FR is one thing, but Phase Alignment is where the Imaging Lives . . .

Steve_Lee

Been playing around with XO types, slopes and amp levels with these Eminence Beta10CX, Selenium D202t CD's and a 12" powered sub lately and wondered what all the fuss was about concerning phase alignment.

So I looked at my MiniDSP 4x10 unit closely and the only thing related to phase (aside from the polarity swapping button) that I could discern was the XO types and slopes.

By ear after having flattened the FR I did some critical near field listening while changing the slopes/XO types and it all became clear.

Subsequently, I focused upon the Sub to MW XO 1st and when it sounded more like the sub was no longer localized and became a part of the MW I moved to the MW to CD (tweeter) XO.

Once the presence and stage imaging became the most pronounced I moved back to the sub to MW XO to finish-up the settings.

All I can say is, WOW!

Unfortunately, these target XO's are currently financially un-affordable and physically un-obtainable to replicate in a passive XO.

Anyone else get this sort of observations from your experiments using DSP?

Can good phase alignment be obtained from low order (less complex) XO's with these drivers?
(Maybe this ^ should be my next experiment since I have the tools to determine the answer to my question).

Wolf

If we knew what it was you did, we could likely make a matching passive xover

jr@mac

Symmetry is where imaging lives, and part of that is phase alignment. Huge improvements in imaging can be realized with room treatments to minimize differences in reflections from left to right. Ensuring the speakers themselves are very closely matched in linear distortion is critical, as well. This is why L-R balance and tone controls have been fundamental EQ options since the advent of stereo. Modern room correction software takes these to the next level, but cannot correct for time of flight differences between speakers when the off-axis reflections are not controlled properly.

The music you listen to has to have it embedded as well. Early stereo recordings were straight up novelty stuff - think early Beatles albums with John singing in one channel and everything else in the other channel. You cannot use that music to test for imaging. It took a very long time for the engineers to figure out proper stereo imaging - which is why very early on 3-channel was recommended. For many reasons, the 2-channel standard was adopted.

Regarding phase specifically, for the most part anything active can be duplicated with passives since phase alignment is nothing more than slope over delay. A common mistake people make when using active to design passive is not accounting for the delay in the passive - the slopes are not adequate when modeling the passive. There are numerous ways to introduce delay in a passive crossover but generally speaking and the reason the 4th LR is so popular is the reverse null at the XO point is a quick and reliable method to visualize phase alignment.

Anyways, sorry for the wall of text but I would argue shooting for symmetry first via FR and room reflections, good music second, and other factors after that. The "smearing" talked about when discussing imaging as it relates to phase should still be identical channel to channel.

A very good way to test is to create a mono pink noise track and listen in the sweet spot. Then, create a stereo track consisting of the same pink noise - effectively creating a stereo mono track. It should image the same since the summed stereo waveform is identical to the mono. You can really troubleshoot a room/system using this technique.

Steve_Lee

@Wolf said:
If we knew what it was you did, we could likely make a matching passive xover

Wolf - How exactly would I communicate "what I did" (with the DSP) in order for this to happen?

dcibel

@Steve_Lee said:

So I looked at my MiniDSP 4x10 unit closely and the only thing related to phase (aside from the polarity swapping button) that I could discern was the XO types and slopes.

Forgot the most obvious phase related feature, delay

By ear ..

How have you concluded that the acoustic effect of your adjustments are primarily a result of "phase alignment"?

@Steve_Lee said:
How exactly would I communicate "what I did" (with the DSP) in order for this to happen?

Start with the transfer function, easily done with a PC soundcard and REW or ARTA, just connect from soundcard output to DSP input, then from DSP output to soundcard input.

Alternative without measuring is to simply input the DSP settings to VituixCAD using active blocks, and make sure miniDSP is selected as the DSP system in the options. VituixCAD will show the complete transfer function, which is an oddly absent chart in miniDSP software.

Assuming you've stuck with IIR filters and haven't gone bonkers overcompensating every little wiggle in response, something rather similar can be replicated passively.

For the VituixCAD example, here's the internal structure of the basic miniDSP2x4:

And here's what that looks like to be completely reproduced in VituixCAD. Here, the PEQ blocks can be anything really that's supported within the DSP.

PWRRYD

Yes delay for sure! Especially with compression drivers mounted to horns where the Z axis is greatly behind where the midbass is.

Steve_Lee

Try this Dcibel:

I did this ^ as an exercise but don't actually need a passive XO for this config - however, I do have another 3-way in the works that I would like to have passives made for - if this exercise works as a communication/translation then I will configure that new 3-way using the Mini DSP and document the settings this way for others to play with and use as they wish.

Thanks for your very detailed response, sir.

Awesome group here . . .

dcibel

Quite the filter you've got on that tweeter, are you sure about the 12dB @ 6kHz correction?

The delay values you've got there are interesting. I have to assume that the time is in milliseconds, not seconds as shown based on the distance of 0.007m and 0.01m. The 10mm delay isn't going to make a lick of difference for the subwoofer, so you effectively have ~3mm delay between the midrange and tweeter, or 1 sample at 96kHz.

Steve_Lee

There is no delay on the Subwoofer, dcibel - check the worksheet again, please. (A = Attenuation, right?).
I'll change the worksheet to reflect ms instead of s for delay.

I should probably go capture some fresh sweeps and post them here before we go any further - I'll get it set up shortly . . .

dcibel

@Steve_Lee said:
There is no delay on the Subwoofer, dcibel - check the worksheet again, please. (A = Attenuation, right?).
I'll change the worksheet to reflect ms instead of s for delay.

Yes, my point was that there is no point in using delay in both the midrange and tweeter, the 7mm delay in the midrange is insignificant to the subwoofer, so might as well have 0ms delay in the midrange and 3ms delay in the tweeter, which was my "effectively 3ms" comment above.

Steve_Lee

I'll give that setting a go by setting midrange delay = 0 and changing nothing on the tweeter (CD).

dcibel

reduce both by 7mm to keep same relative delay...

Steve_Lee

The horns in these coaxes are so shallow it doesn't make much difference in the delays unless I set them for long delays (1" set-back on Z axis) and then it sounds smeared and less well defined so I set the mids and CD's to zero.

I'll update the worksheet and post a FR sweep soon.

rjj45

I have not done this, but shouldn't you be able to align "time of flight" by matching the rising impulse of the woofer and tweeter with some delay?

Steve_Lee

@rjj45 said:
I have not done this, but shouldn't you be able to align "time of flight" by matching the rising impulse of the woofer and tweeter with some delay?

Probably, RJJ but I am new at this and learning from trial and error plus direction from you guys and the learning curve steep while progress is slow . . . but I keep going. Thanks for the suggestions . . .

Steve_Lee

OK, I redid the XO's subsequent to the delays being set all to zero.
This result is from my ears and the graph say it is a keeper in my space/room/dungeon/redneck-laboratory;

~20* off-axis, 20" away:

Updated Worksheet - revisions in RED (I did adjust one (1) PEQ) . . .

dcibel

@rjj45 said:
I have not done this, but shouldn't you be able to align "time of flight" by matching the rising impulse of the woofer and tweeter with some delay?

https://www.minidsp.com/applications/rew/measuring-time-delay