Active Crossovers

[Rudy81]

Will you be using your measurement rig to help dial things in?

Absolutely! Problem is I am no expert on the subject so will be learning as I go along. That is one reason I wanted to mimic the ALK universal, so I could comapre audibly as well as graphically on my pc.

When you say 'dialed in', do you mean set the slope, crossovers frequency, time delay and type of filter in order to get it to sound natural? I know the DC-ONE has all sorts of other settings, but have yet to learn about them. I had planned on chaning as few variables as possible in order to prevent total chaos.

This should be fun, if not really interesting.

[ClaudeJ1]

I would start with LR 24dB for the xover...it's one of the easier filters to dial in properly and then is one of the better sounding ones too. Although with your polar response mismatch, you might find that shallower slopes are gonna sound more natural.

Ok, back up the train for a moment, here, Doc Grasshopper. This is one of the best little nuggets you have thrown out there and sparked my interest.....

I certainly am aware of the polar mistmatch that exists in my "ugly stack." I can also hear anomalies related to driver spacing, polars, etc. when I'm not sitting in the sweet spot (closer to the speakers or to either side of center). When I go into the far end of the room, those anomalies are less obvious an almost disappear since the driver blend is more homogenous.

That being identified in this true confession, are you saying that because I have 1 st order slopes on everything, it makes these polar mismatches more tolerable than if I were to hook up my Behringer and go with high order slopes and Linkwitz-Riley filtering?

[Rudy81]

DrWho: Adding to ClaudeJ1's question, here is another. When you say 'polar response mismatch', is that due to having three drivers in separate locations?

I'm sure increasing the crossover slope will magnify those kind of problems. However, the idea is to get some driver alignment going as well as reducing the driver overlap which creates its own set of problems. Correct?

[Rudy81]

DrWho, another question regarding the DX38. In this thread: http://forums.klipsch.com/forums/t/42024.aspx?PageIndex=161 , 'dtel' mentions that he has a DX38 that you modified to " adjust for the gain difference from Pro to home electronics". Can you explain what you did to the gain?

[Coytee]

'dtel' mentions that he has a DX38 that you modified to " adjust for the gain difference from Pro to home electronics". Can you explain what you did to the gain?

I'll try to explain since he did it to one of mine and I was also watching.

He took maybe 12 or so little resistors and replaced them with a higher value. Simple unsolder, remove... replace & resolder. What it did was lower the output of the Dx unit by a certain amount (I don't know the amount BUT, I can notice the difference here at home)

Lowering the output of the unit also lowered the noise coming out of it. By lowering the noise coming out of it he lowered the noise going into my amps and ultimately, lowered the noise coming out of my K402's.

Did he attenuate it by 3,6,9 db's? I don't know the number. What I DO know is when I would have been cranking it up a little bit before...now requires more gain knob on my Crown amps than it did before (because the signal coming into the amp isn't as strong)

I'm realizing that for me this is interesting... for the times when I'm outdoors, windows open and WANT the system pounding loud.... I don't have those last several db's.

When I'm actually doing sane listening or am actually located within the household itself, all I now notice is I need to either turn my volume knob up more or, I need to increase the gain knobs on the Crowns.

It did serve to knock down some of the background noise.

What I think will be interesting is when I get the HT together and have a JubeScala between the two Jubilees. Now, when I put all the controls on the same levels, one set of speakers, which ever pair are using this specific active (since I've not done this to both) will have an inherently quiter output because of this. Rather than simply say "I need to run my center channel 3 db's less (or is it more?) than my mains"... I'll actually have to stop & say "Well, if I need to run my center channel 3db's less... but... if the gain difference between my two Dx's are actually 6 db's, then I guess I'll have to reduce the gain in my JubeScala dx by 9 db's to get the 3 db decrease relative to the stock Dx"

Maybe I'm worrying too much

[]

[Rudy81]

Coytee: Most interesting. I doubt I'm going to start hacking into the DC-ONE that should be here tomorrow. It will be interesting to see if it has any 'noise', if so, that won't be good. My system is absolutely quiet when I crank it up and when playing soft passages in music. A noisy 'filter' will likely end up back on eBay. I was not aware they had that 'problem'.

I am not familiar with the DX38 controls, but the software I have been playing with for the DC-ONE allows me to attenuate the input and/or the output. I would guess the DX38 has similar ability. That should preclude you having many problems with the center Jubscala.

