Home Theater Forum and Systems banner

1 - 20 of 51 Posts

·
Registered
Joined
·
98 Posts
Discussion Starter #1
Starting a new thread for this issue.

I'm trying to use REW's impulse response measurements for determining the time alignment settings for my car audio system.

Here are the testing parameters:

Using a Behringer ECM8000 and M-Audio M-Track sound card. REW used on a MacBook Pro.

Mic was placed in the driver seat, on a mic stand with the mic held at the center of head position.

My system is comprised of the following speakers:

2 x Horn Loaded Compression Drivers (installed under the dash)
2 x 8" Midbass drivers (installed in the front doors, stock location)
2 x 12" subwoofers (installed in a trunk baffle)

All speakers are active crossover via an Audison BitOne DSP.


For these measurements, only the horns and midbass were tested. Subwoofers are not included in these IR measurements.

These measurements are taken with no delays applied to any of my drivers. I'm hoping to determine how much delay to apply to each one.

The measurements are labelled as follows:

L HLCD = Left Horn
R HLCD = Right Horn
L MB = Left Midbass
R MB = Right Midbass

Each driver was measured independently with all others muted.

Attached is my mdat file with the measurements.


I'm hoping to get help understanding how to align these measurements to determine the correct time alignment for these drivers. As I understand it, the impulse peaks are not the correct point to align the IR. The "initial rise" is the correct point to aim for, but I don't quite understand where the "initial rise" is located.


Thanks in advance for any assistance.
 

Attachments

·
Premium Member
Joined
·
1,562 Posts
The IRs are way too irregular for me to do any detailed timing analysis in the manner I normally do.

> I notice that you ran 2 sweeps for each of these measurements. I don't think that should be an issue, but I would recommend single sweeps. If the relative timing is changed even slightly between the 2 sweeps it will cause measurement problems.

> I see the sample rate is reported in REW info as 12,000? I don't know if that is a problem, or not, but it doesn't seem right. I usually see 41.1k or 48k or 96k. I didn't know REW would use other rates. If there is resampling happening in the measurement (different rates between REW, the audio interface, and XO unit) maybe that is causing a measurement problem.

> When all is normal we should be able to repeat the exact results with repeated measurements of a driver. If we can't do that there is a measurement problem.

> It is easier for me if full range sweeps are used as I can better determine the direct sound phase path. The narrow range sweeps should be okay also, but that is not my normal process so I have little experience that way.

> With the mic close to a driver or at the mouth of the horn a very clean IR should result. If it is clean and repeatable that way and irregular at the LP that suggests the irregular IR is just the result of the car reflections.

I suggest you experiment with this an see if the IRs can be cleaned up. If so, I may be able to help with timing. If not, the timing you have is not too bad. There is another process we could try. I have only tried it once or twice, but it may work better under these difficult circumstances.
 

·
REW Author
Joined
·
6,711 Posts
> I see the sample rate is reported in REW info as 12,000? I don't know if that is a problem, or not, but it doesn't seem right.
If the 'Decimate IR' option is selected in Analysis preferences REW will reduce the sample rate according to the sweep end frequency to reduce the storage required.
 

·
Registered
Joined
·
98 Posts
Discussion Starter #4
My soundcard is 48kHz and that's the sample rate I used in the prefs.

However, you are correct that I had the decimate IR box checked.


Thanks for the suggestions. I'm going to attempt new measurements and see if I can get better results.
Unfortunately, I am traveling right now so it will be a few days before I can do this. I will report back ASAP.
 

·
Premium Member
Joined
·
1,562 Posts
There is no problem in continuing to use the Decimate IR function. The advantage is a smaller file size. The tradeoff is of no consequence to the particular job we are doing. Either way is fine for our purposes.
 

·
Registered
Joined
·
98 Posts
Discussion Starter #6
Following up on this thread:

I am home from my trip and finally had enough free time to attempt new measurements.

I have changed the setup conditions since my previous attempt. A review of the conditions is below:


Computer = MacBook Pro
Sound Card = Behringer FCA610
Mic = Behringer ECM8000

Mic at listening position, driver's seat, ear level, facing forward in horizontal orientation.

