Using REW to Determine Time Delays Between Drivers

[rplace]

I've spent the last many, many weekends (and weeks). Measuring, adjusting, repeating and even started over from scratch at least twice due to what I have learned from Chris and stumbled across blindly.

 

Last Sunday and this morning I've just been listening to 2 or 3 different presets...each with different plusses, none yet have I gotten like Chris'. All 3 put a big smile on my face. Odd thing is what I had before sounded better than anything previous. Now its even better. I whish I still had my passive Khorns just so I could give them another go and make sure I'm not just wanting them to be better.

 

I'm still having problems getting the impulse right. Seems like it just is what it is based on the other settings. @Chris A is there specific ways to influence it or will it just sort of fall into place when you get all the "other stars aligned"?

[Chris A]

Getting better impulse response with your drivers may result in a balancing act between the crossover frequency and the phase growth on the lower-frequency end of its passband.   In order to get a more vertical impulse spike on the spectrogram, you may resort to a separate midrange driver/horn to flatten the phase from 1-2 kHz crossover point down to the crossover frequency with the bass bins.  The Danley Synergy series of MEHs uses a "phase link" approach with their midranges there mostly to flatten the overall phase response in the midrange over perhaps one octave of the midrange driver's passband, and therefore improve the overall impulse response of the loudspeaker.

 

The B&O BeoVox 120.2 (1983-1988): one of the many “Uni-Phase” loudspeakers with a woofer, tweeter and “phase link” driver.

 

 

The Low Pass (black), High Pass (blue), and Phase Link (red) components of a Uni-Phase loudspeaker.

 

Remember that there are MEH loudspeakers using Oris-type horns and MEH configurations:

 

 

Chris

[Chris A]

One other way to get perfect impulse response with your current setup is to use FIR filtering, which requires electronics capable to doing the extra computational load:  https://www.diyaudio.com/forums/multi-way/221434-rephase-loudspeaker-phase-linearization-eq-fir-filtering-tool-98.html#post4679537

 

Chris

 

[Chris A]

On 10/27/2019 at 7:29 AM, Chris A said:

I can now see what the issue is...calibrating the impulse response zero time in each plot.  Below you will see an SPL/phase measurement just after being taken, with the Controls menu turned on to the right.  There is a button on the Controls menu marked "Estimate IR Delay" ("IR" = impulse response).  You need to push that button, then a window will pop up called "Delay estimation complete", and you push the "Shift IR" button on that menu.

 

 

Then you return to the Controls menu and push the "Generate minimum phase" ("3")  button, then the "Unwrap Phase" button ("4")...

The latest beta release of REW (V5.20 beta 29) changed the shape of the controls menu within the SPL & Phase plot window.  Below you will see the new layout with the above notation for the sequence of button pushes transcribed into the new menu format:

 

 

Chris

[schlager]

Did this thread come to a grinding halt? And was there ever a tutorial released about the topic at hand?

[The Dude]

Need to add this to favorites

Sent from my SM-G986U using Tapatalk

[bmeeks]

@Chris A I have read this thread 3 times and have learned many things and feel so more confident in using REW.  So first off thank you for all your time and effort and last do you plan on releasing the document you spoke of to help others.

[Chris A]

I believe that the reason why I halted part two of the tutorial was two-fold:

 

1) I found that few people seemed interested, judging from the lack of response from the participants here in early winter 2019 into early 2020.  So I put the phase-flattening tutorial portion on hold, at least until I could get a bit more time using the technique on other people's setups.  I was planning on giving a face-to-face class in Hope, AR in the spring at the SWAG, but my participation in that event was curtailed due to the arrival of SARS-CoV-II about that time. 

 

Since then I've dialed in perhaps a dozen other setups via email of REW measurement files and DSP crossover configuration files using this technique, and have found that it works very well.  I've since added a couple of modifications to the technique for those folks that apparently like to play their setups "really, really loud"--to add a bit more downslope to the drivers response just outside of their passbands so that there aren't audible artifacts of out-of-band driver response that can be heard at well above 100-105 dB levels in-room. 

