Someone explaining the function of an Anthem pre/pro raised what I consider to be a good question.

It seems that every filter introduces a phase shift that is a function of the knee frequency (among other things). A bass management system that applies a single universal crossover frequency introduces the same phase shift into each channel, and it is relatively easy to resynch.

But in a multi-channel BM system (like my 950), each different crossover frequency introduces a difference phase shift. It seems to me that these would be a pain to re-synch before combining.

How does the 950 deal with the problem?

It doesn't. No pre/pro does. And, get this: even for pre/pros that do simple 5.1 analog pass through: even if you adjust phase correct for DD/DTS/CD if you're using a coax or optical connection to your pre/pro, you still might not be in phase for the 5.1 analog inputs, even if you have the player setup as "all large", "sub on".

BTW, the phase between the mains and sub *more* depends on the slopes of the high pass filters (to the speakers) and the low pass filter (to the sub), and not so much the crossover freqs themselves.

The best suggestion I can offer you, is to pick up the DVD-A setup disc offered by Chesky, "The Ultimate DVD Surround Sampler & 5.1 Setup Disc". This disc has DVD-V and DVD-A phase tests that at least you can characterize your system with.

Since we are talking about phase, and wrt the 950... I had previously reported that the 950 reversed the phase of the 5.1 analog inputs when using its analog 80 Hz crossover with respect to the internal digital crossovers (for DD/DTS/CD). I now know that what the 950 does is unknown. What I do now know, is that if you are using a Pioneer player (the 45A and 47Ai at least), and you compare phase out of the analog inputs vs going digitally to a pre/pro, *then* you are already out of phase. I.e., using the analog outputs on the player, even if "all large", "sub on", the phase relationship between the mains and sub *is* different than going digitally to a pre/pro. I had assumed it was the 950, but it's the player. What I don't know, is how this works with any other DVD player's 5.1 analog outputs...

Theorectically, "all large", "sub on" does not do anything to the phase relation between the sub and mains. But most digital crossovers use a 24 db/octave low pass, and a 12 dB/octave high pass filter, and, usually you get a 90 deg shift for every 6 dB/octave of slope. So *that's* where the *difference* in phase comes from, between the analog outputs on a player, probably *any* player, and a pre/pro using its digital crossovers.

And to confuse things even further, obviously some players do have BM for SACD and DVD-A, but those slopes might not be the same between SACD and DVD-A, or even the same as what the pre/pro is doing!

I'm just surprised that more people and HT mags haven't picked up on this...

Now, there actually are (at least) 2 solutions. One, and I think some Anthem pre/pros do this, is to run "all large", "sub on" to the pre/pro, have the pre/pro digitize those inputs, and then use the same crossover scheme as for DD/DTS for DVD-A and SACD. This is a kludge IMO, because that extra A to D, processing, and D to A conversion *will* affect the sound quality. The 2nd is ... i.Link. You deilver everything to the pre/pro in its native digital format, and then the pre/pro can apply the same exact BM to everything.

I'm hoping that the next Outlaw pre/pro has i.Link...

I would think that the three basic types of crossovers, passive, active and digital, would handle the division of frequencies in different ways. Not being a crossover designer, I may be speaking out of hand, but I’ll offer my thoughts in simplistic fashion for discussion.

If a single stream of data is divided by some means into two or more streams based on frequency, and then each stream put through the same type of D/A conversion, the resulting outputs, even if different in phase from the original input, would, in a decent design, still be properly in phase with each other.

Similarly with an active analog crossover, where the division of frequencies are usually made by complimentary resistor-capacitor networks in association with op-amps, there is a phase shift, but my understanding is that all frequencies shift in the same direction, and at the point of crossover, by the same amount. The points of crossover are the most important, where some small portion of the signal will be sent in two directions and would hopefully be reproduced in phase with each other.

