Home Theater Forum and Systems banner

1 - 16 of 16 Posts

·
Registered
Joined
·
39 Posts
Discussion Starter #1 (Edited)
I posted below over on AVS Forum in the MiniDSP section in regards to sub woofer delay settings.

Goal is clean accurate bass. I am wondering if achieving a super flat SPL could in some way require delaying the timing such that is screws up the transient response. Or does avoiding nulls actually improve transient response? I am new to REW and have not used the waterfall graphs as of yet.

Based on some encouragement and thinking about it I guess I should make the subwoofer response between the BP2002 integrated subs and their own bipolar MTM as smooth as possible. Tuning out any room influence that may have changed Definitive Tech's original design intent.

Secondly layer on the third subwoofer that I have in the system. Original post below.

**********************************************

Did my first dive into subwoofer room correction and nothing but good things to report. The setup guides for 2x4 provided by this site and MiniDSP.com along with REW have made a great improvement in my listening room.

I have a question about time alignment of the subs. I have old Definitive Technology BP2002's in front each with a powered 12. I have a third subwoofer that can be read more about here.

http://www.avsforum.com/forum/155-diy-speakers-subs/1438061-transmission-line-isobaric-subwoofer-home.html

I seem to notice that when I performed the time alignment step by measuring and adding an offset more and deeper nulls occurred in the REW sweeps.

What does this mean? Or rather how should I interpret this change? I rely on a setup disc with "pops" or "clicks" that helps to fine tune timing delays of surround channels vs fronts & centers. Is there a test tone or method to listen and tune low frequency delays? I have a disc from Sheffield labs that does 100Hz bursts but would like to here from the folks here about audible test methods that are used with consistent good results. I am going to drop a post over on HT Shack in the REW forum too I think.

Here some more details about the setup as it stands.

1 - I measure SPL at 6 postions along the main seating area. Lots of data from which to generate biquads so any huge nulls or peaks will be averaged out significantly (IMO).
2 - As the front Left/Right BP2002's have internal cross overs I have left MiniDSP settings as pass through.
3 - The BP2002's are physically closer to MLP so the sub output is delayed by 1.3ms. Also wondering is delaying these subs is okay as they are build in to the main L/R mid & high frequency drivers. I.e. MTM in bipolar setup.
4 - The third sub is a 30Hz transmission line. MiniDSP is used to invert the phase and filter. HPF 20Hz 48dB/oct BW & LPF 85Hz 48dB/oct BW
5 - The averaged results matched the REW predicted curve very well.
6 - No room treatments whatsoever.

Here is a capture is an in process comparison. All three subs active. GREEN is with no timing delay at all. RED has 1.3ms delay for BP2002's based on physical distance delta. I am wondering about the nulls at 60Hz, 70Hz and 85Hz. I did ABA testing so I don't believe it is measurement error. All distance measurements taken from center of sub woofer (side firing and down firing.) I had also thought about taking my measured and adding delay based on the distance to each subs port instead of speaker. But not time yet to fiddle.
 

Attachments

·
Premium Member
Joined
·
1,562 Posts
> The objective should be for smooth SW SPL response and smooth SW handoff to the mains.
> There is no requirement for time alignment for multiple SWs. The TL SW phase response will be at least a little different anyway from the other 2 SWs, so theoretically perfect time alignment at all frequencies is not possible anyway.
> They should be setup to be cooperating with each other however. None should be 180-270° out of phase with another. Maybe this is better expressed as, they all should be within 180° of each other. It's better to be within 90° of each other. The closer they are together, the more the SPL there will be, but it may not be the smoothest that it can be. It may be better that they are not timed aligned exactly to each other so that nulls are reduced and SPL is smoothed. That can create an easier condition for EQ. Having distance setting accurate within 2-3 feet of actual and being sure the polarities are the same for each SW should be good enough.
> After the SW are working together well and the XO and initial SW EQ is active, then the group of SWs can be time aligned with the mains. That will provide good SPL support throughout the XO range. Good timing to the mains impacts the transient response. It is not important exactly how the SWs are timed to each other so long as they work well together as a group as suggested.
------
The Green trace looks reasonable for SW cooperation and EQ possibilities so that is an acceptable setting to use. It's better than the red (timed) setup.
If only the 3 SWs were active and the XO was active then I would be concerned that there is way too much SW SPL above the XO. We don't want SWs leaking too much into the mains. I assume this was not the case so active the XO and add some EQ to smooth the green trace a little more if this has not yet been done. If EQ is already active then this is an okay setup as is. I have seen smoother responses, but there are many that are worse. You must judge if it is good enough.
[You could start over to completely understand how each SW works with the others, move the one SW to a new position, change the Mains or LP position, or change relative SW timing more to try to find a smoother starting point for EQ, but all this work may not impact the final sound quality significantly.]
 