 

2) I lost my saved REW measurements used for this tutorial in March 2020 due to a crashed hard drive.  While I had backed up my demastered music files, etc. to another external HD, I apparently neglected to back up my REW measurements directory, thus losing perhaps 3 years of measurements from my last real REW backup.

 

I have a very rough draft on the "part II" tutorial that I fortunately backed up in January 2020 which is still too rough to release but due to the loss of the measurements in March, I've put this work on hold until I take new measurements to replace the ones lost. 

 

Generally, that task is still there waiting on me, and seeing the interest in this subject being revived here, I'll try to restart the effort and replace the measurements.  It will take a while, however.  In the mean time, a very abbreviated description of the method is reproduced here:

 

Quote

So all you have to do is put the HF and LF drivers together without phase shifts:

1. Don't use the "crossover filters" that come with DSP crossovers--clear any crossover filters if they're set.
2. Set the HF or LF channel delay to get perfect impulse response in the time domain--as seen in the spectrogram view.
3. Flatten each driver's SPL response within their pass bands.
4. Match the channel gains between flattened phase drivers.
5. Use output channel PEQs to trim off response on each end of the bass and high frequency drivers until you've got overall flat SPL across the crossover interference band and smooth handover of SPL vs. frequency.  The drivers themselves will tell you where that transition/crossover should occur.  [If you're using MEHs, you'll have to use multiple PEQs to attenuate the bass bin peaks in response above the first notch frequency.]
6. Use the input channel PEQs to further flatten the overall response within the interference band to correct any dips or peaks in response within that band.

Voila!  Flat phase.  It's really that easy.

Knowing that the description above is probably way too aggregated for many to follow, I plan on continuing this tutorial effort after getting my woodworking shop in order and completing some MEH work that I've promised to others.

 

Chris

[Chris A]

I just realized that I hadn't posted Part I of the tutorial.  Here is a very rough draft (i.e., not yet proofed via users):

 

Dialing In DSP Crossovers-- A Tutorial-Part 1 rev 1.pdf

 

 

Chris

[bmeeks]

I actually was in the process of getting ready to build some new speakers when i found this thread, its really hard to find this type of documentation on proper alignment of active speakers anywhere on the net.  So far this is where I am at with my new build these are 4 way active towers and the measurement is outside of my garage mic 3 feet off the ground and 3 feet away from speaker pointed directly at it with crossovers and delays in place and 3 PEQs in each speaker/way.  I definitely still have an issue with my last way the bass bin which is crossed currently at 120HZ to the mid bass driver but i am getting there slowly.

[bmeeks]

Here is a picture of the progress of my build Morel ET338 with outer ring trimmed to get Morel EM308 as close as possible down to the mid bass Dayton 8' Epique then to the bass dual 8 Dayton HO 8ohm wired in parallel to get the closest sensitivity. Crossovers are 4k LR 12db/ 1k BW 12db/ 120hz BW 12db using minidsp 2x4 hd for each tower.

[Chris A]

10 minutes ago, bmeeks said:

...its really hard to find this type of documentation on proper alignment of active speakers anywhere on the net.

Yes...that's why I created the threads. 

 

I'm actually surprised that the major DSP crossover manufacturers haven't really stepped up to the task to provide hands-on "how to" videos and written tutorials.  I found nothing that was useful as I looked around for material that might help others come up to speed.  Even the book/linked pdf by Mitch Barnett (Accurate Sound Reproduction Using DSP) wasn't at all useful in these particular areas.  Mr. Barnett also selected DSP and measurement tools that most people aren't using (i.e., high-cost). 

 

Finding nothing actually useful, I simply wrote down how I do it from techniques developed over the years.  The little thin book by D'Appolito from the late 1990s--Testing Loudspeakers--was the single best source that I've found, and even that has some limitations.

