Active, DSP, REW, Xilica 101

[Edgar]

19 minutes ago, rplace said:

Are we saying that we need to know more about the two phase plots below (orange and green) to determine which one is better, or maybe that neither is necessarily better than the other? I was under the impression that less wraps = better.

 

It's hard to say unless you "unwrap" the phase, which I did rather crudely below. The orange plot shows a lot of delay above 6 kHz, but beyond that, what you're looking for is how straight the phase line is (even if it's tilted). It's not the slope that causes problems, but the changes in the slope.

 

[rplace]

7 minutes ago, Edgar said:

It's hard to say unless you "unwrap" the phase

 

I can unwrap them, they just don't fit on the screen and I don't know what to look for in the unwrapped version. I've just been treating this like a rubix cube or a puzzle to be solved where the goal was to make the phase have zero wraps from 20-20,000. I'm missing the actual theory here. My cave man speaker building brain just thinks"Phase wrap bad, flat line good".

 

 

 

 

10 minutes ago, Edgar said:

The orange plot shows a lot of delay above 6 kHz, but beyond that, what you're looking for is how straight the phase line is (even if it's tilted). It's not the slope that causes problems, but the changes in the slope.

 

Where/how can you see the delay? For sure my line above is NOT straight.

[Edgar]

1 minute ago, rplace said:

Where/how can you see the delay? For sure my line above is NOT straight.

 

The sudden change in the trend from a slight upward tilt below 6 kHz to a severe downward tilt above 6 kHz indicates that the driver delivering frequencies above 6 kHz needs to be moved forward (closer to the microphone) because its signal is delayed. The individual wiggles are caused by various local frequency response effects, but the trend of the line is caused by timing.

 

The upward trend below 6 kHz probably means that the midrange driver needs to be moved a little farther back (away from the microphone).

[Rudy81]

Isn't that where the whizzer cone kicks in? Mine does that as well.

[Zim.]

x

[Chris A]

1 hour ago, Zim. said:

...how can one say they have improved on settings derived from an anechoic chamber?

Who said that? 

 

What I've found is that by paying close attention to the excess group delay curve, you can effectively see when you've got a minimum phase condition, and therefore know when in-room measurements are effectively approximating anechoic measurements.  My ears tell me that it all works extremely well, but that you need to be careful below the Schroeder frequency and with boundary peaks/cancellations of the microphone position from walls/floor/ceiling (usually around 100-300 Hz). 

 

When you change something in the loudspeaker itself, or you're trying to EQ the minimum phase aspects of placing loudspeakers in a real reflective room, you need something in-room to measure, or another anechoic chamber (which still doesn't release you from doing in-room measurements).  I've learned that in-room measurements and ears works a couple of orders of magnitude better than trying to use something like tone controls and ears only--which is apparently the alternative--right?  The combination of using ears and having the measurements in-room is like the difference between the living and the dead over not having the measurements.  Over time, I've come to trust the measurements even more.  Paying attention to time gating and direct-reflected conditions also help a great deal.

 

Moving the microphone around a bit and taking extra measurements, then averaging will also tell you if you've got a reasonably good in-room microphone/loudspeaker/room position, too, which is precisely what I did to find the best microphone positions in-room.  Once good microphone position(s) are found, the microphone can be put back into its approximate location again and again, getting repeatable results.

 

Chris

[Chris A]

1 hour ago, Zim. said:

Unless one plants their ears exactly where the measurements were taken,  why bother?

You don't mean this, do you?  This is a prime example of argumentum ad ignorantiam. I'm assuming that you don't measure...?  Don't know how?  I can help.

 

Chris

[Zim.]

x

 

 

 

[Chris A]

1 hour ago, Zim. said:

I appreciate your response but would genuinely value an answer to my question.  

This is an interesting subject. 

 

I can answer it this way:  If you do a reasonably good job of measuring and paying attention to the "non-minimum phase" issues at the measurement position(s), you can usually do better than anechoic --below the Schroeder frequency in-room (usually, this means below 200 Hz, but in my room, its more like 100 Hz). 

 

However, if you've got access to an anechoic chamber for the loudspeaker in the same configuration (drivers, horns, relative position of the horns to each other, etc.), that data is usually a little bit better, if you've taken the time in the chamber to really flatten the response to ±1.5 to ±1 dB above the Schroeder frequency of the room you're going to put them in.  If you haven't taken the time in the chamber to really flatten the response (including running out of having enough PEQs using Dx38 crossovers, etc.), then using in-room measurements is the next best thing to anechoic SPL flattening. 