·
Registered
Joined
·
39 Posts
Discussion Starter #3
> The objective should be for smooth SW SPL response and smooth SW handoff to the mains.
> There is no requirement for time alignment for multiple SWs. The TL SW phase response will be at least a little different anyway from the other 2 SWs, so theoretically perfect time alignment at all frequencies is not possible anyway.
> They should be setup to be cooperating with each other however. None should be 180-270° out of phase with another. Maybe this is better expressed as, they all should be within 180° of each other. It's better to be within 90° of each other. The closer they are together, the more the SPL there will be, but it may not be the smoothest that it can be. It may be better that they are not timed aligned exactly to each other so that nulls are reduced and SPL is smoothed. That can create an easier condition for EQ. Having distance setting accurate within 2-3 feet of actual and being sure the polarities are the same for each SW should be good enough.
> After the SW are working together well and the XO and initial SW EQ is active, then the group of SWs can be time aligned with the mains. That will provide good SPL support throughout the XO range. Good timing to the mains impacts the transient response. It is not important exactly how the SWs are timed to each other so long as they work well together as a group as suggested.
------
The Green trace looks reasonable for SW cooperation and EQ possibilities so that is an acceptable setting to use. It's better than the red (timed) setup.
If only the 3 SWs were active and the XO was active then I would be concerned that there is way too much SW SPL above the XO. We don't want SWs leaking too much into the mains. I assume this was not the case so active the XO and add some EQ to smooth the green trace a little more if this has not yet been done. If EQ is already active then this is an okay setup as is. I have seen smoother responses, but there are many that are worse. You must judge if it is good enough.
[You could start over to completely understand how each SW works with the others, move the one SW to a new position, change the Mains or LP position, or change relative SW timing more to try to find a smoother starting point for EQ, but all this work may not impact the final sound quality significantly.]
Thanks for the comments. A few months ago I did sweeps of the subs. No EQ at that time. Moved the front speakers to get decent response but also good front soundstage. More scientific than a basic sub crawl. Based on that I determined the best position of the third TL type. Kind of balanced most SPL vs smallest nulls when possible.

Will try your suggestions. I did use the red curve for 1st EQ but it was averaged with 5 other sweeps from different mic positions.
 

·
Registered
Joined
·
39 Posts
Discussion Starter #4 (Edited)
Just a quick update.

I did a bunch more sweeps comparing the timing delay between the two front subwoofers vs the rear.

Concluded that no advantage or improvement could be had by delaying the fronts as they are physically closer.

Judgement was based on SPL levels below 100Hz or so. My front subs have a fixed XO at 80Hz. I determined 85Hz to be the XO for the TL. MiniDSP HPF at 20Hz and LPF at 85Hz. Both are BW 48dB/oct.

2nd task was to experiment with adding delay to all subs vs. my mains.

I have data from multiple listening positions. I looked at SPL around the XO points and also checked out the Excess Group Delay (EGD) feature of REW. Some positions measured best with 0ms of delay whereas alternate positions measured best with 3 or 3.3ms delay. It seemed to alternate between adjacent positions. EGD sometimes agreed with the SPL conclusions but not always.