I am keeping my hopes up that this will have been a worth while effort. From what I have learned thus far, the active crossover offers so much flexibility and custom tailoring, I am hoping I will like it. As long as it isn't 'noisy', I plan on keeping the active filter system. If I ever go to a Jubilee or Jamboree, I don't need to mess with new crossovers. Hopefully, I will have my ALK's available for sale. Time will tell.

[Rudy81]

One other thing Coytee. Depending on what pre-amp or pre/pro you use for your HT setup, you may or may not have to worry about volume levels on individual speakers. For example, if you use something like my Integra DHC 9.9 it doesn't matter that your center is louder or softer than your mains or your effects speakers. When you run Audyssey, or a HT setup program, it will attenuate or boost each speaker so that you end up with proper volume levels. That is, of course, if your pre/pro controls all speakers.

If not, you will have to calibrate your volume levels with a pc program or a hand held dB meter. I would suggest an RTA program to accurately set volume levels.

What pre/pro are you looking at for a HT setup?

[DrWho]

Will you be using your measurement rig to help dial things in?

Absolutely!
Problem is I am no expert on the subject so will be learning as I go
along. That is one reason I wanted to mimic the ALK universal, so I
could comapre audibly as well as graphically on my pc.

When
you say 'dialed in', do you mean set the slope, crossovers frequency,
time delay and type of filter in order to get it to sound natural? I
know the DC-ONE has all sorts of other settings, but have yet to learn
about them. I had planned on chaning as few variables as possible in
order to prevent total chaos.

This should be fun, if not really interesting.

Backing
up a step...the goal of the xover is to sum the two drivers together so
that the final passband is flat. Just picking a xover frequency doesn't
guarantee a flat passband because you have phase rotation from the time
arrival differences, and from the natural phase response of the
individual drivers...and even from the type of xover slope too.

Here's a crash course in how I personally go about doing it...

When you pick a xover point (based on polars/distortion
numbers), what you want to do is put in the filter and then measure the
MF and HF separately, but plot them on the same graph (use the "all
measured" tab in REW). The frequency where the two lines overlap is the
acoustic xover point, which can be different from the electrical xover
point since your speakers won't be perfectly flat. Then take a
measurement of both drivers combined...what you want to have happen is
for the combined response to be 6dB louder than each driver is
individually at the acoustic xover point. Usually what happens
(especially with horns) is that you'll measure a dip instead of +6dB.

To
fix the phase, what you want to do is add delay to the tweeter to
time-align the two signals. You can actually measure the time delay
using REW, which would be a good starting point, since that compensates
for the propogation delay. Using a LR xover type ensures that the xover
is phase aligned (since it is by definition). So all that is left is to
compensate for the individual driver phase. The easiest way to do this
is to adjust the delay on the HF.... Start by playing a test tone at
the acoustic xover point and measure the SPL. You'll notice that you'll
be able to make the SPL at the acoustic xover point go up and down as
you change the delay...so stop when you hit +6dB. Also make sure you
limit yourself to within +- 1/2 cycle so that you don't mess up the
time alignment (every 360 degrees of phase the on-axis magnitude will
be +6dB).

Now measure again and you should see that your dip got filled in real nicely and your passband is flat.

Keep
in mind that this is just the basics...there's a lot more involved with
voicing a speaker than just setting the xover frequency and getting
proper summation through the passband. There's a few other nuances that
show up with the xover settings too that are hard to describe without
having measurements handy.

[DrWho]

That being identified in this true
confession, are you saying that because I have 1 st order slopes on
everything, it makes these polar mismatches more tolerable than if I
were to hook up my Behringer and go with high order slopes and
Linkwitz-Riley filtering?

That's a difficult question
because it's already a problem that the polar response is
mismatched....so now you're forced into a situation where the on-axis
response has to be sacrificed for the off-axis and it becomes a nasty
game of sloshing energy around to achieve a proper voicing.

It
seems the difficulty with a higher order slope when there is a large
polar mismatch is that it creates a very sharp tonal shift off-axis. In
principal, a slower slope could theoretically allow for a slower
transition in the off-axis behavior, but in practice you still have to
fight against the polar lobing from two sources that aren't perfectly
in phase everywhere (due to their physical separation). Is it better to
have a sharp change in the power response, or a slower change that
yields random dips and peaks that cause each individual reflection that
hits the listening position to have a different tonal balance?

It should also be noted that steeper slopes are generally going to exhibit less overall system distortion than shallower slopes.