Passive crossovers, now there’s a different animal. At each crossover point, both a capacitor and a coil are usually used in the division of frequencies, the phase shift being in opposite directions. Depending on the complexity or simplicity of the crossover, if the output to one driver leaves the crossover in near opposite phase to another, one driver may be wired to the crossover in opposing phase to another driver in order to help restore the proper phase relationship of the moving drivers.

In a pre/pro or other device dealing with bass management, it would seem to me that three reasons exist for not giving the consumer phase adjustment controls. One, the implementation is unnecessary. Two, the implementation is complex and/or would add cost deemed by designers/manufacturers not to be an advantage when marketing a device to consumers, unless that device is meant expressly to deal with the phase issue. Three, giving most consumers more to adjust might lead to increased frustration for some users who would rather have things as simple as possible.

Most powered sub manufacturers give the consumer the option to switch the phase relationship by 180º, and the same option exists for passive subs by reversing the speaker leads. Some powered subs have a variable phase adjustment. If the method of dividing the frequencies is consistent for all frequencies, a one time adjustment of a subwoofer should be all that is necessary. I don’t that kind of consistency is too difficult to achieve.

As far as I can tell the Anthem allows one to set separate high-pass freqs for each channel, but uses a single low-pass for all channels, keeping the phase of the combined result coherent. Problem is, that leaves coverage gaps between the high and low freqs and you end up throwing info away.

If the whole idea is that the loss of info is compensated by the coherence of the bass, then phasing seems to be a big problem (at least in their minds).

I'd like to hear the Outlaws address this issue, too.

bang, that's pretty much how I understand the different types of filters to work, too. I think you're right about the output phase shifts being coherent between the two resulting streams. That isn't a problem when the stream is going to a standalone transducer, but when you have 7 streams potentially shifted to 4 different angles, then combined with a 5th stream into a sub out, what happens when the same signal is present on all 5 sources? I'd think it could get chopped up pretty good.

Another problem with phase shifting controls (adjustments) is that as the frequency goes to zero, the delay time goes to infinity. They'd have build in an absolute high-pass filter and stop quoting response down to DC.

I wonder whether the 'distance' configuration might be used to tweak bass stream phases? I wonder how you'd measure it without a 4-channel scope?

I’m not familiar with the Anthem’s handling of bass management, but my understanding of the Outlaw 950 leads me to believe that choosing different crossover points for main, center and surround does not produce a ‘gap.’ In digital mode, if L/R are set at 60Hz, the center at 80Hz and the surrounds at 100Hz, the sub receives all that is below each setting, channel by channel. When using analog bypass mode with bass management ‘on,’ all speakers get the same 80Hz crossover split.

As to timing and phase, I suppose an ideal adjustment could be achieved with enough course and fine control over time delay for each channel, but only for one point in space. Since some of us will be listening with two ears and those ears may move from place to place within the listening area, maintaining any exactingly obtained phase relationship for the listener would seem impossible. ‘Ballpark’ would be about the best one could hope for. Then there are the myriad effects of reflected sound and/or standing waves. There is almost no end to the things that need to be taken care of in the pursuit of an unattainable ‘perfect sound.’ I think we end up with the pursuit of an ‘acceptable’ sound.

Four imperfect solutions:

One, put the subwoofer/s under the chair or over the head of the listener. If that means the sub is four feet from the listener’s ears, then all the other speakers should be four feet away also, and all delays at zero.

Two, similar to one, however most speakers’ placement is normal and the delays are set normally. The sub, under or over the listener, would need to be delayed even more than the other speakers because it is closer to the listener.

But sub placement near the middle of the room is rarely good for low frequency response.

Three, each loudspeaker has it’s own subwoofer adjusted in phase with that one loudspeaker. But such subwoofer placement is also rarely the best.

Four, the subwoofer phase of one or more subs, some people liking stereo subs, some people liking a separate LFE sub, are set to agree with an aggregate of either the main L and R, or L, C, R, or all speakers in operation at once. As a listener shifts position, the listener receives the best compromise available within realistic means wherever they are.