Having a bunch of data I made three MiniDSP configurations that I am in the process of evaluating.

1 - 0ms (best measured data at some listening position)
2 - 1.2ms (Not the best measured data but based on the physical distance only)
3 - 3.6ms (Best measured data at MLP)

Same PEQ from REW is used for each delay set point.

I have some bass quality test discs. It is difficult for me to hear any audible sound difference so far. I may have heard a slightly more thick and perceived louder bass for one of the settings? Guessing that might be an indicator of a less optimal timing but that is a stretch. It could all be just in my head.:coocoo:

Now my preference is the 3.6ms setting but more critical listening ahead.
 

·
Premium Member
Joined
·
1,562 Posts
It sounds like your are pretty well dialed-in for the TL-SW delay setting.

The range of 0 to 3.3 ms is only about 90° at 80Hz so the difference in SPL is not likely to be too significant. I suppose a room mode may make one look better than the others at particular mic position, but that is not as important as it may look on the chart. Chose the one that sounds best, or has the smoothest SPL on average, or the one measuring the best at the LP. It probably would not make a meaningful difference.

The EGD is not very reliable in helping to fine tune settings that are this close to each other so that would not let that influence the decision for me. It's more helpful to show a significant timing error as when the timing is off by something approaching a full wavelength. I would just chose using the SPL chart.

The delay timing to the mains will probably have much more impact to XO range SPL and sound quality than which of those 3 TL-SW delays is chosen. Even there, the same value for mains delay is likely to work reasonably well with all 3 of your choices for TL-SW delay.
 

·
Registered
Joined
·
21 Posts
When using delay control, you'll only ever get the best results through various listening positions, when all listening positions maintain the same distance ratio between listener and speaker.

Generally, you can make the assumption that other listeners do not have the same ability to discern sound as yourself, and especially in the bass region. This assumption affords some leeway when finding the best compromise.

I shared some thoughts on timing control between multiple subs, and subs/mains over here.
 

·
Registered
Joined
·
39 Posts
Discussion Starter #7
When using delay control, you'll only ever get the best results through various listening positions, when all listening positions maintain the same distance ratio between listener and speaker.

Generally, you can make the assumption that other listeners do not have the same ability to discern sound as yourself, and especially in the bass region. This assumption affords some leeway when finding the best compromise.

I shared some thoughts on timing control between multiple subs, and subs/mains over here.
I had read many comments including yours on the MiniDSP page. So thanks for the comments and link.

I am a bit lost on how to make a "sounds best" judgement. As I mentioned previously I have a surround test discs that makes a pop sound and demos how a slight delay can easily shift image off center. Tweaking delay between Left Main/Surround & Right Main/Surround was easy. But the LF is harder to tell.

But that's what weekends are for....
 

·
Registered
Joined
·
21 Posts
Maximum SPL is obtained when multiple sound sources reach a position at the same time. I start with the impulse response to find the two speakers, get the timing into the ballpark, then I'll typically run binary sweeps around the ballpark setting.

I look at the sweeps and decide which one produces the highest SPL at and around the crossover setting. Verify with the phase response to ensure I'm not 360 degrees out of phase (giving good SPL support but poor timing).

For timing sub's only, there's generally no crossover, so I look at frequency response of the entire sub band. For more then two sub's, I'll start by aligning the two furtherest sub's (A being the farthest of the two, and B being the other) from the listening position, then switch out B for the next farthest sub (C) and align C with A. Keep repeating until all sub's are aligned with A.

Sounds best is subjective. For me, if the kick drums sound snappier, I find I'm on the right track.
 

·
Registered
Joined
·
39 Posts
Discussion Starter #9 (Edited)
Here are some of the measured results.

I went right for a room curve. The Transmission Line is tuned to 30Hz so I am hesitant to try boosting anything below 30Hz. I suppose I could lower my sub gains and get a flat EQ from 20Hz up to 300Hz. But just don't feel like torturing myself for a result that isn't listenable in reality.

