[Kevin Haskins]

Here is the Anarchy DXT design.    I performed these measurements, outside in my parking lot, with the fork lift.  That is my standard measurement procedure when the weather is good.    All this data is un-smoothed, raw measurement data that I import into LSPCad for simulation of the crossover.   The measurement is NOT simulated, the crossover is.   I find that these pretty much match the build exactly as long as you use real off-axis measurement data.

None of my off-axis angles are exact.   I rough estimate them based upon some chalk marks I have in the parking lot.   They don't have to be exact as they just generally reflect a point along a continuum. 

This is the active design.   I don't know if I can do this good with a passive.   I should be able to get close but the active design will probably have an edge in performance.     

My on-axis measurement was goofed so I only have the off-axis data starting at 20deg.   On-axis should be fairly close or I may have to adjust the crossover a little.   The on-axis data is actually not as pretty as the slight off-axis due to the baffle diffraction, which is probably not audible. 




45 deg.




70 deg.



Vertical off-axis from above.   I would guess this is somewhere around 25-30 deg.



Vertical off-axis below.   I would guess this is the same 25-30 deg. below.




You get different vertical performance because of the physical separation of the tweeter-midwoof.    Typically the null you see off-axis in the direction of the tweeter is the largest off-axis null in a 2-way with a traditional driver arrangement.    This is excellent. 

[Carl V]

whoa.....

[Kevin Haskins]

whoa.....

Yea.... my thoughts exactly. I'm trying to take a couple components out of it before I call it done but it is hard to beat that with a stick.     I'll build and measure the final circuit in a couple weeks time.   I have PCBs but not all the component values to build it.   

[poseidonsvoice]

Kevin,

Those frequency response curves look magnificent. Does it really matter what the on-axis curve looks like with what you've got here off axis?

Anand.

[Kevin Haskins]

Kevin,

Those frequency response curves look magnificent. Does it really matter what the on-axis curve looks like with what you've got here off axis?

Anand.

It matters.... what we hear is a combination of the direct radiation, the first reflections and then the total room reflections which arrive later.   You want even spectral balance as much as possible.   

For the best performance, you want the direct sound + early reflections to be as close as possible to the same.   The top octave stuff probably isn't critical, because high-frequency stuff is mostly absorbed by the room anyway.   . 

Having it nice and smooth and even, up to 10K out to 70 deg. is fantastic though.    That is why I'm excited about the measurements.   They are proven, without a doubt to have subjective relevance.   

[Craig Treusdell]

That's wild Kevin!

I remember back in my days at Ga Tech using SPICE to do crossover simulations for a class and trying both active and passive, with active capable of just about any response. No comment on passive

I seem to recall a conversation with you and Dan W (I think) where you discussed having an intentional slight dip somewhere in the mid to upper octaves for some particular reason, possibly with the LCRs. Was this you guys and if so could you elaborate a little please?

[Kevin Haskins]

That's wild Kevin!

I remember back in my days at Ga Tech using SPICE to do crossover simulations for a class and trying both active and passive, with active capable of just about any response. No comment on passive

I seem to recall a conversation with you and Dan W (I think) where you discussed having an intentional slight dip somewhere in the mid to upper octaves for some particular reason, possibly with the LCRs. Was this you guys and if so could you elaborate a little please?

That was probably a BBC dip.   It was a dip in the on-axis response in the upper midrange to compensate for the fact that you get a hump in the off-axis data in the same range with a typical 1" dome tweeter crossed over to a large midwoofer.     As I said, we hear a combination of the direct + reflected sound so when the frequency response of the off-axis data, that represents the first reflection has a large peak in that range, the speaker will sound brighter or harsh in comparison with one with even off-axis response.    The BBC dip was just a band-aid for the problem.   

Here is an example:

http://www.soundstagenetwork.com/measurements/speakers/usheraudio_be718/

Notice the peak in the response in the off-axis 45-75 deg.   That represents the first reflections and it adds significantly to the later arrivals.   It has a good 4-5dB peak in the response centered around 3K-4K.    The old BBC "trick", didn't really acknowledge or recognize the cause of that "brightness".   It just adjusted the on-axis data, to have a dip in the response, which correspondingly would even out the off-axis data.