 

If however you have trouble keeping the early reflections out of the measurements, and/or somehow convince yourself that the corrections that are identified in REW's EQ facility are not needed because "it's too much EQ" (I see a lot of this kind of rationalization on the forum and elsewhere, without any backup as to why someone might think this, much less prove that it's true), then you can find yourself chasing your tail--as I did around the crossover frequency of 425 Hz and the microphone/floor/wall boundary cancellation frequency.  It wasn't until I realized that I could correct these areas with excellent results that I began to find even better performance. 

 

What I've found is that the flatter you can make the SPL on-axis (and by association with the minimum phase argument--correct the off-axis behavior, too), the better it sounds.  This is from Floyd Toole, and I have to say that he's right on.  I didn't try it, though, until perhaps a couple of years ago.  It really got my attention when I turned on a high quality stereo recording (and multichannel recordings later on after I reworked the other loudspeakers in the setup) and felt much more presence in the entire setup.  This was unexpected.

 

1 hour ago, Zim. said:

 

3 hours ago, Chris A said:

This is a prime example of argumentum ad ignorantiam

What is it that cannot be proven?  I don't follow what you are saying...

What I mean is that the idea of shifting the burden of proof to the guys that are doing the measurements in-room and updating the DSP crossover PEQs/crossover filters is probably not what you intended to imply.  The burden of proof is really on those that say that it doesn't work (which I find is quite contrary to my own experience over the past 5 years or so).

 

1 hour ago, Zim. said:

Did I misinterpret?

Yes. You imply that there are anechoic settings for the configuration that I was discussing in the text that you quoted.  To my knowledge, there aren't such anechoic measurements.  I developed the fractional-order crossover Xilica DSP settings from in-room REW measurements of shortened-down K-402s on top of Jubilee bass bins.  Those settings have continued to astound me even today through subjective listening (i.e., using my entire 5.1 setup), and the measurements have proven to be repeatable and stable in-room over multiple measurements/calendar time.  And they have produced the effect that I mentioned. 

 

I urge you to try it yourself since the cost of doing it is time only if you've got something like a Xilica or DC-One having enough PEQs/channel.  To not try it is actually saying to me (at least), "I don't believe what you're saying, because what you're saying doesn't have value, but I'm going to try to negate what you are saying anyway because I don't like what you're saying."  Frankly, that's unfriendly. If you don't like it but don't intend to try it--but feel compelled to express your negative feelings on the subject nevertheless,  I instead recommend other threads to read and participate in (just like I do--as there are many threads that I choose not to participate in because of the subject matter). 

 

If on the other hand that same person had asked for direct help using REW measurements and had received return email Xilica .xdat configuration files with updated settings, then s/he could say, "I tried it, and didn't like it", or "I tried it and there is definite merit to it", or even "I tried it and it really works--I like it. And it was also a very good learning experience: it cost me nothing but a little time."

 

Chris 

[Zim.]

x 

[rplace]

18 hours ago, Zim. said:

If minor changes in mic. locations significantly affect measurements, and measurements taken at the main listening position are inconsistent, how can one say they have improved on settings derived from an anechoic chamber

 

15 hours ago, Zim. said:

Yes, and I do apologize for the 090 level question.

 

10 hours ago, Zim. said:

I'm simply trying to better understand some fundamental aspects of the process

 

I'll take a stab at this, based on what you wrote I think this is what you are/were asking....often it is hard to tell intent from internet postings. You were thinking that if a given speaker was developed in a anechoic chamber and was now sitting in your room with all the room's flaws, how could you possibly improve on it. Especially, given my microphone placement questions. If so the key here is that these don't have a passive crossover or network. Also the speakers in question were not tested, developed and marketed as a finished product. There is no starting point that was good, great or mediocre to begin with. So there was nothing to improve on

 

By measuring nearfield (close to the speaker) you are getting what the speaker is doing in your room without the problems of the room. Measuring form the listening position has all the influences of the room. By carefully measuring and adjusting PEQ/XO points/delay/etc. you can (hopefully) get the speakers to perform the best they can. Treating the room can tame the bad effects as well, but that is a different topic all together.

 

I don't want to speak for @Chris A but I believe he may have thought you were like many others that show up only to cast stones. I hope that is not the case. People form opinions without anything to back it up. Kid: I don't like mustard. Mom: Have you ever tried it? Kid: no, but I know I don't like it.

 

I think he was just saying give it a try, see what you think, ask plenty of questions, form your own opinion. Don't just pop in and say "why bother" ie casting stones....which by the way I don't think you were doing, but I cold see how it comes off that way in the earlier post, not the latter.

