simple impulse wave question - Page 4 - Home Theater Forum and Systems - HomeTheaterShack.com

Old 09-06-12, 10:01 PM Thread Starter
Shackster

Join Date: Aug 2011
Posts: 81
Re: simple impulse wave question

Quote:
Barleywater wrote: View Post
Ouch!

28ms mid delay and 29ms tweeter delay = 1ms, or about 33cm. This is multiple wavelengths at 4kHz, and doesn't seem right.
I was tinking, these time delays includes room and air loss to get at the listening chair.

I'll think again for this tomorow
rewjack is offline

Old 09-07-12, 02:44 PM Thread Starter
Shackster

Join Date: Aug 2011
Posts: 81
Re: simple impulse wave question

It is true that 1.0ms is 4cycle at 4Khz = 34cm. But IR measurements, between left tweeter and mid is only 0,1ms that I will have to add to tweeters, = 3.4cm.
Of course I will correct this, or move my head, but still ?
The physical distance of tweeter and mid transductors is about 3cm. So the final 1.1ms correction represents what? Mids latency?
rewjack is offline
Old 09-10-12, 12:35 AM
Senior Shackster

Join Date: Dec 2011
Posts: 239
Re: simple impulse wave question

Hello Jacques,

I come back here from Flatfinger's thread.

Bob has plenty of background, but seems stuck with his FFT analyzer, and just a little short on information theory for sampled discrete time systems. REW uses swept sine technique, which turns out to be very powerful. Low starting frequency guarantees time interval long enough to capture low frequencies. Inclusion of sufficient time after sweep signal ends captures remaining reflected energy, and this all shows up in resultant impulse response. Other measurement method exists as well: Maximum Length Sequence techniques use signal that appears as white noise, but may be used with same convolution techniques as swept sine technique to get virtually identical system impulse response.

Basic dynamic speaker driver is expected to behave as rigid piston for wavelengths greater than the diameter of the speaker's cone. As this point is passed speaker break up modes become important. Full range speakers such as RTF64 attempt handling this in several basic ways; cone may be made with intentional flexibility that includes a controlled amount of damping. At low frequencies cone moves sufficiently as one solid piece. At higher frequencies central portion of cone moves more, and as sound energy propagates towards edge, damping occurs; energy is radiated as sound, some is absorbed by cone itself, and acoustical impedance causes sound energy reflection back toward center of cone. Break up behavior is often explored with simultaneous tones, allowing examination of intermodulation distortion.

Driver with break up modes makes phasing of many frequencies dependent on which modes are excited, which in turn depends on source material.

Sub to woofer and woofer to mid crossovers typically occur at frequencies below break up modes of drivers, and thus problems from this are less often encountered. Mid to tweeter is highly affected by break up modes. And even used alone as full range drivers, imaging suffers when left and right speaker drivers have different modes excited.

The full phase response combined with full frequency response describes complete information in impulse response, and visually are much easier to see. What appear as tiny changes in shape of impulse response often show as large changes in both phase and frequency responses. REW makes working with both time domain and frequency domain easy.

My 4kHz crossover demo is really nothing more than importing crossover filter. You should be able to do this with crossover filters created with JRiver software. File needs to be 32bit integer for REW. Filters are also fairly easy to create with Audacity. Audacity also makes linear phase/FIR filters, allows sample rate conversions and changing between integer/floating point/double precision formats. Trick with REW is that when importing impulse response, it always sets peaks at t=0, so beginning of filter impulse responses need to be zoomed on and realigned using whole sample adjustments. Preferences need sub sample timing adjustment turned off too.

Impulse timing adjustments much less than single sample may be used, and delays may be preserved when filters are exported. I never use delay settings in config file for convolution, always incorporate delays directly in filters.

What length crossover filters are you using?

