Home Theater Forum and Systems banner

21 - 40 of 51 Posts

·
Registered
Joined
·
98 Posts
Discussion Starter #21
Posting some new measurements taken today. These are only the right side horn and midbass drivers.
No delays applied. Only crossovers have been set, and a tiny bit of EQ to take down some peaks.

I'll be interested to see if we can determine anything about the phase from these?

I was attempting for an acoustic crossover @ 1200Hz. Linkwitz 24 slopes.
 

Attachments

·
Premium Member
Joined
·
1,562 Posts
I reviewed the data.
As before, the IRs show very strong early reflections that make it impossible for me to adjust timing using the phase tracking method. If this were in a room situation I would question if both L+R Horns and L+R MR units were both operating during their measurements. The reflections are basically as strong or stronger than the direct sound.

I quickly checked your current timing for SPL support and found it be very good. There is support across the XO range and I did not improve it when I checked a few timing adjustments. Given this setup I would say the current timing is pretty well optimized. In looking at the IR locations the MB is leading the horn a little, but the distance is well within the range that is normal. I believed this alignment is as good as possible.
 

·
Registered
Joined
·
98 Posts
Discussion Starter #23
I reviewed the data.
As before, the IRs show very strong early reflections that make it impossible for me to adjust timing using the phase tracking method. If this were in a room situation I would question if both L+R Horns and L+R MR units were both operating during their measurements. The reflections are basically as strong or stronger than the direct sound.
Not surprising. The horns are mounted underneath the dash so there is almost do direct path to the LP. The sound is going to be reflecting quite a bit.

I quickly checked your current timing for SPL support and found it be very good. There is support across the XO range and I did not improve it when I checked a few timing adjustments. Given this setup I would say the current timing is pretty well optimized. In looking at the IR locations the MB is leading the horn a little, but the distance is well within the range that is normal. I believed this alignment is as good as possible.
This was actually my exact thought, too. When I compared the overlays of each driver independently with the combined response, I saw the SPL support was good. Keep in mind that my house curve has quite a tilt to it because in a car environment we skew heavy on the bass.

This is a higher crossover point than I have traditionally attempted to use. My horns can play down to about 800Hz if I use 24 dB slopes. This time I was attempting a 1200Hz acoustic crossover point. However, it would appear that my measured crossover point was more like 1300Hz? Do you agree?

I was initially surprised that the SPL support was so good without any delays on the drivers. But when I think back to some of the threads about Linkwitz 4th order crossovers being 360 degrees out of phase, then it makes sense that a midbass driver which is closer to the LP might align well without any delay. Especially when the distance between the horn and the midbass is about 1 wavelength @ the crossover range. I think 1300Hz is about 10 inches, correct? That's about how much closer the midbass driver is to the LP vs. the horn.

Is my understanding of this on target?
 

·
Registered
Joined
·
106 Posts
Have you ever considered performing the alignment using the phase trace?

Edit: oh , missed that part about early reflections making it impossible........
 

·
Registered
Joined
·
98 Posts
Discussion Starter #26
I believe in my situation, since I am using horn-loaded compression drivers that are mounted under the dash, there really isn't much direct sound. Everything is reflected in some way. So I don't know if even windowing could clear the phase traces up enough to be helpful?
 

·
Registered
Joined
·
106 Posts
No you're probably right,man that sounds very interesting. I'd love to get ahold of that project and throw my "rolly cart" on it..........I'm actually working on a really cool experiment right now, I'm at work, but is slow......
 

Attachments

·
Registered
Joined
·
98 Posts
Discussion Starter #28
What are people's opinions on whether or not I should be sitting in the driver seat during measurements?

These measurements thus far have been with me outside the car. I'm wondering if I should be in the car during measurements, however?
 

·
REW Author
Joined
·
6,711 Posts
Another plot which can be interesting when looking at time alignment is a wavelet spectrogram, I've been adding that to REW lately. Here is a normalised wavelet spectrogram of the combined right side data, with the peak energy arrival shown by the dashed line. Beta versions with the wavelet spectrogram are available for Windows and OS X.

wavelet.png
 

·
Registered
Joined
·
98 Posts
Discussion Starter #30
Another plot which can be interesting when looking at time alignment is a wavelet spectrogram, I've been adding that to REW lately. Here is a normalised wavelet spectrogram of the combined right side data, with the peak energy arrival shown by the dashed line. Beta versions with the wavelet spectrogram are available for Windows and OS X.

View attachment 96241
Very cool, although I have no idea what I'm looking at. LOL :D

Is the goal to have the dashed line be smooth, with few of those vertical "jumps?"
 

·
Registered
Joined
·
106 Posts
So, as far as sitting in the car or out, I have definitely tuned both ways, i honestly think that for the most part, it doesn't matter. The parameters that we adjust when optimizing a car audio system, its been my experience, that any differences that you'd end up with at the end of the process would be so small that you couldn't rule out normal variation/tolerance. Do ya know what i mean. in fact, now that i think of it, Id say in most cases out of the car would be recommended because sometimes when were in a car and trying to tune a system, we have a greater chance of accidentally (unknowingly?) affecting our measurements, due to our physical body positioning (legs, arms blocking/unblocking sound propagation) and us moving about. Id say, try both and see what differences you come up with.
 

·
Registered
Joined
·
106 Posts
I've been doing experiments with measuring the step response of a system before and after alignment. Very interesting, this could be another way to verify the coherence of the different drive units thru crossover. I wish I had more time to post the pics today, but we're closing in 30 min.
 