 

I've been using DSP crossovers for about 13 years now and helping others dial in their setups remotely via email round-robin (REW measurements emailed-->DSP crossover configuration/preset files emailed back), repeating until their systems are in dialed in...for about 4 years now. 

 

I find it's like riding a bicycle--once you see how to do it, it stays with you and is comparatively easy to master once becoming able to do the basic dial-in tasks of time alignment and using PEQs to flatten SPL response.  Everything else comes along as you dig deeper into issues you see in the measurements. 

 

Chris

[bmeeks]

The worst part is when I google search terms active speaker alignment this thread doesn't show up, if somehow we could get this thread properly worded for google keywords this thread would blow up.  I am very new to speaker building and learning 3 years so far and this speaker is my 4th iteration of trying to find the sound i am looking for.  I have bought many books and read for 100's of hours trying to figure out all the things it takes to build a good loud speaker.  The problem is so many people talk about all the attributes needed for a good loud speaker but don't always have the answers of how to get to them.  I have listened to many people considered professionals and when you try to get down to the brass tax of how to accomplish some of these things they are not always able to explain as you have throughout this thread.  So again a huge thank you to all of your time and efforts of helping me and all the others, your expertise are invaluable.

[Wirrunna]

Chris, I agree with what you said above which is why I have spent a day or two going back over the rough procedures that I wrote down to dial in my DCX464 in my K-Horns with a miniDSP and have assembled REW measurements, EQ settings and the resulting REW measurement and RePhase settings and the resulting REW measurements for each driver. Today I intend to set up the measurement microphone again, get into the miniDSP and turn off delays and level matching and run some REW measurements to show before and after plots, spectrographs and GD. I have expanded the resulting document and will break it into into a series of .pdf files rather than a great long thread full of REW plots and screen captures.

I suspect that a lot of members of the forum are daunted by the technical challenge of adding a DSP and a couple of amplifiers to go "active".
I have the motor board from my K-Horn's top hats, complete with original K401 / K55M and K77M, it is only 4 screws to swap  this from the DCX464 / Eliptrac motorboard. So I'm thinking hard that it may be worthwhile doing an AK3 to Active project some time and document it. The resulting REW measurements and EQ, RePhase tweaking and miniDSP settings from a K-Horn would be good starting point for a La Scala or Belle .

[Chris A]

^^^^^^^^^ What he said.

 

28 minutes ago, Wirrunna said:

I suspect that a lot of members of the forum are daunted by the technical challenge of adding a DSP and a couple of amplifiers to go "active".

One of the reasons why I offered my dialing-in services free of charge was this perception.  I don't believe that this is something that should deter anyone.  The gain is much larger than the risk, in my experience.

 

Chris

[Wirrunna]

Chris, I have added the remaining measurements including the EQ'd but no delay in the miniDSP.

 

Now, a question - how flat should the phase be ? Certainly the first music I played back in early January when I got this flatish phase cross over working was a revelation. There were some tracks that you could swear it was a mono system and the speaker was disguised as a slow combustion wood fire sitting near the wall in the middle a long way from the two corner cupboards with Klipsch labels. While other tracks had instruments coming from further out than the corner cupboards. And Dark Side of the Moon when the alarm clocks went off frightened everyone.

Is there some acknowledged measurement for phase, e.g. + - 30 degrees from 200 to 10,000 ?

[Chris A]

15 minutes ago, Wirrunna said:

Now, a question - how flat should the phase be ? ...Is there some acknowledged measurement for phase, e.g. + - 30 degrees from 200 to 10,000 ?

I really don't know.  I had assumed that once you get under 90 degrees of total phase swing (+ or -) that the effects of further phase flattening should be pretty much impossible to pick out.  I infer this from playing with multiple entry horns (MEHs) where the only real requirement is that all the drivers' throat exits must be within 90 degrees of phase...or cancellations will begin to occur that affects the measurable SPL response from within the aperture of an MEH (different sources within the horn acoustically couple and sum with each other). 