In
the end, I've found that it's easier to get an acceptable voicing with
slighlty shallower slopes, but I also find myself wanting to tweak and
change things between every song. Going steeper can sometimes sound
better, but I think a lot of that is heavily room dependant...and it's
usually a lot harder to dial in too. I'm yet to find a very happy spot
when polars aren't very close, but I think one can do less damage with
shallower slopes. However, I personally don't like to go shallower than
12dB/octave due to possible excursion related issues.

[DrWho]

I'm realizing that for me this is interesting... for the times when I'm outdoors, windows open and WANT the system pounding loud.... I don't have those last several db's.

Even with the Crown volume controls cranked up all the way? I'll have to recheck the numbers, but I coulda sworn you should be able to drive your K2's into clipping with everything cranked up all the way.

The real purpose for lowering the output gain of the Dx38 is to allow the preamp output to be cranked up much further without tearing your head off. This gets you more resolution through the DSP on the Dx38, which in turn lowers the digital noise floor.

Btw, I realized that I could just wire a resistor in parallel with the ones already on the board, which gets rid of the whole desoldering process. It takes way less time to do the work, and makes undoing the mod even easier.

What's the time frame on your HT, Richard? I'm hoping to move to a Dx38 in the hopefully not to distant future, so we might be able to do a trade if you don't want to deal with volume matching your modded Dx38.

[Coytee]

Coytee: Most interesting. I doubt I'm going to start hacking into the DC-ONE that should be here tomorrow. It will be interesting to see if it has any 'noise', if so, that won't be good. My system is absolutely quiet when I crank it up and when playing soft passages in music. A noisy 'filter' will likely end up back on eBay. I was not aware they had that 'problem'.

I am not familiar with the DX38 controls, but the software I have been playing with for the DC-ONE allows me to attenuate the input and/or the output. I would guess the DX38 has similar ability. That should preclude you having many problems with the center Jubscala.

I am keeping my hopes up that this will have been a worth while effort. From what I have learned thus far, the active crossover offers so much flexibility and custom tailoring, I am hoping I will like it. As long as it isn't 'noisy', I plan on keeping the active filter system. If I ever go to a Jubilee or Jamboree, I don't need to mess with new crossovers. Hopefully, I will have my ALK's available for sale. Time will tell.

I've not read further down yet (in case it's been mentioned)

I don't think this was done because the Dx is noisey per se'.... (it's not). I don't remember his comments about it but what I think I do remember is, the Dx wants to be maximized on the input section. Doing this will help maximize the signal/noise ratio. When I first got mine, I once talked to Roy, ingrigued as to why the Dx HE was using had a bunch of "blinky lights" on it and I really had zero lights on mine (input gain indicators). Long story short, I guess I had the gains on my amps maxed and was using VERY little juice through the Dx to drive them to listenable levels. Once I turned the gains down, that required me to turn my preamp volume up and doing that, drove the input of the Dx harder, which in turn, helped maximize the signal/noise ratio.

Somewhere in this process, DocWho came up with the idea of doing this modification to help further lower something in there...not that it was a problem per se... but I think that "if it can be improved, why not improve it" kind of thing.

Sometimes I wonder if I create more harm on some of these technical subjects than I help since I don't always know the technical angle and kind of use some general comments....

oh well... you'll figure it out soon enough and I'm sure he'll be by here about 2:45 AM some evening and answer it! (darn punk never goes to bed!)

[]

[DrWho]

lol, you don't give yourself enough credit Richard.

Btw, there are other digital artifacts that happen when your input inside a DSP is near the noise floor...I've often wondered if that doesn't inject some graininess to the higher frequencies. Sadly, I've never had the opportunity to do a side by side comparison on the mod, but my biggest driving factor for investigating it was to see if the Dx38 couldn't become more liquid sounding.

I might need to be reminded, but I have the numbers written down at work. I should be able to tell you exactly what the extra attenuation is.

[Coytee]

I'm realizing that for me this is interesting... for the times when I'm outdoors, windows open and WANT the system pounding loud.... I don't have those last several db's.

Even with the Crown volume controls cranked up all the way? I'll have to recheck the numbers, but I coulda sworn you should be able to drive your K2's into clipping with everything cranked up all the way.

The real purpose for lowering the output gain of the Dx38 is to allow the preamp output to be cranked up much further without tearing your head off. This gets you more resolution through the DSP on the Dx38, which in turn lowers the digital noise floor.

Btw, I realized that I could just wire a resistor in parallel with the ones already on the board, which gets rid of the whole desoldering process. It takes way less time to do the work, and makes undoing the mod even easier.

What's the time frame on your HT, Richard? I'm hoping to move to a Dx38 in the hopefully not to distant future, so we might be able to do a trade if you don't want to deal with volume matching your modded Dx38.