[Edgar]

I'm going to try a kind of off-the-wall analogy:

 

I can go to the indoor range with my target pistol and hit the X-ring with every shot. Take it to the outdoor range, though, and suddenly I can't hit the side of a barn. Wind, humidity, temperature, and who knows -- maybe gnats and butterflies -- all make hitting the target much more difficult. I can employ some Kentucky Windage and mostly compensate for the new problems, but tomorrow I'll have to start all over again because conditions will be different. Or I can measure the wind, temperature, and humidity, and actually compensate for them. (Can't do much about gnats and butterflies.) And tomorrow I can measure them again and compensate for them again.

 

The point is that the performance under ideal conditions is just the starting point. In the real world we strive to duplicate that ideal performance, and measuring the things that affect the performance is a better way to do that than trial-and-error.

[Zim.]

x

 

[Edgar]

48 minutes ago, Zim. said:

  If only it was as simple for me as environmental influence on internal and external ballistics.....

 

Actually, it kinda is. In both cases the performance is strongly affected by the environment. In the ballistics case you have no control over the environment so you just try to compensate for it. In the loudspeaker case you actually do have some control over the environment, so you get your choice: compensate for the environment, alter the environment, or a combination of the two.

[Zim.]

On 12/18/2019 at 12:54 PM, Edgar said:

x

[rplace]

Is school back in session, @Chris A ?

 

I've run into a road block. New OB project that you are aware of. So far it  has been the simplest to date for me to flatten and get (I think) the delay right. No matter what I do I have problems around 3K to 4K. Over in @Rudy81 Rudy's double stack thread there was one phase "problem" on the 12" wings that you said to ignore as it was a measurement issue. I'm hoping this is similar and I can just ignore it. If so can  you please tell me why it may be happening and how/when you know to ignore stuff.

 

I've actually had it a lot flatter, but these recent measurements reflect me moving the speaker, absorption and even the microphone around trying to see if I can make all those rapid zig-zags go away. No luck so far. Below are the usual suspects of Phase/SPL, GD and Spectrogram. Drivers are dual Eminence Alpha 15s and Voxativ 1.6 for the HF.  

 

Extra credit if you can explain the bumps/humps in the Spectrogram around the same 3-4K off to the right of dotted line (what is that line called again). I think perhaps on the Oris it was reflections arriving behind the initial sound.

 

With minimal time spent it's looking pretty good with the exception of that GD that tracks with the zig-zags in the phase.

 

 

 

 

 

 

[Edgar]

26 minutes ago, rplace said:

With minimal time spent it's looking pretty good with the exception of that GD that tracks with the zig-zags in the phase.

 

Wow, that's weird. I count seven phase wraps centered at about 3500 Hz. That corresponds to a little over 2 feet of delay. And your group delay seems to extrapolate to about 10 milliseconds in the same region. What is your crossover frequency and order?

[rplace]

Just now, Edgar said:

 

Wow, that's weird. I count seven phase wraps centered at about 3500 Hz. That corresponds to a little over 2 feet of delay. And your group delay seems to extrapolate to about 10 milliseconds in the same region. What is your crossover frequency and order?

 

I'm using the zero-order or no name XO that Chris has described. Basically just a couple of PEQs to heavily attenuate the HF or LF as desired. Based on Chris' suggestion I found a spot where the HF and LF are close in phase. My attempt for this was to cross around 500Hz. If I do just a HF sweep only I still have those problems. I'm really new at this but that suggests to me that if no real XO involved and singe HF driver it can't be something form the LF drivers. I've tried the other HF driver with same result. Is it something in my room perhaps? Is it reflected energy back at the microphone? I'm not really sure if that is possible but seem to remember reading about that a few months back when I was trying to understand all of this.

 

Here is a screen shot of Xilica console so you can see what I am doing to roll off the HF/LF accordingly trying to build of my no-name crossover.

 

LF only:

 

Here is the HF added into the LF above:

 

 

[rplace]

Here are the HF sweeps only, no LF. It seems to be there with and without any processing applied

 

With all the PEQs, 1.5ms delay and -7db Gain:

 

 

 

And here is it with nothing added or subtracted just raw sweep 20-20K for the HF driver only:

 

 

 

 

[The Dude]

On 11/25/2019 at 9:21 AM, rplace said:

If your speakers are dialed in to your liking and swap out a piece of equipment (specifically amps

I would think that there would be some differences with the gain structure?  But as I am still learning about gain structure, this may not be an issue.