My own real world example is the running of stereo subs via an Outlaw ICBM between pre/pro and amp, each sub not too far from, but not co-located with, the my main L and R speakers. Each sub is in general phase ‘agreement’ with the appropriate L and R speaker, which certainly helps with stereo listening. For surround listening, phase ends up approximately correct for the center speaker also, whose low frequencies are sent to both subs equally, so center and sub sounds reach the middle of the room at nearly the same time. The rear speakers are about the same distance from the center of the room as the fronts, so there is general timing agreement for those speakers as well, if one listens near the middle of the room. Listening from near either end of the room means ‘all bets are off’ if one is concerned with maintaining some type of exacting timing/delay relationship, but this hasn’t detracted from my enjoyment.

With no perfect adjustment for everywhere in the room, and with a fully enjoyable result obtainable with correct coarse adjustments, it doesn’t surprise me that most equipment treats this issue in a minimalist fashion.

One of the major reasons I use an external crossover (24dB slopes for all crossover points which yields in-polarity response at each crossover point) for the subwoofer-to-mains split is to maintain phase relationships regardless of which source, analog or digital I am listening to. This is essentially the same thing the ICBM does except for the 24dB/octave aspect.

Yes, that’s right. Now, go to your local guitar center to get yourself a 24dB/octave Linkwitz-Riley active crossover so you can stop using that bass management and run your system like big boys do!!

An external crossover won't solve the bass routing problems that most people have, but the ICBM will.

But this isn't how it happens. Like I mentioned, most pre/pros and receivers when they do digital BM use a 24dB/octave low pass, and a 12 dB/octave high pass. Because the slopes are different, the phase delays are different. That difference is 90 deg per 6 db/octave of slope. So ... for the low pass, the shift is 360 deg, and for the high pass, it's 180 deg. Hence the ubiquitous 180 deg phase switch on subs. But that isn't all there is. Unless your sub is right on top of a main speaker, there will be additional phase delay introduced by the actual difference in that distance. Plus, even though digital crossovers are designed to have the same phase characteristics of analog crossovers, there *is* an extra delay due to the digital processing itself. Hence the fact that it is really difficult to get all the phase relations to match up between using digital crossovers in a pre/pro, analog crossovers in a pre/pro (if present), or whatever you're doing with BM in the player (if anything).

This is why Soundhound is right on the money when he says he uses identical 24dB/octave slopes for HP and LP filters for all sources.

But... There is a small problem with the identical 24dB/octave approach. Real example. 80 Hz crossover (with low pass and high pass filters). 24dB/octave LP means 24dB/80 Hz for the LP slope. But the high pass slope is also 24dB/octave which is 24dB/40 Hz. They are not equal. Hence I don't see how you get flat freq response through the crossover.

The funny thing, is that after the 950 I went with a Lexicon pre/pro, which actually uses identical 24 dB/octave slopes. So there's something there, but no one has yet explained exactly how it works.

No, man. This routing problem has been discussed before and it is not good. The best way is to avoid routing and using clean full range signal. Many people often say they can’t run their speakers full range (center, surround). I don’t understand why this is so difficult to do. All one needs to do is to find a powered subwoofer that has high level input and output level, or even better, internal Xover bypassing, that similar to this one , to do the passive bi-amping combo with their speaker. This method is also called "loudspeaker overlaping"

For example, if you want do a full range center. Just connect the full range center channel signal from the amplifier’s binding post to the LFM-1's High Level Inputs. Then, connect the high-level output terminals on the back of the LFM-1 back to your center channel speaker. Now, it is full range. This connection will let you run full range without using bass redirecting. You can also use this method for full range surround or any channel you feel like full range.

The best, buy yourself a an active Xover 24dB/octave Linkwitz-Riley active crossover from guitar center and connect it with your LFM-1 by flipping the switch on the back of LFM-1 to internal Xover bypassing mode for more precise fine tuning.