Regards,

Andrew
Barleywater is offline
Old 09-10-12, 07:37 PM Thread Starter
Shackster

Join Date: Aug 2011
Posts: 81
Re: simple impulse wave question

Hi Andrew
Quote:
Barleywater wrote: View Post

Basic dynamic speaker driver is expected to behave as rigid piston for wavelengths greater than the diameter of the speaker's cone. As this point is passed speaker break up modes become important. Full range speakers such as RTF64 attempt handling this in several basic ways; cone may be made with intentional flexibility that includes a controlled amount of damping. At low frequencies cone moves sufficiently as one solid piece. At higher frequencies central portion of cone moves more, and as sound energy propagates towards edge, damping occurs; energy is radiated as sound, some is absorbed by cone itself, and acoustical impedance causes sound energy reflection back toward center of cone. Break up behavior is often explored with simultaneous tones, allowing examination of intermodulation distortion.
This may explain the high dirtortion I encounter often when I use 100hz and 120hz warble tones to test some of EQ's settings. Most of added gains provoques horrible distortion noises that I have to eliminate. This behavior sounds like the Break ups you have discribed.

Is this related to your explanation?

Quote:
Barleywater wrote: View Post

What length crossover filters are you using?
Well, will have to do some research on the JRiver's MC17 Low and high-passes I'm using to create crossovers. Settings are limited to frequency, and slope. I don't know to what I can refer for "lenght"?

There is no "Q" factor for High-low-pass filters.

Have to work more, again.

best regards
jacques
rewjack is offline
Old 09-12-12, 03:34 PM
Senior Shackster

Join Date: Dec 2011
Posts: 239
Re: simple impulse wave question

How do you create warble tones? And yes, depending on number of frequencies present, and their spacing, intermodulation distortion may be what you are hearing.

For crossover filters I was referring to length in samples.

Mixing filter slopes in crossovers to me is begging for alignment difficulties.

Do you have other software good for recording/editing. Have you used Audacity?

Andrew
Barleywater is offline
Old 09-12-12, 07:52 PM Thread Starter
Shackster

Join Date: Aug 2011
Posts: 81
Re: simple impulse wave question

Hi Andrew
Quote:
Barleywater wrote: View Post
How do you create warble tones?
I got a full tone CD on the web last year. I do not remember where but I just found the same exact file names on this one, must be it.
http://binkster.net/extras.shtml

Quote:
Barleywater wrote: View Post
For crossover filters I was referring to length in samples.

Mixing filter slopes in crossovers to me is begging for alignment difficulties.

Do you have other software good for recording/editing. Have you used Audacity?

Andrew
OK I usually take 512k lenght. I already have chosen the approximate location for crossovers and I focussed on specific frequency ranges; for sub-woofer, a 10-200 Hz sweep, 20-300 Hz for woofer-mids and 1-8Khz for tweeters. I tryed Long and short sweeps and did not makes any differences for the targeted location readings. I can not use these readings for wider examination.

I am awared of difficulties in mixing slopes and the only place I still have a 24db facing a 48db/oct is between the Sub Lpass ans woofer Hpass. It happened after several measurements where I have experienced unperfect phase and delay matching at XO point using equal slopes? The best one was strangelly this 48db/oct one the sub side and 24db/oct on the woofer side.

I know Andrew, I still have some more reading on the subject, some grey zones, like fast identification of coupling, isolation, combination and combing zones at and near crossover points.
Quote:
Barleywater wrote: View Post
Do you have other software good for recording/editing. Have you used Audacity?
I don't have Audacity, but I have the recording/editing Total Recorder software. Can I do something with that?

regards
jacques
rewjack is offline
Old 09-17-12, 09:13 AM
Senior Shackster

Join Date: Dec 2011
Posts: 239
Re: simple impulse wave question

I worked with H13 sub-woof results, forming A-B result. IR Window for results was set to optimize notch:

Indeed your mixed slope crossover appears to work quite nicely!

I sense however that more optimizing of mid-tweet may be possible.
Barleywater is offline
Old 09-17-12, 08:29 PM Thread Starter
Shackster

Join Date: Aug 2011
Posts: 81
Re: simple impulse wave question

Hello Andrew
Nice work on H13 data, thanks. This confirmed the rigth track I may be on.

I want to get back on Butterworth filters, which I have to work with using JRiver without convolution VST.