Black trace is average of three points. MLP plus about 1.5' to the left and right. Nothing funky going on between 80-90Hz where the XO takes place. These positions are nearest to the TL.
Red trace is average of six positions that span from one side of the room to the other. All about 1.5 to 2' spacing. Some weirdness around XO. I wonder if a fourth sub would smooth this?

Here is the biquad that came from REW as loaded into the MiniDSP for the 3.6ms delay. The MiniDSP and any EQ can only effect 100Hz and below based on the system setup.
 

Attachments

·
Registered
Joined
·
21 Posts
I went right for a room curve. The Transmission Line is tuned to 30Hz so I am hesitant to try boosting anything below 30Hz. I suppose I could lower my sub gains and get a flat EQ from 20Hz up to 300Hz. But just don't feel like torturing myself for a result that isn't listenable in reality.
You're 20dB up at 30hz, with the curve being pretty steep around 110hz, and back to +/- 0dB at 20hz. If that was my system, I would try starting the boost curve around 160hz, with less "step" in the boost. So a more gradual increase with decreasing frequency. I would probably only go 10-15dB up at low frequencies also.


Black trace is average of three points. MLP plus about 1.5' to the left and right. Nothing funky going on between 80-90Hz where the XO takes place.
Everything looks fine with the SPL graph. SPL tells you very little about what's going on in the time domain though. Care to share an mdat?
 

·
Banned
Joined
·
4,838 Posts
I had read many comments including yours on the MiniDSP page. So thanks for the comments and link.

I am a bit lost on how to make a "sounds best" judgement. As I mentioned previously I have a surround test discs that makes a pop sound and demos how a slight delay can easily shift image off center. Tweaking delay between Left Main/Surround & Right Main/Surround was easy. But the LF is harder to tell.

But that's what weekends are for....
Don't know if this is useful to you. It is a mono Pink Noise track (stereo format, same signal to both speakers, so it sounds mono), centered between L and R mains. it is full-range, can be used for centering L & R mains or subs if you wish. It can be downloaded from my Google Drive by anyone logged into a HTS account from the following link:

Link...

I am compiling a series of 2-channel and 5.1 test signals which I will make downloadable. Have to use Google Drive because of the file size/type limitations on the forum. But it is a HTS account, so it is legit and will persist beyond the limits of time, space, matter, energy..........:rolleyesno:
 

·
Registered
Joined
·
39 Posts
Discussion Starter #12 (Edited)
You're 20dB up at 30hz, with the curve being pretty steep around 110hz, and back to +/- 0dB at 20hz. If that was my system, I would try starting the boost curve around 160hz, with less "step" in the boost. So a more gradual increase with decreasing frequency. I would probably only go 10-15dB up at low frequencies also.




Everything looks fine with the SPL graph. SPL tells you very little about what's going on in the time domain though. Care to share an mdat?
I am limited to EQ from 100Hz and below as my miniDSP only processes the lows. Mid-Bass, Mid and Treble are all straight from an OPPO to Analog inputs of the AVR (1999 so no Auto EQ features). I think I get your point to not have so much at 30Hz as to leave the 20Hz response so much lower. The miniDSP is cutting so if I drop the overall gain 20Hz will go down also.

The bass test discs that I have sound natural but with more substantial deep bass due to the TL sub.

Here's one of my go-to discs. Oldie but a Goodie. :T

https://www.murfie.com/albums/various-artists-music-for-bottom-feeders-boston-acoustics-bass-disc-2

Also been listening to some reissues rock albums in surround. So far the bass is great and not overpowering. I should do a full sweep up to 20kHz and get a better idea of how loud the bass is reality to mid frequencies like 1kHz - 3kHz

Here is a mdat file from a single position with no delays at all. I am very curious to start using other features of REW like the water fall graphs etc.... but haven't found the time just yet. :huh:
 

Attachments

·
Registered
Joined
·
39 Posts
Discussion Starter #13 (Edited)
Don't know if this is useful to you. It is a mono Pink Noise track (stereo format, same signal to both speakers, so it sounds mono), centered between L and R mains. it is full-range, can be used for centering L & R mains or subs if you wish. It can be downloaded from my Google Drive by anyone logged into a HTS account from the following link:

Link...