Before I upgraded into pure active Tri-amping, I use this connection for a while and it works great. The center channel would put out some serious bass like I experienced in cinema. No more hollow vocal; Just strong, well defined mid-bass.

The sub placement is also very easy. Just put it under your center channel speaker (many people put center channel above their TV) and between your front main.

No, man. The 24 DB Linkwitz-Riley crossover has a characteristic of absolutely flat amplitude response through out the passband with 24 dB/octave rolloff. Its low pass and high pass outputs are everywhere in phase with zero phase difference between drivers at crossover. That was what all about. Different crossover type has different characteristic. There’s also Butterworth, Bessel crossover. Linkwitz-Riley is pro standard.

I'm not sure why pre/pro designers would do this. The outboard analog active crossover units I've looked into have been 24/24 and with a rather flat sum of the outputs.

Well, yes, I totally agree - that's what I do. The problem is that most people don't want to mess with this stuff, so I would think the ICBM is a good compromise for them.

Personally, I think the whole bass management scheme is a load of junk that shouldn't have been foisted on the public. It was all done for economics so that small and cheap speakers that the wife would like could be made rather than full range jobs with real 12" or 15" woofers.

Dubbing theaters where movies are mixed do not use bass management.

Well, I have full-range mains (down to 32 Hz) and have been planning to do the hi-level config with "Large" L/R and "No Sub" to drive both subs in an 'envelopment' configuration. (Room is still in process.) That would eliminate re-direction problems on those channels (combining the LFE should not present a problem since there are no filters involved).

But I don't have full-range center and surrounds. If I set them to "Small" then the phase problem is back. If I set them to "Large" the bass content of those channels is lost.

Which are you suggesting?

teod and bb4tb:

See that's just it. I *do* understand why they use 12 db/octave for the LPF and 24 dB/octave for the HPF: because the slopes on either side of the crossover are identical: 24/80 = 12/40 (for example, for an 80 Hz crossover).

These are just words that aren't backed up by how the slopes add together. Others have stated it before too, but the slopes do not add up. Can you add some more info? (I think I saw once "flat amplitude response" maybe vs "flat energy response" or something.) ??

I actually do now also listen to SACD and DVD-A all large, sub on. (My "weakest" speakers still do -3 dB to 45 Hz. Mains are -3 dB to 30 Hz.) Music doesn't have much content below 40 Hz, and I personally feel you lose more with bad BM than by speakers that roll off too soon. But for movies, I do use digital BM and TA.

Yes, man. But I really don’t understand why most women hate to see large speakers in the living room. That is really unreasonable. A wife should enjoy audio as much as her husband does. That is a happy marriage. If that is a problem, I would rather not marry. I think stay single will give you more freedom. For example, I don’t have to worry about kids, wife, family expenses and whole bunch of troublesome issues I could imagine of that would later come up in the marriage. I could do whatever I want to do. No restriction at all by the wife and family. But, that is just my opinion. Maybe I am just too young. What do you think, man?

By the way, in cinema they don’t use bass management for movie soundtrack playback either. A very powerful stage LCR channel setup will beat any wimpy 5.1 or even 7.1 that is improperly setup.

My understanding of a –24db per octave L-R crossover is:

At an example crossover frequency of 80Hz, the low-pass section has started it’s roll-off below 80Hz. At 80Hz the signal is already –6db down. The high-pass section is complimentary to this. As the frequency rises, the slope upward crosses through 80Hz at –6db down and does not approach 0 db until higher than 80Hz. Since complimentary resistor-capacitor networks are used, and at the same rate of –24db per octave for both high-pass and low-pass, the phase shifts but remains coherent. With the same resistor-capacitor values chosen to halve the amplitude at 80Hz for both rising and falling frequencies, combining the two outputs for a test would show that at 80Hz, the two –6db levels sum to 0db, resulting in a close-to-flat response through the crossover point.

