Home Theater Forum and Systems banner

1 - 20 of 30 Posts

·
Registered
Joined
·
25 Posts
Discussion Starter #1 (Edited)
hello
I made some measurements with 1 sub in a corner, and I got some ups and downs in the frequency graph. When I move the sub to the middle of the wall, it is better, but still a dip between 50Hz and 60Hz.
In order to improve that, some people had suggested adding another sub. But if I buy the same type of sub, it would cost me 600$. That would be expensive if there is no improvement. So I wondered if I could predict the performance from a sub in the corner and a sub in the middle combined, from the measurements of a sub in the corner and middle separately.
I tried it out with the calculation (A+B)/2 in the allSPL control. First I unsmoothed the measurements, calculate, then smooth all with 1/6. Then I created the waterfalls with time range 600ms and window 500ms.
The waterfall of (A+B)2 however shows a more rapid decay which looks unrealistic to me.
Please note: measurements were done with the same settings, like same sample rate etc. I use an UMIK-1 USB microphone.

I do not want to start a discussion what is best, multisub, room treatment or PEQ.

My question is:
- Can REW predict the 2sub performance from single sub measurements?
- If yes, did I do it correct? Or what settings or procedure in REW should I use?
 

Attachments

·
Registered
Joined
·
1,298 Posts
To be honest i am not sure about the "predict 2sub performance".

Second sub can certainly help. Can you try the second sub, most suppliers let you take one home to try for a certain time?

What is the Sub and does it i have phase control?

Have you tried shifting the listening position, might be sitting in a null? Try other positions for sub location, listening position.

BTW setting the graph scale to 45 - 105 for posting is good.

What is your crossover?
 

·
Premium Member
Joined
·
6,711 Posts
Before you could combine measurements in that way you would need to make sure they were properly time aligned. Time alignment needs a timing reference for REW, which is provided using a loopback connection - but that is not possible with a USB mic.

An alternative, if your room is rectangular, is to look at the predicted results of a simulator such as the one at http://www.hunecke.de/en/calculators/loudspeakers.html
 

·
Registered
Joined
·
261 Posts
Before you could combine measurements in that way you would need to make sure they were properly time aligned. Time alignment needs a timing reference for REW, which is provided using a loopback connection - but that is not possible with a USB mic.
Is that process correct if you have it aligned? How accurate will that prediction be?
 

·
Registered
Joined
·
25 Posts
Discussion Starter #5
Before you could combine measurements in that way you would need to make sure they were properly time aligned. Time alignment needs a timing reference for REW, which is provided using a loopback connection - but that is not possible with a USB mic.

An alternative, if your room is rectangular, is to look at the predicted results of a simulator such as the one at http://www.hunecke.de/en/calculators/loudspeakers.html
hello John,
Thanks for your answer. I am just a beginner, but I would like to learn more.
Can you explain a bit more please what time alignment is? Why is it needed for the prediction? Why does a lopback provide it. Why is it not possible with USB?
That is a lot of questions. Is there any reference that I can look at to learn more?
 

·
Premium Member
Joined
·
1,562 Posts
Is that process correct if you have it aligned? How accurate will that prediction be?
Not JohnM but:

Yes, assuming the SWs would be identical, this is a very good way to predict impact of multiple SWs if you currently have only one. [Note that, the correct math is the sum (A + B) rather than the average.] REW loopback feature must be used to provide the proper alignment of the IRs for the calculation to agree with the actual measurement of 2 SWs. [If a loopback can't be done because of a USB mic or other issue, then a distance measurement between the two SW locations can be converted to a ms offset in the IRs, but this is a little more problematic.]

Below is an example of from my AV setup. I have 2 Identical SWs that are placed; front left at 22 ft from the LP (SW-L) and Rear Right at 11 ft from the LP (SW-R). Neither are in corners. The positions were selected by looking at the individual SPL responses to see that they compliment each other, when one is peaked the other is dipped resulting a sum that is smoother than either alone.

I have the ability to delay the closer "SW-R" so the two IRs are aligned closely as shown below. If you don't have this capability I recommend the locations be kept approximately equidistance from the LP. A difference of 2 ft is probably okay. The measurements/calculation will tell.

ir.jpg

Below is a measurement of the individual and combined SPL of the two "SWs".

spl a.jpg


Note that, the "SWs" SPL is as smooth as shown only because all these measurements include the EQ that I applied (same EQ to SW-L and SW-R). In your process the measurements should be without any EQ applied.

Below is the "SWs" (measured) Vs the "A + B" (Calculated) SPL. All the other charts hold up equally as well.

spl b.jpg


Remember that in order to get a high accuracy calculation it's necessary to have identical SWs and the appropriate IR offsets. The IRs above were only aligned that closely as that is the way that they are adjusted using a delay feature in my DCX2496.
 

·
Registered
Joined
·
25 Posts
Discussion Starter #7
Not JohnM but:

Yes, assuming the SWs would be identical, this is a very good way to predict impact of multiple SWs if you currently have only one. [Note that, the correct math is the sum (A + B) rather than the average.] REW loopback feature must be used to provide the proper alignment of the IRs for the calculation to agree with the actual measurement of 2 SWs. [If a loopback can't be done because of a USB mic or other issue, then a distance measurement between the two SW locations can be converted to a ms offset in the IRs, but this is a little more problematic.]