·
Registered
Joined
·
106 Posts
Another plot which can be interesting when looking at time alignment is a wavelet spectrogram, I've been adding that to REW lately. Here is a normalised wavelet spectrogram of the combined right side data, with the peak energy arrival shown by the dashed line. Beta versions with the wavelet spectrogram are available for Windows and OS X.
John, this wavelet analysis look really interesting. Can't wait to experiment with that one, for sure!! On a different note, I've been experimenting with a way to be able to reform a transfer function measurement on a system that has no means of getting a signal into it, like a lot of OEM Car systems, when all they have is CD or USB. What I did was come up with a 131k periodic pink noise signal with a synchronization pulse every 3 seconds, this way I can line up the measure and reference in SysTune and once I get them aligned, turn on auto delay tracking to account for the drift (because the two clock sources aren't syncronous) and WHAMO, it worked!!........kinda, I got good data up till 4500Hz. Here is a pic of the signal I am using.

Any chance you might incorporate some type of triggered sweep, where we can play the excitation signal thru a cd or USB stick?
 

Attachments

·
Premium Member
Joined
·
1,562 Posts
This time I was attempting a 1200Hz acoustic crossover point. However, it would appear that my measured crossover point was more like 1300Hz? Do you agree?
The crossing point changes a little depending on smoothing used. I just looked at 1/3 smoothing and that way it is near 1400. Minor differences like this between the electrical and acoustic XO frequencies is common. There is nothing to be concerned with.

I was initially surprised that the SPL support was so good without any delays on the drivers. But when I think back to some of the threads about Linkwitz 4th order crossovers being 360 degrees out of phase, then it makes sense that a midbass driver which is closer to the LP might align well without any delay. Especially when the distance between the horn and the midbass is about 1 wavelength @ the crossover range. I think 1300Hz is about 10 inches, correct? That's about how much closer the midbass driver is to the LP vs. the horn.

Is my understanding of this on target?
Yep, 1300Hz is about 10.4" WL.

Your MB is leading the Horn a little and as such the direct sound phase tracking is probably not ideal. It is likely, as you suggest, that the phase is crossing near the 1300Hz XO with the MB leading by maybe 1 WL. This is not a bad situation. It is in fact a very good setup. This can provide a reasonably flat phase and GD response with the only the XO range phase being irregular. The more important SPL support is still strong.

In a room with vertically offset drivers the frontal lobe would shift up and down a little in the XO range, but even there, that is not a major influence in most cases. In your car setup I would think this is no issue at all.
 

·
REW Author
Joined
·
6,711 Posts
Very cool, although I have no idea what I'm looking at. LOL :D

Is the goal to have the dashed line be smooth, with few of those vertical "jumps?"
Pretty much. A conventional spectrogram is produced using a fixed window width, which gives it the same time resolution at all frequencies. The frequency resolution is similarly a fixed number of Hz, 10 Hz if a 100 ms window was used, for example. At low frequencies that is a big octave fraction (1/1.4 octaves at 20 Hz), at high frequencies a very, very small octave fraction (1/1386 octaves at 20 kHz). For a time-frequency plot it would be handier if the tradeoff between time and frequency resolution varied with frequency, with higher time resolution at high frequencies and lower at low frequencies. A wavelet transform can achieve that, specifically a constant Q wavelet transform. A constant Q wavelet transform is mathematically equivalent to using a frequency-dependent window to produce the spectrogram (which is what REW does). For reference, this is how a perfect impulse looks with the same settings.

impulse.png
 

·
REW Author
Joined
·
6,711 Posts
What I did was come up with a 131k periodic pink noise signal with a synchronization pulse every 3 seconds, this way I can line up the measure and reference in SysTune and once I get them aligned, turn on auto delay tracking to account for the drift (because the two clock sources aren't syncronous) and WHAMO, it worked!!........kinda, I got good data up till 4500Hz. Here is a pic of the signal I am using.

Any chance you might incorporate some type of triggered sweep, where we can play the excitation signal thru a cd or USB stick?
I'm not sure why you need the sync pulse, the PN is periodic anyway so I'd have thought SysTune could align to the reference without anything else.

Generating responses with offline signals is on the REW todo list.
 

·
Registered
Joined
·
106 Posts
You're right John, I was over complicating it. What had me confused was the fact that in SysTune, the custom file option for excitation signals only allows just over 500,000 samples, so at 44.1k I think is right around 11 seconds. So any custom signal used, if longer than that, is chopped to 11 seconds. It never occurred to me to just record a long time file (30+min.)of the built in generator of SysTune, then burn that to disc, then use the built in generator of SysTune as my reference, and play the disc. So that's what I did today, and it worked like a champ!
 

·
Registered
Joined
·
106 Posts
Got another question for ya John, I can't figure out why, when I select Periodic Pink Noise in the generator in REW, and select the file button, to save the signal to wav file, then import the signal into SysTune, it looks like this: And I've tried reducing the level of the signal in REW before saving, as low as -22dB, doesn't seem to matter? The signal looks clipped, or am I missing something? The second pic is the built in generator.
 

Attachments

·
REW Author
Joined
·
6,711 Posts
The PN sequences REW generates are optimised to have a crest factor (ratio of peak level to rms level) that does not exceed 6 dB.
 

·
Registered
Joined
·
106 Posts
The PN sequences REW generates are optimised to have a crest factor (ratio of peak level to rms level) that does not exceed 6 dB.
Ok, awesome! See that's very handy to know, I never would have known that had I not imported the signal to SysTune. Even when I try to view it on my scope, it isn't as apparent as in SysTune. So it's not clipping at all, which is what I found when I zoomed in real close, hence the confusion. Thanks John!

One more thing, can you elaborate a little on why you made the signal this way. Potential benefits and downfalls?
 
21 - 40 of 51 Posts
Top