Maybe I need a better education on this?

24dB/octave Linkwitz/Riley crossovers are indeed amplitude flat through the crossover point. This can be proven by summing the two sections (lo pass and hi pass) through two equal value resistors and measuring the frequency response through the crossover region. It will be ruler flat. The phase also has to be zero degrees too since any other phase angle would introduce amplitude variations in the crossover region with the simple summed resistor network, or any active summing netork.

I think you should stick to giving advice about subjects you have experience in and know something about. Just my humble, married with a six month old son and never been happier in my life opinion.
Another result of those small, non-full range speakers foisted upon us unsuspecting consumers is that we can have multi-channel home theaters in reasonably sized spaces without having a multi-million dollar/year salary.

As long as the bass management is done in the digital domain there will be no phase errors. That's one of the beauties of digital filters. Check out this technical article for a good explanation. The discussion is centered around CD player "brickwall" filters but the concepts are the same when applied to digital filters implemented as a crossover. The most commonly used type of digital filter for audio use is the FIR (finite impulse response) which is further discussed here . This not to say that digital filters are perfect, they have their own set of problems but the introduction of phase errors is not one of them.
Bass management done in the analog domain can introduce phase errors and having staggered cutoffs for the different channels could have some interesting results. However, in the case of the 950 analog bass management is only done on the 6ch input AND the crossover point is fixed at 80Hz for all channels. Therefore, any phase error introduced by the analog circuitry will be uniform across all channels. I also believe that these phase errors would be largely swamped by the effects of the room and speaker placement when considering the long wavelenths at 80Hz and the distances between the speakers and the sub(s).

I don't believe that that's true.

1) I was told by someone in the industry, that the digital filters used for BM are designed to mimic analog filters.

2) If you hunt around Secrets, there is an article by Brian Florian on why multiple (individual) crossovers are a bad idea. I can actually refute most of his arguments, but he does state that when you use individual crossovers, the shifts are different for the different frequencies. That has been backed up by tech support people at a few companies I know.

Plus, Anthem (one of them) has a special setting in their software to do the following. They have individual high passes but a global low pass. The wrong way to do it IMO, because you either leave holes in the coverage or overlaps unless you set all the filters to the same exact number. But anyway. There is a scenario that allows the full range LFE signal to be combined with the globally low passed info from the mains. But with the shift that takes place for the low passed info from the mains but not the full range LFE info, you get phase effects (and cancellation). So there is a special setting to artificially adjust the delay of the full range LFE info so it combines more in phase with the globally low passed mains info.

Now, if the global low pass filter didn't introduce any phase effects, you wouldn't need that special adjustment. But it's there nonetheless.

But personally, I don't understand *why* digital crossovers are designed this way. But that seems to be the way it is.

TD- your first link is for an antialiasing filter which is an FIR filter. Obviously, a different application than crossovers in a pre/pro. Good info, but I can only conclude that this *isn't* the type of filter (usually) used for digital crossovers for BM. I still have my connections at Anthem and Lexicon, so maybe I'll ask. Good info!

SH and BB4TB- Yes, I understand the phase effects. Same slope, same phase shift. I guess I need to do lttile bit more research into:

a) If a L-R 24/24 crossover is the best, why don't more companies use it?

b) Why my simplistic argument about the matching slopes (by freq) on either side of the crossover doesn't work.

Don’t forget the crossover used inside your passive speakers (If you have one). That is the most critical part.

One reason is that the bass drivers used in most speakers today are so small (6-8", which hardly qualifies as a woofer in my opinion) that they don't have enough bass extension to allow such steep slopes. Ideally, the response of a speaker should extend an octave above or below the point where it is to be crossed over. If this is not the case, the rolloff of the speaker itself is added to the slope of the crossover netowrk. A crossover that has a 12dB/octave slope when used with a typical speaker, yields a much steeper slope than the crossover alone.

