Active Crossovers

[Rudy81]

Here is the Butterworth simulation with the above mentioned settings.

BTW, here is a handy page to help with the DCX2496 auto-align: http://www.dcx2496.fr/en/calage_en.php

ButterworthOpt.pdf

ButterworthOpt.pdf

ButterworthOpt.pdf

[KdAgain]

Rudy,

Thanks very much for your earlier informative answer to my post on the DCX2496. I have followed the links and read your further posts - it is seeming like this may be a great way to go.I'm anxious to give this a try, but don't want to give up my 2496 that is presently being used as crossover/equalizer for our new Danley DTS-10 sub. So looks like I'll need to get another 2496 - that may be a little while.

For amps we are presently using an old HK-430 on our Khorns, which does quite a job for it's low price. I do have a couple of Tweak City Audio 'Gizmo's' laying around. These are Class D stereo amps, so I might try them for the midrange and tweeters, and keep the 430 for the bass bins. If this whole scheme works out I would love to build up some amps using the Hypex UCD modules.

Thanks again Rudy for all your sharing on this exciting project.

Rod

[Rudy81]

Rod. Hopefully you might find the DCX satisfactory.....bang for the buck. Please be careful plugging the DCX outputs directly into a consumer amp. From everything I can find and my conversations with Parasound, the voltage difference between pro and consumer gear may be enough to damage your consumer amps. Word to the wise. I have been using the Samson s-convert, but find it inadequate due to noise issues it adds to the line. Mine is going back to the retailer.

[KdAgain]

Rod. Hopefully you might find the DCX satisfactory.....bang for the buck. Please be careful plugging the DCX outputs directly into a consumer amp. From everything I can find and my conversations with Parasound, the voltage difference between pro and consumer gear may be enough to damage your consumer amps. Word to the wise. I have been using the Samson s-convert, but find it inadequate due to noise issues it adds to the line. Mine is going back to the retailer.

Things are never as easy as they should be, eh. Have done a little Googling on interfacing pro and consumer and it gets complicated. Since I wont be getting another DCX for awhile this will give me some time to study the issue further. Rudy, I can't remember if you are using pro amps now? I do remember you going thru many iterations before you were satisfied.

I do have some of the Behringer hum issue with my DCX which is being fed from an old HK Citation 11 Preamp. Haven't taken the time yet to totally resolve it.

So thanks for the warning - I'll definitely heed it!

Rod

[Rudy81]

Rod, yes I am using a Crown K1 for the woofers, a Hafler P1000 for the tweeters and currently a consumer Parasound that is using the Samson S-convert to lower the input voltate.

I should receive a Hafler P1500 this week for the mids.

[Rudy81]

Finally got a chance to properly hook up the Hafler P1500 and set it up for the proper gain vs. the mid JBL drivers. I also got a chance to do some A/B comparissons with the L-R 24dB filter vs. the above mentioned Butterworth 18dB filter setup. Doing A/B comparrisons is not as easy as the EV DC-ONE made it due to the pc based software.

There is something about the way the Butterworth is set up that focuses the imaging and broadens the soundstage just a bit. I really prefer the Butterworth setup thus far.

I am including the individual driver responses with the Butterworth filter in play.

[Rudy81]

This graph includes a full requency sweep.

[DrWho]

The way the blue and red intersect to create the purple line is how
you want the LF to look....which might be kinda difficult because of
how the Khorn LF spills the sound out to the sides, and the fact that it's
harder to ignore reflections at the lower frequencies, etc... but I
just wanted to comment that the HF to MF xover looks very good.

One
thing you might look into is pushing the highpass frequency for the HF
slightly higher than the lowpass frequency for the MF, or maybe just
applying some EQ to bring down that peak at ~6kHz a bit. I almost
wonder if the slight change in imaging you hear isn't related to
changing the shape of the frequency response with the different filter
types moreso than the actual behavior of the xover itself. Although it
should also be noted that a 3rd order filter is gonna shift the polar
lobe about 45 degrees or so, which is gonna interact with the room
slightly differently. Of course if that's a desired sound, then
definitely keep going that route.

But for what it's worth, it
looks like there might be a polarity issue between the MF and the LF,
or at least that big notch at ~520Hz on the purple line looks pretty
representative of that.