Well crud...I should have looked further down and not had to type all of that! You're early lol

Regarding the last couple db's... I honestly don't know... I was trying to more make a point how it lowered the sound a bit. Truth be told, I can still have the Dx driven into overload BUT fortunately, haven't really had the need to do so. I certainly like to crank things up but... if you can keep a secret between the two of us, my reputation for cranking it might not be as warranted as rumor has it.

I am interested in getting my own Dx back that Brian has (Nola). He's going to piddle around with it but... has had some water damage to his abode and I said not to worry. I don't need it for a while.

I'll get it back and probably swap them out for a while. I have no complaints (at all) about the change that was made. Now, regarding

so we might be able to do a trade if you don't want to deal with volume matching your modded Dx38.

this suggests to me that my gut feeling is right? meaning, if the center speaker is supposed to be offset by a couple db's, relative to the main speakers (I'm using JubeScala so maybe it will have a natural offset of db's?)

Anyways... if I had two bone stock Dx's then I might have to adjust the output to the center channel?? If so, then my thinking was right that I might need to adjust it while also, keeping in mind the changes you made? If so... what are the changes in a numerical sense? 3 db reduction? 5, 9 142??

My time frame on HT? lol.... 3 years thus far? lol

Actually, I should receive my last batch of Mogami for the cross room XLR runs on Monday. I'm putting two runs of XLR wires in, one will be terminated with RCA's, the other with XLR's. Two runs of 12g speaker wires, 2 runs of Cat 6 and 2 runs of RG6 to the left/right locations. The center location will get the same as this PLUS two more runs of Cat 6 and maybe a HDMI run as well.

Right now, as in I just got a quote today, I'm trying to find a faceplate that will let me run these to a single box location and have a single faceplate for them. I'm finding that if I keep the XLR's seperate, I might be able to get it cheaper, but at the 'cost' of having more boxes and faceplates on the wall (all of which the wife will love). Then again, I have in addition to my normal 120 volt wall outlets at ankle level (or what ever their standard height is), I have three orange isolated hospital orange grounded outlets in each corner and a single orange one in the center. Two of these, are of course, above the height of the Jubilee bass bin so they're what... 40 inches off the ground? The third one is ABOVE those such that if I put a shelf over the K402 while in the corner (to install an amp on top of it) this plug will be available to me up there so it's something like 60" off the ground. I realize these will be hidden by the Jubilees when I stuff them into the corner... the wife though... is currently walking into the room and gagging at all these neon orange outlets screaming "I will clash with anything you decorate with in here and there is nothing you can do to hide me" at her. [6]

Each drop (left, center, right) of orange outlets will be on their own 20 amp circuit AND each outlet will be switched on the top and unswitched on the bottom. Since I might never USE these outlets (she says she wants all my amps/stuff in a cubbyhole space I'm building above a LaScala in the rear) I will run a duplicate line of these outlets to that same location, also on seperate circuits WITH the addition of yet another 20 amp circuit. In this location I will have 4 double gangs of outlets, each on one of these circuits and will be 'live' outlets. Above these I'll have a duplicate double gang of outlets and these will be switched to the same redline that is switching the top halves on the other side of the room. This way I'll have a wall of 4 switches when I walk into the room that will control some things if I want.

All of this is creating a timelag for me since I'm not an electrician. I originally installed some 12/2 w/g UF (underground wire) since I already had it. Found it might not be correct for code. Yanked it out and replaced with normal 12/2 w/g. Was then told I needed 12/3 w/g wire (before I decided to have switched outlets) and bought some 12/3 wire...

Furher chat, indicated I didn't need 12/3 wire that the person telling me I needed 12/3 wire actually calls 12/2 wire w/g "12/3" wire since it has 3 wires. He really MEANT 12/2 w/g so I took the 12/3 back and rewired everything with 12/2 w/g.

Yes... I then decided I really DID want to have the switched outlets since it's easier to have them now then retrofit them... so I went back to buy the 12/3 w/g (12/4 in his mind) that I had returned days before... ripped out the 12/2 w/g that I had installed days before and just a couple days ago, finished up the outlet side of the room, pulling it all back to the switchboxes.