I see SH’s point. If my speakers have a -12db/octave roll-off starting at 60-80Hz, and my crossover is has a –12db/octave roll-off at a similar frequency, my effective acoustic roll-off will be –24db/octave. In order to have acoustic continuity using an –24db/80Hz crossover, a full-range loudspeaker would need a rating down to 40Hz.

One method sometimes used to help fill in a gap in a situation where there is a 80Hz crossover point used with a full-range loudspeaker not useful down to 40Hz is to build a mild custom boost into the crossover, as in a Linkwitz Transform Circuit, to the lower end of the desired frequency range. This helps compensate for a combined roll-off of crossover and loudspeaker, but is usually not something that can generally be applied. The designer must know the characteristics of the loudspeaker in question before determining what frequency and amplitude to use in such a compensating circuit.

In general, it’s better to have speakers where such compensation is not needed. Also, if the speakers performance is too limited, too much compensation leads to distorted performance. I suppose that such compensation only works when the assistance the loudspeaker needs is only mild to moderate

SH- Wait- with steeper slopes, you don't need as much bass extension. With a less steep slope, you need more.

But I do agree with the "rule of thumb" that dictates setting the crossover a half octave to a full octave above the -3 dB point of the speakers specifically such that the natural roll off of the speaker doesn't "add" to the imposed slope from the crossover. I personally believe that a lot of people have their crossovers set too low.

Soundhound

I sent you a PM.

If the speaker itself is already rolling off in the vicinity of the crossover point, then it's slope is added to that of the crossover. If the speaker's natural rolloff occurs an octave above/below the crossover point, they are offset enough so that addition does not take place - there is minimal interaction.

Yes, a steeper crossover does take place over less bandwidth than a crossover with a gentler slope, but nonetheless if the speaker is already rolling off in the crossover region, then it's slope is added to the total slope. A steep crossover is best if there is adequate excess bandwidth available below or above the electronic crossover point.

This is an issue because many satellite speakers simply don't have solid response down to the 40Hz that would be required for no interaction with an 80Hz crossover network.

The manufacturers really have to design for the worst case, that that worst case is the reality that many speakers are already rolling off in the vicinity of 80Hz by themselves because of their small woofers and cabinets.

The 950 manual indicates a HPF slope of -12db/octave using 2nd order Linkwitz-Riley and a LPF slope of -24db/octave using 4th order Linkwitz-Riley.

The ICBM manual indicates, in ‘normal’ setting, both HPF and LPF slopes are 2nd order Butterworth.

Assuming that the ICBM’s Butterworth arrangement will cause a gentle rise to a +3db boost approaching the crossover point with a sum of the outputs, this would tend to compensate for a loudspeaker characteristic that already is -3db down at the chosen crossover point. If a loudspeaker and sub are down -1db at the crossover point, the acoustic in-phase sum would be +2db. However, a sub and loudspeaker will be slightly out of phase in all but a few ideal spots in the room, so I’m guessing there wouldn’t be the full +2db acoustic boost in reality.

Seems the ICBM designers were making some assumptions about real world equipment/environment and figured if the end result of the crossover would keep things within +/- 1db at the crossover point acoustically, they were doing pretty well – similarly for the 950.

SH- That's one thng I don't understand at all. The large number of speakers that are supposed to be used as surrounds and rears that are only rated to -3 dB at 80 Hz. That might work for a THX rated speaker where the speaker's roll off is purposefully combined with the crossover, but most people don't use a THX crossover setup with non-THX speakers. I even specifically asked Mirage about this with the OM-R2's, the speakers that are supposed to be used with what I have up front, and they either didn't understand the question, or pretended not to. So I went with their Omni 60's which are -3 dB to 45 Hz. And I even run that crossover a little high at 100 Hz, because 3 of the 4 are up against walls. The higher crossover helps deal with boundary low freq reinforcement.

Bump, hoping for some input from the Outlaws.