Btw, when/if you EQ out that little bit
at 6kHz, you'll probably find that you'll also need to put a wide Q
(low number) filter centered around 3kHz or so to help flatten the
midband even further. Then the MF could be attenuated another dB or
two, and then you should add some EQ to the bottom of the MF to flatten
out its response from 500-700Hz (push the highpass filter for the MF
down to like 350Hz with as steep a slope as possible temporarily while
the passband gets flattened before adding the xover). What I would
expect you to find when doing all these little tweaks is that the
midrange will sound smoother with a more natural extension into the
higher frequencies. Male voices and guitar will sound more full bodied,
and your low frequency sounds will appear to have a bit more authority
without sounding exagerated.

Just one other comment....try to
keep in mind the polar response of the speakers when doing EQ changes.
If you know the sound is starting to beam, then you want to let the
on-axis sound rise a little bit to keep an even power response in the
room. If you can find the plane wave response of your driver, then you
can compare your raw on-axis response against the plane wave
response....places where you have more output on the horn are places
where the horn is beaming the sound (thus a higher directivity index).
One thing I'll do after identying where all these filters want to live
is to then change them by ear while listening to music. What I've
always found when measuring after tweaking by ear is that I'll end up
with another frequency response that still looks "flat", but it's clear
that the emphasis has shifted around. I've also found that the
better the horn's polar response, the less iterations I need to take
before I'm satisfied. Basically what you're doing is striking a
compromise between the on-axis response and the off-axis response.

Btw, you can drive yourself crazy if you try to get too tweaky with
the EQ...it's real important to learn to EQ the system and not the
music because as you swap between different albums you'll find yourself
changing the EQ back and forth. I find that it's really hard to fight
this urge because you'll realize just how much control you have on the
sound and then when an album has a minor shortcoming, the first
instinct is to start cranking stuff to make it sound better.

And just a quick caveate, this is all just my opinion at this
point...there's a million different ways to approach voicing a system.

[Rudy81]

DrWho, your opinions are way better than any ideas I have on my own, so thanks for all the input.

BTW, the numbers I am using for the crossover points were generated by the '3D' excel program I mentioned a few posts back. The settings it comes up for my my Khorns based on the auto calibration function of the DCX is this:

"Butterworth 3rd order filter, crossover points at 434Hz for woofer, 569Hz-5223Hz for mid, and 6843Hz for the tweeter. Mid polarity inverted with delay of 8.83ms. Tweeter was normal polarity with delay of 9.86ms. "

Based on those numbers, the MF should not have any problems providing good response from 569Hz on up. The woofer should be ok below 434Hz as well. I'm not sure why the LF, MF crossover is so jumpy and it may be, as you suggest, a polarity issue. When I get a chance I will run some plots changing the polarity or tweaking the phase of the woofer vs. the MF.

One issue I need to keep in mind is that tweaking the EQ too much is a waste of time for me. My system uses Audyssey to fine tune the system, and it will compensate for any big EQ anomalies on its own when I run the Audyssey program. I normally listen to every type of 2 ch and multi channel audio with Audyssey.

Thus far, the system sounds really good and just needs some fine tuning. Who knew it would take this long......but worth the frustration.

[mikebse2a3]

But for what it's worth, it looks like there might be a polarity issue between the MF and the LF, or at least that big notch at ~520Hz on the purple line looks pretty representative of that.

Awesome work Rudy..!

The notch could just be a measurement anomaly due to early room reflections. I also would advise using EQ for voicing withen small adjustments ranges (usually less than 3db or 4db) and don't necessarly try to get the measurement curve perfectly flat because as DrWho pointed out you are EQing both on and off axis.

mike tn

[Rudy81]

Thanks mike. So here's an idea. I have been wondering if it would be useful to create a new tophat that would slightly angle the mid horn and tweeter 'at' the listener. Say, 10 degrees down so the horns are not firing over the seated listeners head. I have been thinking about this ever since I had problems getting the Audyssey mike to properly register the tweeter performance. Turns out if I place the mic at the listener height, Audyssey boosts the tweeters too much. The listener's ears are well below the center point of the mid and tweeter horns.

In my mind's eye, I see a Jubilee bass bin with a custom 'top hat' that angles the mid and hi horns at the listener, thus avoiding some of these problems.

