Full range line array for wall or corner placement

You can see the ringing is still in the IR, but not in the windowed IR, window defined by 5 cycle FDW. So the IR really is ringing and not due to any EQ filtering as the ringing appears in the raw responses also. What could make it ring?

Could it be low damping factor due to 8 drivers in series? I don't want to rewire the array only to find that wasn't the cause. Perhaps I could put some R in series with the amp and see if it gets worse. But I also had this ringing in my Synergy project so perhaps its something in the sound card.

Coming along nicely! Are you fully pleased with the side panel? It looks well done.

re ringing....Unless you have a soundcard issue, my thinking is it's multiple arrivals along with their combing, showing up in the impulse response.

Try close mic'ing and see if they diminish.
Also, if you can mic at greater distance, to see if the squiggles move closer to t=0 (less distance variation to drivers)

Oh, can you get a coherent phase trace, or at least easily readable, with REW without using FDW? (after shifting the impulse response)
I can't.....FDW has been mandatory ..
 
Those FRs are all with FDW.

Yes, i realized that. My question is about phase traces without FDW.
Are they readable?

I can not get REW to produce readable phase traces, after shifting impulse to remove group delay, for my array.
I attribute it to multiple arrivals, not knowing what else to blame.
Which also leads me to also attribute what appears to be ringing in the impulse response, to multiple arrivals.

I remember the measurement files ra7 kindly posted....phase couldn't be read without FDW with his either . So I'm very interested in yours, wesayso's, any others experiences here...thx :)
 
Back home now. No more interruptions from darn eye doc.

Yes phase is ugly w/o FDW but its not due to being in an array. Its due to latent reflections in the environment.

attachment.php


The array phase is a little bit messier but I'm better able to absorb around the small test box than the tall array.
 

Attachments

  • Compare test box to array no FDW.jpg
    Compare test box to array no FDW.jpg
    339.2 KB · Views: 879
Yeah, eye doc is about to get me in the shop too...
Hope all is well..

I'm going to try REW on the array outdoors...hopefully today.
I hope I'm wrong about early arrivals messing up impulse just due to multiple drivers.

Smaart, that i normally use for measurements, does produce a decent phase trace even indoors. No clue why it does and REW doesn't...although i know smaart is routinely used to measure large live-sound line arrays.
I've done so, helping sound providers align their lines with their subs, prior to concert.
Lines are often tough to pinpoint acoustic distance...multiple sources again....
This is probably why I'm in' multiple driver suspect mode' for our arrays.....
 
The ringing you see can be one of many things, lets start by looking at a generated Dirac pulse (generated with Rephase):

attachment.php


As you see there's even some ringing there (due to the abrupt end of the impulse).

Now put in the DAC, and loop back to see how that performs within REW. Guaranteed it will not look any cleaner than the Dirac pulse. Next put in the amplifier...

But, the series of pulses in the first raw measurement:
757684d1558281985-range-line-array-wall-corner-placement-ir-garage-door-jpg


That is a series of peaks of the separate drivers arriving at the microphone.
Once we EQ out all of the dips and peaks, you'll immediately see the clean up in the impulse. The impulse is "just" a rendering of the total data with an emphasis on high frequencies. EQ-ing this flat isn't that easy, as one would need quite a bit of resolution in the top end to see the summed result of all those tiny pulses.
It's easier to run it trough DRC-FIR, with enough cycles up top at high frequencies so it can try and correct it back to a 'Dirac like' impulse. If you put EQ on top of that (for instance with the waterfall plot open with very short rise time and short window settings) you can EQ it to close to perfection in one single spot. Which will only be valid at this one spot. However, the result of for instance DRC-FIR on the impulse will be far better than what REW can achieve with EQ automation (much too course a correction is applied within REW).
REW's phase plot follows time. In other words, for every disruption we will see, there is something happening over time that creates this. If we remove these disruptions, the phase will show up to be quite normal (once we have fixed the raw pulse).
If another program shows great phase plots indoors, it won't accumulate the energy over time like REW does. I quite like what REW does here, as you can find these disruptions in all of the other plots (or ways to look at the same data) as well, as these plots are all derivatives from the original impulse. So first correct the frequency, preferably with a tool that tries to correct back to a Dirac pulse, next look at the other disruptions in the IR, as these will cause the phase to "wrap". Solve the reflection/diffraction completely and phase will look shiny and bright.
But it should not look like that indoors to begin with. Maybe in an anechoic chamber (preferably with a floor and ceiling to mimic the infinite array) after correcting the impulse at a certain distance.
 

Attachments

  • Dirac pulse.jpg
    Dirac pulse.jpg
    61 KB · Views: 880
Last edited:
The spread of those ringing peaks matches my calculation of the delta arrival times at the listening distance of 120" or roughly 3m.

The number of peaks also matches. I expect 15 or 16 because of symmetry about mic at mid height. There are 13 strong peaks and then count 2 or 3 more as the burst tails off.

But I also saw similar ringing in the IR of my single driver test box.

If I look at the raw unwindowed or FDWed response, I doubt all that HF hash can be EQed away but I can do better than what I have so far if I get the FIR path going. REW auto=EQ with Rephase is easy for me. DRC would be more difficult.

EQing should not affect the delta arrival times but should do good things for the IR and step

I did manage to integrate a sub today so I feel good about that....but my poor dual 10" sub has >5% THD at even modest listening levels. Have to switch to my TD15s....
 