I am compiling a series of 2-channel and 5.1 test signals which I will make downloadable. Have to use Google Drive because of the file size/type limitations on the forum. But it is a HTS account, so it is legit and will persist beyond the limits of time, space, matter, energy..........:rolleyesno:
Thanks for the offer. In addition to what REW can generate I have correlated and uncorrelated pink noise etc... from a Sheffield Labs test CD.

I probably have lost count of how many copies of pink noise that I have in various CD, DVD and BD formats. Having a good set of digital files may prove handy down the road.

When measuring pink noise from various disc sources I seem to have settled on +18dB as a SPL target for the sub channel compared to the rest of the channels played individually.

Here's the link to the test disc that has some great timing/delay tests for mains vs surrounds. Bass test tracks also. Might be a good reference for useful material.

http://www.walmart.com/ip/The-Ultimate-DVD-Surround-Sampler-5.1-Set-Up-Disc/15075610
 

·
Registered
Joined
·
21 Posts
I think I get your point to not have so much at 30Hz as to leave the 20Hz response so much lower.
Not really. More the fact that you have a large chunck of frequency response some 25+ dB greater then the frequencies above it. The curve is good, probably just a minimum of 5 dB to hot.

If only the 3 SWs were active and the XO was active then I would be concerned that there is way too much SW SPL above the XO. We don't want SWs leaking too much into the mains.
These things are subjective, so it's your call. jtalden makes a good point regarding SW leakage into the mains though.


Here is a mdat file from a single position with no delays at all. I am very curious to start using other features of REW like the water fall graphs etc.... but haven't found the time just yet.

Exercise caution with sine waves below 20hz. Damage is far to easy, and you have no useful data below 20hz anyway.

An ideal waterfall should look like the first image in this thread (not including any room curves).
When looking at the waterfall, we have SPL from top to bottom, frequency response from left to right, and the results of both SPL and frequency response through the time domain (near and far).

Each line in the waterfall represents the SPL and frequency response at some point in time beyond the initial signal hitting the microphone. It's like an echo.

At the back of the waterfall is the frequency response as shown on the SPL tab in REW, and as the waterfall moves closer to the viewer, we have the frequency response through time (the echo).


On first judgement of your impulse response, there are some timing related issues (as expected), but without further data it's impossible to make any recommendations. If you have loopback capability with REW, you could measure each speaker with the loopback active. If you don't have loopback capability, accurate data can still be obtained, just the procedure is more convoluted.
 

·
Registered
Joined
·
39 Posts
Discussion Starter #15
That's for the comments.

Started to read up on the REW101 about waterfall graphs and played with REW some.

Also got about 5 pages through the link on HTS that focuses on these graphs and pros/cons.

Will re-read the Floyd Toole book where he describes waterfall graphs. He is not a big fan but I wonder if his comments are more geared towards speakers design and evaluation and not room analysis?

Also I have lots of household items that rattle in sympathy to various bass frequencies. This appears to be showing up in the graphs despite the gated nature of the original sweeps. At least that is my concern.
 

·
Banned
Joined
·
4,838 Posts
Started to read up on the REW101 about waterfall graphs and played with REW some.

Also got about 5 pages through the link on HTS that focuses on these graphs and pros/cons.

Will re-read the Floyd Toole book where he describes waterfall graphs. He is not a big fan but I wonder if his comments are more geared towards speakers design and evaluation and not room analysis?
All due respect to the experts and their research and writings, there is a lot of good info in them. Perhaps the bigger question is how to use that info wisely.
 
1 - 16 of 16 Posts
Top