Just thinking out loud....

The whole Jubilee bass bin idea is what started all this active crossover stuff anyway.

[Rudy81]

BTW mike. You should try the Le Cleac'h calculations for setting a crossover when you are bored. If you try it, let me know what you think of the results. I still can't beleive his calculated crossover settings sound so nice. I would have guessed there would be 'holes' in the response, but that is not the case.

Also, If you ever get rid of the the DC1, I have first dibs. []

[mikebse2a3]

BTW mike. You should try the Le Cleac'h calculations for setting a crossover when you are bored. If you try it, let me know what you think of the results. I still can't beleive his calculated crossover settings sound so nice. I would have guessed there would be 'holes' in the response, but that is not the case.

Also, If you ever get rid of the the DC1, I have first dibs.

Will do Rudy...[]

mike tn

[DrWho]

The notch could just be a measurement anomaly due to early room reflections.

I might argue that the microphone should stay in a single position when doing these types of measurements, otherwise you have no way of truly knowing if you're getting proper phase alignment, etc...in which case it wouldn't be a room reflection, but since I wasn't at the measurement there's really no way to know.

[DrWho]

One issue I need to keep
in mind is that tweaking the EQ too much is a waste of time for me. My
system uses Audyssey to fine tune the system, and it will compensate
for any big EQ anomalies on its own when I
run the Audyssey program. I normally listen to every type of 2 ch and
multi channel audio with Audyssey.

The problem
with tools like Audyssey is that they give you very different results
based on where you choose to place the calibration mic. Automatic tools
also have to make assumptions and run curve fitting algorithms to
determine what kind of anomaly it's trying to fix. The pattern
recognition and cognition of a human is always going to be better than
an automatic tool because there is so much contextual information that
the tool could never know...even assuming the author of the algorithm
perfectly understood the psychoacoustic principals at play (which isn't
always the case either).

But heck, now that you've got the
tools, you could easily measure what the Audyssey program does to your
frequency response and determine for yourself just how well it works.

It should also be mentioned that correcting minimum phase
frequency response aberrations will result in a flat phase response
ONLY when fixed with filters that match the Q and amplitude of the
aberration. If you flatten the frequency response after you dial in the
crossover, then you're going to end up with filter shapes that don't
necessarily match the abberations induced by the individual drivers
near the xover frequencies...which basically leaves extra phase
rotation on the table because you chose to address the final system
response instead.

BTW, the numbers I am using for the crossover points were
generated by the '3D' excel program I mentioned a few posts back. The
settings it comes up for my my Khorns based on the auto calibration
function of the DCX is this:

"Butterworth
3rd order filter, crossover points at 434Hz for woofer, 569Hz-5223Hz
for mid, and 6843Hz for the tweeter. Mid polarity inverted with delay
of 8.83ms. Tweeter was normal polarity with delay of 9.86ms. "

Is this feature something the Behringer comes with? I checked out the link you provided and it seems more like a how-to-mangle your DCX [] I'm kinda curious about the specifics of the theory behind their magic.

[Rudy81]

DrWho, the DCX comes with the ability to autimatically take driver time delay readings and if desired make other changes. The information derived from the automatic run is then plugged into the spreadsheet provided in this page: http://freerider.dyndns.org/anlage/LeCleach.htm which then gives you frequency and polarity settings to plug into the DCX.

Those readings have provided the results you saw in the latest graphs I posted.

As I said earlier, I don't quite understand the ratinale behind the settings, or the math behind the formulas, but is seems to work. That's why I wondered if mike had time to try it on his system.. I have compared these results with the basic LR settings I first set and the derived results with the Butterworth filters have a much better quality. I found it very interesting.

[KdAgain]

I've been giving quite a bit of thought this past week to using my DCX2496 as an active crossoveror for the Khorns. All my source material is ripped to a PC and it presently feeds an EMU-0404USB external DAC that sounds suprisingly good for the money. Something that bothers me is taking the output of the DAC and sending it to the DCX which includes an ADC/DAC set of conversions. That's two DAC processes, which must add some colorations. A long, long way from a straight wire with gain!