Last edited:
So, inferring from fact that ringing is in both single driver and array, then its not due to multiple arrival times.

I took the measurement of the single driver test box, imported it into Vituix, and looked at its IR. I next used Vituix calculator to create a filter that made the FR perfectly flat.

Attached pics show the before and after
the measurement viewed with FDW5
attachment.php


The IR and step in Vituix
attachment.php


The simulated IR with a Vituix mirror filter
attachment.php


Its just as you said. A ringing IR just means there is more EQ work to do

Thanks!
 

Attachments

  • sb65 test box no eq.jpg
    sb65 test box no eq.jpg
    37.6 KB · Views: 737
  • sb65 test box no eq IR.jpg
    sb65 test box no eq IR.jpg
    25.4 KB · Views: 760
  • sb65 test box mirrored flat IR.png
    sb65 test box mirrored flat IR.png
    6.3 KB · Views: 736
So, inferring from fact that ringing is in both single driver and array, then its not due to multiple arrival times...

Sounds right :)

Have couple of points that maybe can be reason some of the HF hash seen, one is the relative hot ringing and high HF reach on axis for SB65 8kHz and up another is for many sound cards of past decade it had been modern argument their sharp low pass filters should be of minimum phase instead of using linear phase filters.

Here is HF hash of SB65 (orange) verse TC9 (blue) on same big cardboard baffle, sample rate for session is at 192kHz and had it been at 44 or 48kHz guess hash would be much worse:
attachment.php



Her is loopback session at 44kHz with a quite old souncard using linear phase low pass filter (green) verse a moderne one using IIR low pass filter (gold):
attachment.php
 

Attachments

  • 1007.png
    1007.png
    12.7 KB · Views: 749
  • 1006.png
    1006.png
    33.8 KB · Views: 750
Last edited:
Valid points made by BYRRT here, which is why I said the ringing could be caused by many things. Another note, when you start correcting, you'll want it to sound good. Preferably everywhere. That is why I encourage the use of a package like DRC-FIR.
It gives you many variables that help decide what should be fixed and what shouldn't.
Most of my time was spend on finding my own preferred mix of setting to achieve just that. Not to make the prettiest graphs, but to get the best sound.
You've got to be aware of the restrictions. Fixing the impulse with all it's warts is probably only valid for the exact point you've measured at. There are lots of ways around that, such as using measurement averages etc. But also the internal settings and variables can help optimise the results without overdoing it.
I've spend a lot of time to figure out that part. Don't stare at IR's alone, use the other representations to learn what belongs to what. See how the waterfall changes over time to learn what happens at the listening spot etc.
In my theory I want the best first wave front I can get, including FR and phase. But the tonal balance is set over a much larger window. The mid/side actually EQ is a logical extension of that theory.
 
Thanks BYRTT and Wesayso for helping me grasp this critical point about the ringing. We all know that IR and FR are related but its not so clear what ringing in the IR says about the FR; too used to just looking for evidence of reflections.

re' DRC
I'm with you Wesayso. At the same time you are way ahead of me. I still have another speaker to build and I need to fashion a base for the first one. I will no doubt need your help finding my way through new software and dealing with these higher level issues when I get there.

I think I will have a much easier time with the room than I did last time. What I get at 3m listening distance in the garage is amazingly clean, way cleaner than in my previous home. The garage has a few bits of absorption but all kinds of dispersion, filled as it is with my toys and tools. The problem is it has very little free wall space for the speakers. I will have to set them up in front of the garage door, tame them and the room with JR and DRC there, and only then bring them inside. Fortunately, they are easy to move weighing only 35 lbs fully loaded.
 
Having flattened my test box response in simulation I'm wondering if there is a quick and dirty way to do the same for the array but in the real world.

I could take the measurement of my array plus sub, which is about +/- 3db for most of the range, create a mirror filter to flatten it the rest of the way, modify the mirror filter to keep from over-boosting the sub, then bring it into rephase to create a convolution file to be executed in JR.

Thoughts?

Maybe I should just dig into DRC.
 
Hi guys, had some time to test my 'multiple arrivals' hypothesis...as the source of 'ringing'
So I started ground up with a soundcard measurement to eliminate that variable.
Then i just built upwards from single driver on an open baffle, to the full array, which has 6 groups of four TC9's per section.
Measurements were out on a deck with the mic about 10ft away centered on the array. Not as reflection free as out on driveway, but all i had time for....


Following plots in order:
Sound card
Single TC9 on open baffle
Single 4-driver section from array
Two 4-driver sections
Four 4-driver sections
All six 4-driver sections
 

Attachments

  • soundcard.jpg
    soundcard.jpg
    45.4 KB · Views: 98
  • single driver.jpg
    single driver.jpg
    51.2 KB · Views: 101
  • single 4-driver section.jpg
    single 4-driver section.jpg
    53 KB · Views: 106
  • two 4-driver sections.jpg
    two 4-driver sections.jpg
    51.4 KB · Views: 102
  • four 4-driver sections.jpg
    four 4-driver sections.jpg
    56 KB · Views: 104
  • all six 4-driver sections.jpg
    all six 4-driver sections.jpg
    58.1 KB · Views: 107
Last edited:
Thanks for doing those tests and uploading mark100! From your screen captures it appears there is more ringing from the full array vs the single/small array.

You bet !

My take is that we can correct the squiggles under 200us....I think that's diffraction. My open baffle was the same width as the array baffle.

Past 200us is multiple arrivals combined with their comb filtering, imo.
I don't think it can be corrected.