Speakers tested are as follows:

L HLCD = Left Horn
R HLCD = Right Horn
L MB = Left Midbass
R MB = Right Midbass

This time I ran sweeps from 20Hz - 20kHz.
Only 1 sweep instead of 2 as before.

I took measurements with the car windows open, and another set with the windows closed. (attempting to change reflections)


Please review the attached .mdat file and help me determine if these measurements look more valid, and more importantly, the correct impulse alignment if I am trying to change my time alignment delays.

Lastly, I have included the measurement for my sound card calibration. I'm hoping someone can tell me if the sound card is inverting the inputs or not?


Thanks in advance for any assistance!
 

Attachments

·
Registered
Joined
·
98 Posts
Discussion Starter #7
I took a swing at this myself, but I'm just not sure if this is aligned correctly or not.
Only used the measurements taken with the car windows closed.

Here is what my impulse overlay looks like after I shifted the impulses by the amount of delay I think is required. Does this look correct now?

Delay times applied to the impulses were:

L HLCD = .5491 ms
R HLCD = 0 ms
L Midbass = 1.4072 ms
R Midbass = .6326 ms
 

Attachments

·
Premium Member
Joined
·
1,562 Posts
The cal file looks normal. The SC (soundcard) does not invert the phase; the SC phase follows along the 0° line of the SC phase chart. To see the efficacy of the SC cal file we could just do a loopback measurement after the SC cal file is created and active. A good cal file will measure the loopback cable at flat SPL over maybe 10-20k Hz and the phase will track at 0° over that same range. The phase may roll up or down a little at the top end as that is dependent on the exact positioning of the IR. In that we case could always manually adjust the IR position a little to confirm the phase is indeed flat. REW auto IR positioning is very good, but always perfect in this regard.

This is just background FYI. Your SC cal is good - no action needed.

Measurements:
Great start! :sn:
The IRs in your chart now look very normal. The IR positioning in the chart looks exactly how I would align them. The next step is to see which relative polarity of the MR to TW works best. You can swap the polarity on one set of the drivers (I would chose the TWs in this case). The relative polarity to that provides the most SPL support through the XO range is the correct one to use. A final step would then be to make small changes to the delay to assure the XO range SPL support is as smooth as possible. Changing the delay by about 20-30° relative to the XO freq is a good choice. This last step in usually not critical to the sound quality, it is more for those of us that like to fine tune the system for best measurements.

I will next take a look a the file and see what additional adjustments I would have suggested.
 

·
Premium Member
Joined
·
1,562 Posts
I reviewed the data file. While I was optimistic of being able to provide optimized timing advice for your situation, I was not able to do that. There is still a lot of TW phase congestion in the XO range making is impossible for me to do any better than your chart timing above.

I am guessing that this may be because the horns may not have a direct sound path to the mic. If they are under the dash or pointed away from the mic then all sound arriving from them is reflected to some extent. There is still first arriving sound from the closest path, but the other reflections are about as strong and very closely timed, hence the chaos.

If this is the case, Then your chart above is probably the best that can be done. Minor deviations in timing from that setting is not likely to have much impact at all on the SPL support through the XO range like it would when there are fewer close reflections. Even a TW polarity change is not likely very significant although that may still be worth a try.

Possibly there is better advice from someone experienced in car audio.
 

·
Registered
Joined
·
98 Posts
Discussion Starter #10
Measurements:
Great start! :sn:
The IRs in your chart now look very normal. The IR positioning in the chart looks exactly how I would align them. The next step is to see which relative polarity of the MR to TW works best. You can swap the polarity on one set of the drivers (I would chose the TWs in this case). The relative polarity to that provides the most SPL support through the XO range is the correct one to use. A final step would then be to make small changes to the delay to assure the XO range SPL support is as smooth as possible. Changing the delay by about 20-30° relative to the XO freq is a good choice. This last step in usually not critical to the sound quality, it is more for those of us that like to fine tune the system for best measurements.

I will next take a look a the file and see what additional adjustments I would have suggested.

Awesome! Thanks for the help. I definitely feel like I understand this a lot better now.