I have read the Joe Rasmussen" paper as flatfinger told me and there is a some conclusions that came from this reading. I am more awared of the digital processing power of my actual setup. This is so clear when I see the hard way to make adjustments to filters if one needs to do it almost peace by piece with electronic devices. Active digital proceccing is a world ahead analogue passive filtering.
It remenber me when I was doing soil geochemistry without computer in the early 90's, for a couple of months we were able to get a small set of maps covering a small amount of chemical elements and add a little processing the hard way. I am now able to create maps for all elements and try any stat treatments I want in a single day. Try this, try that, it's a game.
It's almost the same here. It is easy to try many possibilities of slope, frequency cuts and time delays having Jriver MC 17's DSP parametric EQ in operation while measuring REW sweeps. Measurements follows any modification right away. At the end, of coarse, I worked every nights since several weeks anyway to fix things, trying different "XO" position for accurate instrument sounds. I'm now faster on it and If I can't get a phase slope enough similar with different filters, I change a bit the frequency cut location (the higher angle phase slope = the higher phase delay) backward-foreward and find the optimal one, correcting time delay when needed depending on the variation size.

So as it is said, Joe prefers first orther Butterworth (12 db/oct)??. If it's right, my tests do not match this approach. I was not able to get satisfying results with single-order Butt. Have a look at fig.1 (H28). It is one example where 2th orther gives a lot better summation. Here I just compared 1th and 2th for woofer-Mid XO, but I did it for them all (1th to 4th). The red summation line is for 12db and the blue one for 24 db.

Every time I create a new crossover the higher orther filter is always the easyest to work with?
THe H32 files show good phase matching at 36 db/oct between woofer-mids "XO" at 132Hz. H33 are for left Tweeters-mids at 48 db/oct.

H30 is an test on phase effects of filter orthers 1th to 4th, for main driver at 3.3Khz. The H29 black line is the raw (no filter) crossing the "0" degree line at 3.3Khz, used for reference. It looks like they go around the circle, 73deg(1th), -176deg(2th), -102deg(3th) and -3.9deg(4th) which match the raw no filter.

How does it look from your side?

best regards
jacques
Attached Thumbnails

Attached Files H32_Mid_Woof.mdat (651.7 KB)  H33_Twt_MId_left.mdat (6.25 MB)
rewjack is offline
Old 09-17-12, 10:42 PM
Senior Shackster

Join Date: Dec 2011
Posts: 239
Re: simple impulse wave question

H33_Mid impulse response has bad ringing artifact 5ms before main peak, and appears to be ruining A+B v A-B notch. I try 2ms gate and can't get the math to behave, even though phase slope and cross points look good. Artifact is not present in earlier measurements.

Andrew
Barleywater is offline
Old 09-18-12, 03:35 PM Thread Starter
Shackster

Join Date: Aug 2011
Posts: 81
Re: simple impulse wave question

Andrew
Quote:
Barleywater wrote: View Post
H33_Mid impulse response has bad ringing artifact 5ms before main peak, and appears to be ruining A+B v A-B notch. I try 2ms gate and can't get the math to behave, even though phase slope and cross points look good. Artifact is not present in earlier measurements.

Andrew
Yes I can see it clearly at 15ms on the left mid driver, it's big. Well I've quited to early on this yesterday I guess. This is what happens when I try to go faster at the end of a work session.

I just took the same measurement, no change and this artefact is gone?

I saw your that you're using Blackman-Harris 4 for previous H13 dataset A-B IR Window. You say it's set to optimize notch. Can you developed on this?

thanks
jacques

PS, How come this file is that big, 2 mb? Is it the lenght of the sample?
Attached Files H33b_Mid_L48.mdat (2.08 MB)

Last edited by rewjack; 09-18-12 at 03:46 PM. Reason: add H33b file
rewjack is offline

 Bookmarks

 Tags impulse , question , simple , wave

Message:
Options

Register Now

Random Question
Random Question #2

User Name:
 PLEASE READ BELOW PRIOR TO ENTERING AN EMAIL ADDRESS! ATTENTION! YOU MUST ACTIVATE YOUR ACCOUNT!Activation requires you reply to an email we will send you after you register... if you do not reply to this email, you will not be able to view certain areas of the forum or certain images... nor will you be able download software. AN INVALID EMAIL ADDRESS WILL CAUSE YOUR ACCOUNT TO BE DELETED! See our banned email list here: Banned Email List We DO NOT respond to spamcop, boxtrapper and spamblocker emails... please add @hometheatershack DOT com to your whitelist prior to registering or you will get nowhere on your registration. Email Address:
OR

Log-in

Human Verification

In order to verify that you are a human and not a spam bot, please enter the answer into the following box below based on the instructions contained in the graphic.