So...when will I be done?

hahahahahahaahahah

[DrWho]

Yikes, running 12/3 on switched outlets is gonna create ground loops... []

j/k of course....I couldn't resist

[Coytee]

How/why so? Both halves share the same ground wire. Ultimately, as you follow the lines back to the source, they'll both share the same black & white feeder wires as well. (starting out 12/2 w/g to the powered outlets and taking the 12/3 out from there with the red (as I understand it) traveler line, tied to the black hot somewhere (probably wired into the back of the recepitical since I'm using the rear connect type which will allow 4 wires into each side, 8 wires total as I recall seeing)

Although I'm not an electrician for a reason (I remember being fascinated as a 5 year old, with jamming a screwdriver into one of these cool things I see on the wall that other "big people" stick things into.... don't ask what happened [:#])

Anyways, I'm not an electrician for a reason, but I sure hope you're yanking my chain since I really do not want to rewire it all again [8o|]

??

[Rudy81]

DrWho, thanks for the basic setup procedure. I had thought LR would be best since the 'theory' behind it is that you get a flat response at the crossover point. I'm a little confused about having +6 dB at the acoustic crossover. I will try it all out and see what I get. I now know why active crossovers are not more mainstream in the home market. The average person is not going to want to jump through hoops to set up a stereo. But.....we're not avearage, are we?

[Rudy81]

DrWho: Are you using the Impulse tab to look at the time delay? If not, how are you doing that, and can you post a graph example? Gracias.

[DrWho]

DrWho, thanks for the basic setup procedure. I had thought LR would be best since the 'theory' behind it is that you get a flat response at the crossover point. I'm a little confused about having +6 dB at the acoustic crossover. I will try it all out and see what I get. I now know why active crossovers are not more mainstream in the home market. The average person is not going to want to jump through hoops to set up a stereo. But.....we're not avearage, are we?

Each individual driver is 6dB down at the acoustic xover point such that a perfect +6dB summation yields a flat passband.

The butterworth is only 3dB down at the acoustic xover point, but it only gets +3dB summation because the phase is off by 90 degrees, which is why the polar bubble fires off-axis. If you adjusted the delay so that they were in phase, the butterworth be 3dB hot in the passband, but your polar bubble would then fire forward.

Btw, any properly engineered xover - whether passive or active - needs to go through these steps. It's just a lot harder to do with a passive because the only control over phase resides in the inductors/caps being used to make up the filter. That's why it kills me to see so many passive xover "experts" playing with filters without measuring the final acoustic output....it's practically impossible to tweak out ever last ounce of performance by ear.

[DrWho]

DrWho: Are you using the Impulse tab to look
at the time delay? If not, how are you doing that, and can you post a
graph example? Gracias.

No, REW is a bit backwards in that regard because it normalizes the peak to time=0...which is uber lame.

Are
you using the loopback feature? You need to use that in order to
measure the time delay. The REW forums are down, but I found this
picture to help you know where to look:

See
where it says "Details (15-Dec-2005....)" ? Under that tab it will have
a "System Delay:" readout IF you used the loopback feature during your measurement. This is the amount of time that the impulse
response has been normalized to define where time=0. In other words,
it's the amount of time it took the sound to reach your microphone.

What
you want to do is measure the MF with a highpass, but no lowpass, and
then measure the HF also with just a highpass. Then subtract the MF
System Delay from the HF System Delay to get the amount of delay that
needs to be added to the HF.

The reason you don't want a
lowpass on any of the drive units you're measuring is because rolling
it off will make it measure a longer delay. If you think about it,
something that can play a high frequency can change its direction very
quickly. If you limit how fast it can change its direction, then its
going to take longer to hit its peak amplitude.

There's an
article from Syn-Aud-Con that discusses this in great depth and even
provides a means for calculating the true delay for a bandwidth limited
signal (which would be the case for your khorn LF, even without a
lowpass since it acoustically rolls off)....but I don't have it readily
available. An easy way to see it for yourself is to compare the ETC or
impulse plots of your HF versus just your LF. You'll notice that
you get very narrow spikes for the HF, but you get a soft rounded shape with the LF.

Btw, just one other comment....the impulse tab in REW is showing a log squared value for the impulse, which is why it never swings negative. A true impulse looks more like this:

[DrWho]

Anyways, I'm not an electrician for a reason, but I sure hope you're yanking my chain since I really do not want to rewire it all again

??

lol, I guess I didn't make the fine print bright enough.....here it is again in a different color:

Yikes, running 12/3 on switched outlets is gonna create ground loops...

j/k of course....I couldn't resist

[A][A][A]

I never did the screwdriver thing, but I did strip back an extension cord to see what was on the inside...I shorted the wires together thinking it would juice up the power going to the light in the room. I guess technically it did make the room brighter....but for only a second or two. []