My crossover point for the midbass and horns is 800Hz @ 24dB Linkwitz. I was looking at the phase chart at 800Hz and saw that the Left Horn and Right Midbass drivers aligned very well.
The Left Midbass looks like it would align better if inverted, however. I have an invert polarity button in my DSP for each channel, should I use that and then fine tune the delay to get the phase dialed in closer?
The Right Horn is about 80 degrees off from the Left Horn. Inverting it looks like it goes more out-of-phase.
Should I use delay to bring that driver's phase more in line with the others?
That horn is the furthest driver from the LP, so obviously I would just increase delay for all the other drivers by the same amount. I attempted this shift in REW but it took about .28ms of delay, which seems like a lot.

Thoughts?
 

·
Premium Member
Joined
·
1,562 Posts
Both channels drivers need to be the same. The 2 MR the same and the 2 TWs the same polarity. The relative polarity can be reversed for MRs Vs the TWs though. Just select the setting that gives the best SPL response. There may not be much advantage either way in this case however - just guessing.

I don't think you will learn much from the phase chart due to the various reflections. Moving the window range slightly make the phase tracking look very different. There is no clear way that I found to suggesting one setting is correct or even clearly better than the other. I would rely in the SPL chart to decide.
 

·
Registered
Joined
·
98 Posts
Discussion Starter #12
Just spent an hour listening to the system after applying the delays from my latest measurements. I can happily report that we have achieved the best imaging my system has presented yet. Many thanks!

I did not have to invert any of the drivers' polarity. I tried inverting the left midbass, but that killed the center image.

I did invert the subwoofer by 180 degrees, and then delayed it by about 3 ms. The bass is now well anchored in the front of the soundstage.


Here are some new measurements after the delays were applied. The impulses line up almost exactly as predicted before.
 

Attachments

·
Premium Member
Joined
·
567 Posts
If that was mine I'd swap the phase of the 2 MB elements or the others (either would work) to get everyone moving the same way (Positive or Negative) out of the gate. May not be noticeable due to the difference in the impulse waveforms but I'd be inclined to try anyway.

Just my $.02

GCG
 

·
Registered
Joined
·
98 Posts
Discussion Starter #15
If that was mine I'd swap the phase of the 2 MB elements or the others (either would work) to get everyone moving the same way (Positive or Negative) out of the gate. May not be noticeable due to the difference in the impulse waveforms but I'd be inclined to try anyway.

Just my $.02

GCG
OK, I'll give that a try after the weekend. :T
 

·
Registered
Joined
·
98 Posts
Discussion Starter #16
If that was mine I'd swap the phase of the 2 MB elements or the others (either would work) to get everyone moving the same way (Positive or Negative) out of the gate. May not be noticeable due to the difference in the impulse waveforms but I'd be inclined to try anyway.

Just my $.02

GCG
Thanks for the tip. You were correct. I measured my response, and then inverted the phase on the horns and measured again. I got better SPL support through the crossover range with the inverted horns. :T
 

·
Registered
Joined
·
98 Posts
Discussion Starter #17
I have another question about this topic...


When recording a measurement for the impulse response in order to determine time alignment, should we use a full range sweep? Or is it ok to limit the sweep range to an area closer to crossover point?

For example, if my midbass and horns are crossed over at 800 Hz, is it ok to sweep from 400 Hz - 1600 Hz?
Or is 20 Hz - 20,000 Hz still preferred?
 

·
Registered
Joined
·
98 Posts
Discussion Starter #19
Should IR measurements for time alignment be done before or after EQ and X-over are set?

I was trying another go at this today but I couldn't get very good IR measurements. I had done my EQ and X-over settings first this time. Previously, the measurements for time alignment were done first before EQ and X-over.
 

·
Premium Member
Joined
·
1,562 Posts
The XO changes the delay. EQ around the XO range impacts the delay a little also.

I'd suggest; set the desired XO filters, adjust the driver levels, adjust the delay, and then EQ.
It's a good Idea to confirm the delay is still good after EQ.

For additional delay adjustment after the initial setup just leave the EQ on.

That's my approach. It's the end result that counts so order is not required.
 
1 - 20 of 51 Posts
Top