 

I also don't know about the effects of "higher order modes", HOMs, but these are (I find) not issues for pyramidal shaped MEHs.  Here the effects of Huygens-Fresnel principle apparently control the physics of the horn, where wavefronts are created at the inner surface of the horn, and anything higher order behind the surface of the horn gets reflected back to the source and not passed through the horn entrances (i.e., throat, off-axis ports).  The only horns that I've found having audible HOM issues are those with slots in their throats and a long chamber from the slot exit back to the phase plug/throat of the compression driver itself that apparently acts as a resonator chamber.

 

So the subject of audible phase distortions is one that is also wrapped up with Huygens-Fresnel re-radiation from a single horn aperture in MEHs, too.  If you can hear less that 90 degrees of phase swing (...and it might be possible...) then this would have implications on MEH design...a subject of interest. 

 

On the other hand, if the audibility of phase (especially within an MEH) gets homogenized into one source within the 1/4 wavelength rule, then life gets considerably easier.  This is apparently what occurs, at least in my experience and readings of the experiences of Tom Danley.

 

Chris

[Chris A]

There is one other effect or observation that I believe is relevant: the audibility of phase shifts from speech, etc. to form spikes in the resulting waveforms at the eardrums.  A PowerPoint slide from David Griesinger:

 

Quote

Once in every fundamental period the PHASE of the harmonics align to form a peak pressure.

⦁    If there are four harmonics inside a critical band, once in every fundamental period the pressure increases by a factor of four, giving a 6dB increase in the signal to noise ratio.
⦁    If the (cochlear) hair cell outputs are sent to an autocorrelator with a length of four periods, there is an additional S/N improvement of 6dB.
⦁    A 12 dB advantage in S/N is an enormous advantage
⦁    Speech recognition algorithms are just beginning to realize the importance of PHASE!
⦁    Because both clear speech and music waveforms have SPIKES!

 

Original waveform (spoken "one, two"):
 

Same waveform with phase shifts:

This is really the root issue with audibility of phase shifts...

 

Chris

[Khornukopia]

On 1/28/2021 at 4:42 AM, Chris A said:

• Don't use the "crossover filters" that come with DSP crossovers--clear any crossover filters if they're set.
• ...
• Use output channel PEQs to trim off response on each end of the bass and high frequency drivers

 

Do "crossover filters" add or introduce phase shift on purpose? Why?

 

Does an equivalent "PEQ filter" not add any phase shift? Why not? 

 

I am trying to mentally visualize how the difference occurs during the electronic process if the user selected filter frequency and slope are exactly the same.

[Chris A]

41 minutes ago, Khornukopia said:

Do "crossover filters" add or introduce phase shift on purpose? Why?

Yes.  At its simplest, it's the inductive phase lag in the low pass filters and the capacitive phase lead of the high pass filters that combine to form 90 degrees of phase shift on the lower frequency drivers relative to the higher frequency drivers.  The following is a plot of the amplitudes and phase shifts of a so-called Bang & Olufsen  "phase link" loudspeaker:

 

 

Note the 90 degree phase shifts of each "way" of the loudspeaker, summing to 180 degrees of crossover filter-induced phase lag from the woofer to the tweeter.

 

41 minutes ago, Khornukopia said:

Does an equivalent "PEQ filter" not add any phase shift? Why not? 

PEQs (active crossovers) do not introduce systemic phase lead or lag to the entire channel relative to the other channels, rather, their phase effects are local to the vicinity of the notched (or boosted) frequency.  The following -25 dB PEQ (red trace) at 1 kHz from Xilica's XConsole application shows this phase response (white trace):

 

 

The equivalent plot using a second-order high pass Butterworth filter (12 dB/octave):

 

 

The steeper the crossover filters used, the more phase lag on the lower frequency drivers relative to the higher frequency drivers--45 degrees of lag per order of the filter (high pass).  When combined with a symmetrical low pass filter, the total phase lag is 90 degrees per order of the combined high-pass/low-pass filters on the lower frequency drivers.

 

Chris