the upper limit, which is set at 20K from the factory.

I was playing with the upper end limits on the DEQX this evening and for grins, set it not to correct the FR above 3K, and guess what, it sounded better uncorrected.

In wondering why, given that many FR variations above 3K are baffle related and change as one moves off axes, correcting them as the DEQX does on one measurement axes, would likely cause issues elsewhere. Given that most passive tweeter crossovers do not EQ much if anything above the crossover region, I wonder if this is the key to the passive sounding better?

Will have to break out the speakers that sounded better on passives and retest.

branwell

[Davey]

I think you answered your own question. Correcting a response so it's flat on a certain axis doesn't necessarily yield an improvement in the overall performance.

The DEQX does its job very well. Unfortunately the job/concept it's designed to perform is misunderstood by many folks. Way too much money for the capabilities IMHO.

My two cents...worth what you paid for 'em.

Davey.

[Evil Twin]

In the past, I noticed that I prefer a well implemented passive crossover to the DEQX ( mid to tweeter ).

Never gave any thought as to why, just accepted that in an apples to apples comparison, the passive sounded better.

Had an interesting discovery this evening.

In general, the DEQX will correct the mic’ed FR within predefined limits that are system and user adjustable. While the system will adjust the lower limit depending on various variables, it doesn’t adjust the upper limit, which is set at 20K from the factory.

I was playing with the upper end limits on the DEQX this evening and for grins, set it not to correct the FR above 3K, and guess what, it sounded better uncorrected.

In wondering why, given that many FR variations above 3K are baffle related and change as one moves off axes, correcting them as the DEQX does on one measurement axes, would likely cause issues elsewhere. Given that most passive tweeter crossovers do not EQ much if anything above the crossover region, I wonder if this is the key to the passive sounding better?

Will have to break out the speakers that sounded better on passives and retest.

branwell

Now you are beginning to sense the true complexity of the problem, and why shortcuts to easy solutions rarely prove satisfactory.

Ponder well these things you have learned, and how you might benefit from them and grow in wisdom and skill. In simplistic solutions lie the Dark side...

[Branwell]

I understand oh black one. I must unlearn everything I have learned….

[cjd]

I understand oh black one. I must unlearn everything I have learned….

No!

You may have to re-learn it, however. And it may be a different understanding you reach after your learning this time.

So, I'm curious what all is being done to correct and how it works. And why it may be something you hear as detrimental.

C

[Rick Craig]

In the past, I noticed that I prefer a well implemented passive crossover to the DEQX ( mid to tweeter ).

Never gave any thought as to why, just accepted that in an apples to apples comparison, the passive sounded better.

Had an interesting discovery this evening.

In general, the DEQX will correct the mic’ed FR within predefined limits that are system and user adjustable. While the system will adjust the lower limit depending on various variables, it doesn’t adjust the upper limit, which is set at 20K from the factory.

I was playing with the upper end limits on the DEQX this evening and for grins, set it not to correct the FR above 3K, and guess what, it sounded better uncorrected.

In wondering why, given that many FR variations above 3K are baffle related and change as one moves off axes, correcting them as the DEQX does on one measurement axes, would likely cause issues elsewhere. Given that most passive tweeter crossovers do not EQ much if anything above the crossover region, I wonder if this is the key to the passive sounding better?

Will have to break out the speakers that sounded better on passives and retest.

branwell

I think the way you design / measure your passive crossovers will account for the difference that you hear versus the DEQX response. If you can explain your methodology I can offer some help.

[Dennis H]

Can you tell the DEQX software to smooth the frequency response before it calculates the FIR filters? Seems like that would help it provide broad filters for things like baffle step or tweeter rolloff while ignoring narrow diffraction artifacts.

[TacoD]

Risking becoming rude and way offtopic, but I think a good design starts with a good designer. DEQX is just a tool, more flexible then most, but still a tool. Like having a shed full of filter components a soldering-monkey and a xo-simulator and a measurement setup in one. Therefore like Jon said, when you learn more and more about loudspeaker design, your speakers will becoming better and better with each subsequent iteration. It's not about passive better then DEQX or vice versa, your skills improve with each iteration

[Branwell]

Here are two graphs of the millennium tweeter taken on axes that might answer some questions.
One is raw, the other smoothed ( not corrected ). If they look unusually lumpy, look at the resolution the system is set at.

You can see what I assume is the “M” shaped diffraction ripple with a high at 1700hz, low at 3K and high at 6K. I say assume because it flattens out if I measure 15 degrees or so off axes. The bump at 15K is something to do with the Millenniums, as it doesn’t show up with other tweeters I have.

Hi C,
The DEQX can correct Amplitude, FR, Phase and GD. How it works? I haven’t a clue, but likely gerbles manipulating 1's and 0's like this :bash:

Hi Rick,
Thanks for the offer, and I might well call you up one day and try to convince you to give me some lessons :B , but to go into it online would take a heck of a long writing…I appreciate the offer though ;x(

Dennis,
Yes, and when one tells it to run the corrections, one uses maximum smoothing.

[Branwell]

Graphs

[Branwell]

TacoD,

You think I am dangerous with a DEQX, should see what I can do with a Bulldozer :T

B