I've been researching feeding the DCX straight from an SPDIF from the PC, but life then gets very complicated. Not too bad now that I'm only 2.1, but when I get to 5.1 down the road it becomes quite difficult. Especially since it seems that for max resolution the volume control should be after the DCX - or at least a part of its analog output.

Rudy I believe your source is also a music server and that you feed it
to the internal DACs on your pre/pro. Have you noticed anything
negative that you might attribute to the additional conversions? I'm also wondering if you feel the loss of resolution by having volume control before the DCX? You seem to be very pleased with the SQ so perhaps the benefits of the active crossovers and time alignment outweigh what I perceive as negatives in the system.

Thanks,

Rod

[Rudy81]

Rod: It definately gets complicated if you want to send digital directly into the DCX. It would be ideal, but it is complicated. Although I have not been focusing too much on the quality of going throught two sets of DACs, I'm not sure I might notice. I gave that some thought early on, and figured I would only find out if I tried it. So far, the active system is a keeper.

Frankly, I think we are already way far away from a straight wire with gain. Even before you do the double DAC thing, think about how many electronic components your source bits encounter from the time the hard drive head reads the zeros and ones. I had mentioned something about that a while back. Heck, my original reluctance to going active was for the same reason. However, I am almost a convert.

It is an expensive experiment, althought if you purchased good used gear, you can always sell it back for a minimal loss at worst. That is what I have found.

I would say that for me, the results thus far have been well worth the price of admission. I am just now starting to live with the results.

[ClaudeJ1]

Frankly, I think we are already way far away from a straight wire with gain. Even before you do the double DAC thing, think about how many electronic components your source bits encounter from the time the hard drive head reads the zeros and ones. I had mentioned something about that a while back. Heck, my original reluctance to going active was for the same reason. However, I am almost a convert.

I go back to being 12 years old with my audiophile beginnings (speaker building and such). I remember expericing the 2:1 DBX compander on Jazz and Classical recordings in the 70's. I was a member of the Audio Engineering Society, as recommended by PWK. I witnessed the evolution of the CD and heard the master DBX recordings from a local sound engineer, who was involved as a consultant on Klipsch products. We went from about a 60 db Signal to Noise ratio on reel to reel recordings, to 96 db plus on CD and DBX. The Delta Modulation demos I heard from the DBX boxes in the early 80's went beyond 110 db S/N in the "digital world."

That being said, if we assume a 30-40 db gain via the evolution of digital technology, that's a 10,000 times improvement on Signal to Noise. With the advent of 96 Khz. resolution, with anything higher being decreed as "designed for bats" by Tomlinson Holman himself (Mr. THX), I think we can afford to canoondle around quite a bit (no pun intended) in the digital filter domain before our ears tell us we are messing up the "purity" of the sound.....puhlease.

My experience with "golden ears" has shown that they are more "tin ears" when presented with ABX techniques and hard science.

I applaud your objective journey and support your conclusions wholeheartedly as a "reasonably semi-informed" opinion.

[KdAgain]

Rudy & Claude,

Thanks very much for your encouraging views on using the DCX as a crossover and for minimizing my concerns. So now it's time to quit theorizing about it and give it a try.

I presently have a DCX2496 that is being used for crossover duties for our Danley DTS-10 sub. This is only consuming one of the six channels as we are running our old Khorns full range. That leaves 5 channels available and we need 6 for the Khorn drivers. What I'm thinking of doing for a first stab at this is to only use the DCX for the mids and tweets. Our Crites Type A crossover only has a single choke inline with the bass bin so I could keep that in use as a low pass. This would negate some crossover tweaking, but since the bass bin has the longest time delay shouldn't hurt me in the time alignment. Any thoughts on this approach? BTW if I end up happy with how the system works I could later on get another DCX to allow dealing with all 6 drivers.

The bigger issue to work out for now is the high output voltage on the DCX outputs going to my consumer amps inputs. There are various mods available on the internet to add volume control for each channel and to improve the output stages. For now I would like to try this with what I have if it can be make workable. A simple voltage divider or pad on the DCX outputs could lower that voltage, but might present too high of an impedance to my SS amps inputs. I may also find high noise floor levels as I believe you did Rudy. But I'd sure like to give this a try now utilizing what I have. I'm sure open to suggestions on lowering the voltage to the amps, or anything else here - getting anxious to give this a try!

Thanks,

Rod