Mixing two and three way speakers

[EEE272]

I could not find an explanation anywhere else but my apologies in advance if there is a post on it.

Even in one and the same series you can often find 2 and 3 way speakers. Typically, you might have three-way floor stand speakers suggested for the fronts but the surrounds are often two way speakers. Some manufacturers also sell these complete systems with such a mix saying that they perfectly match.

If I understood correctly, the filter that is needed to transition between the tweeter and woofer or mids, typically causes a phase shift. Because of this, if two and three way speakers are mixed, they should at some point be completely out of phase, or not? If so, would it not cause quite audible issues?
Or are they adjusting the phase of the two way system to artificially follow the 3 way system?

Thanks in advance!

 

[Juhazi]

You will get lots of replies, but my reply is the right one!

It is crucial that L/C/R speakers have exactly same topology and xo types. 95% of info comes from those, most important is dialog and localization cues for everything happening on the screen. Rear and surround speakers give mostly effects, like added delay (echo) or independent sounds. Ideally they should be identical to LCR, but you don't loose much wih smaller/cheaper 2-way speakers.

I have two multich systems now, but only the one in HT room follows my guidelines, and yes, the difference is huge!

 

[TurtlePaul]

Some crossover slopes call for inverted mids and some dont. If it is a problem you can just change the polarity on the 2-ways.

I always assume the bigger concern is that the 3-way will have much wider directivity than the 2-way if the mid on the 3-way is smaller than the woofer on the 2-way. For example, several speaker lines use 4” mids but use 6.5” woofers on the 2-way versions. This would result in them being voiced a little differently.

 

[ernestcarl]

If the LCR are identical, then there likely won't be any audible issues. If they are different, I would check to make sure that the phase deviation is hopefully no greater than 60 degrees. More than that and you might get a not so insignificant dip in the summed response from coherent sources. Would you be able to really hear this? Maybe... I dunno. I don't have a real center channel to test.

My own LR main speakers are linear phase in between their mid-woofer and HF xo by default, but my surrounds (LsRs) are not and use an LR4 xo around 1725 Hz. The satellites are phase compliant (PC) in the region that matters the most which is esp. around the bass management xo region so I shouldn't really worry.

If we do a sine sweep measurement sent to all channels (precisely time-aligned) we get the ff:

(FDW 15 cycles) *center channel is merely simulated with an equivalent signal split between left and right by adjusting to the correct dB level

FIR correction (less than 3ms time delay) applied in the surrounds just to linearize their phase. Delays for time alignment were adjusted to be equivalent between the two saved DSP presets.

For real surround channel or stereo media sources where the signals are rarely if at all coherent mix (unless a mono track), it would be very difficult/impossible to tell if there is any difference at all. If I isolate just the surround channels where I applied my post xo phase linearization, I can hear a very subtle difference. But that is just the thing, you have to absolutely isolate it and focus really, really hard to be able to tell. I've already done the requisite ABX tests with foobar to confirm this for myself.

In the early A/B tests I've done listening to actual live stage classical music of stereo sources upmixed to pseudo-surround (JRSS or JRiver surround sound mixer), I thought there was a subtle but audible change in the perceived soundstage -- although, nothing that could be described as destructive or wrong. In much more recent tests I've done -- in fact, just one today -- I couldn't tell between the two at all. Either way it doesn't seem to matter much.

However, in order to alleviate my "audiophile nervosa" doubts which does creep up from time to time, I will likely continue using my FIR phase correction for the surrounds.

Now, if I inserted a completely different type of center channel -- esp. one that isn't phase linear as my existing default coaxial LR channels, say, a passive 3-way -- then I'd be much more worried. I would, at the very least, make sure to match their frequency and phase response even though other important discontinuities like directivity and timbral characteristics are sure to crop up as well.

 

[DVDdoug]

This isn't something I'd lose sleep over...

ernestcarl uses the word "coherent". Phase is ONLY an issue when the sound from different speakers is identical from multiple speakers. Thats super-common for "centered" vocals and other "centered" sounds in a stereo recording and of course your left & right speaker should be identical.

In surround recordings it's more unusual to have the same-exact sound coming from multiple speakers. The centered-sounds are isolated in the center channel. There probably are some common sounds but in the overall mix it's probably not something that would be noticeable.

Distance also creates a time-delay which results in a phase shift. At 1kHz the wavelength is about 1-foot. A distance-difference of about 6-inches (about the distance between your ears) means the waves are 180 degrees out-of-phase (at 1kHz). It's very likely that your center speaker is more than 6-inches closer (to your ears) than your left & right speakers, and who knows about your rear (and/or side) speakers. And heck, I don't usually sit exactly in the center between left & right, and if there are multiple people listening they can't ALL be exactly in the perfect position... That's not even considering the reflected sounds in the room...


---------------------------------------------------------------------------------
I was doing some experiments with an SPL meter and some high-frequency test-tones (maybe 5-10kHz) in my home office awhile back - Just turning or moving my head slightly made huge loudness differences as the left, right, and reflected sounds went in-and-out of phase. It's something we don't really notice with music but it was drastic with test tones. The other interesting thing was that the SPL readings changed quite a bit when I moved around behind the SPL meter (changing the way the sound was reflected in the room). That's also something we don't notice with normal listening... We don't hear (or we don't notice) the sound changing when someone walks around in the room.

I don't remember what the experiment was about but something related to reflected sound... Something like adding "reflectors" to get wider stereo from closely-spaced speakers...

 

[EEE272]

I was wondering because with Atmos, there is a lot of panning between, e.g., front and surround and surround back.

For active speakers, I can see that there is a way to adjust the phase but for passive that seems difficult. For different speakers, I can also see that it will not work well.

But some manufacturers offer sets, where in the same series of speakers have different xo and even membranes. I asked the manufacturer and they just said that within a series, they optimize crossovers and phase to avoid issues and it should not be possible to hear it. Yet, technically, I was wondering if this is even possible. Probably @ernestcarl is right that they try to keep them close and the effect is then potentially indeed very low.
Maybe I could test it by playing the same pink noise on both speakers.

In any case, thank you for all the answers!