Below is an example of from my AV setup. I have 2 Identical SWs that are placed; front left at 22 ft from the LP (SW-L) and Rear Right at 11 ft from the LP (SW-R). Neither are in corners. The positions were selected by looking at the individual SPL responses to see that they compliment each other, when one is peaked the other is dipped resulting a sum that is smoother than either alone.

I have the ability to delay the closer "SW-R" so the two IRs are aligned closely as shown below. If you don't have this capability I recommend the locations be kept approximately equidistance from the LP. A difference of 2 ft is probably okay. The measurements/calculation will tell.

Below is a measurement of the individual and combined SPL of the two "SWs".

Note that, the "SWs" SPL is as smooth as shown only because all these measurements include the EQ that I applied (same EQ to SW-L and SW-R). In your process the measurements should be without any EQ applied.

Below is the "SWs" (measured) Vs the "A + B" (Calculated) SPL. All the other charts hold up equally as well.

Remember that in order to get a high accuracy calculation it's necessary to have identical SWs and the appropriate IR offsets. The IRs above were only aligned that closely as that is the way that they are adjusted using a delay feature in my DCX2496.
hello jtalden,
"unfortunately" I have a USB mic. But "fortunately" the subs I intend to combine will be the same type, and aequidistant to the listening position. Listening position is along the centerline between the subs. Under these provisions, is it possible to calculate A+B from USB measurements and get meaningful results?
 

·
Premium Member
Joined
·
1,400 Posts
John,

Thanks for the great post ( should be a sticky ), plus the detailed explanation ( including ; How to use 2 subs for smoother response ) . :T

:sn:
 

·
Premium Member
Joined
·
1,562 Posts
hello jtalden,
"unfortunately" I have a USB mic. But "fortunately" the subs I intend to combine will be the same type, and aequidistant to the listening position. Listening position is along the centerline between the subs. Under these provisions, is it possible to calculate A+B from USB measurements and get meaningful results?
Yes, this likely to work, but there is risk. REW automatically offsets the IRs to place them near 0 ms. The routine works very well, but the accuracy between 2 different measurements relies on the shape and uniformity of the 2 IRs. The different room location will change the IR shape so it will not be identical for the 2measurements. Since you know they are equidistant, you can inspect the IR alignment to assure it appears to be correctly set.

I can also offer an opinion if you post a screen shot of the 2 IRs or attach the .mdat.

It looks like REW did it correctly above and with those particular SW locations it does not look like there will be much SPL smoothing.
 

·
Registered
Joined
·
25 Posts
Discussion Starter #11
Yes, this likely to work, but there is risk. REW automatically offsets the IRs to place them near 0 ms. The routine works very well, but the accuracy between 2 different measurements relies on the shape and uniformity of the 2 IRs. The different room location will change the IR shape so it will not be identical for the 2measurements. Since you know they are equidistant, you can inspect the IR alignment to assure it appears to be correctly set.

I can also offer an opinion if you post a screen shot of the 2 IRs or attach the .mdat.

It looks like REW did it correctly above and with those particular SW locations it does not look like there will be much SPL smoothing.
hello jtalden,
Thanks for the help. Here is the mdat file. I would appreciate what you think about the prediction for the combined sub performance.
As a second sub would cost me 600$, would PEQ 125$ be better bang for the buck?
 

Attachments

·
Premium Member
Joined
·
1,562 Posts
For this purpose we need measurements of just the SW in the 2 locations. Your current data has SW+FL. For this purpose that does not work. The high frequencies dominate the IR and totally obscure the shape and location of the SW portion of the IRs.

Also the Interaction of the SW and FL may be contributing to the rough SPL response if they are not well timed.

To comment on just the 2nd SW I just need to see an .mdat containing:
> SW Corner
> SW Center

I can comment on both the 2nd SW and also the current timing of the SW+FL, but only with an .mdat containing:
> SW Corner
> SW Center
> FL
> SW+FL [you already provided this one, but if the mic position at the LP is going to change a little then it would be better to measure it again. I do not need the SW+FL at both SW positions, but it wouldn't hurt.]
 

·
Registered
Joined
·
25 Posts
Discussion Starter #13
For this purpose we need measurements of just the SW in the 2 locations. Your current data has SW+FL. For this purpose that does not work. The high frequencies dominate the IR and totally obscure the shape and location of the SW portion of the IRs.

Also the Interaction of the SW and FL may be contributing to the rough SPL response if they are not well timed.

To comment on just the 2nd SW I just need to see an .mdat containing:
> SW Corner
> SW Center

I can comment on both the 2nd SW and also the current timing of the SW+FL, but only with an .mdat containing:
> SW Corner
> SW Center
> FL
> SW+FL [you already provided this one, but if the mic position at the LP is going to change a little then it would be better to measure it again. I do not need the SW+FL at both SW positions, but it wouldn't hurt.]
hello jtalden
Thanks for your help. I probably have to do some measurements again. I have two questions:
- Normally I do 15Hz - 20kZ sweeps. Is that Ok for subwoofer? If not, what sweep range settings should I choose?
- I use a laptop with HDMI connected to an AVR. If I select the subwoofer(LFE) channel, the AVR adds 10dB so that the level is higher than the redirected bass when selecting a front channel. I had done SW measurements where I lowered the AVR by 5dB in order to prevent it getting to loud, but then it cannot be compared directly with redirected bass. When doing SW(LFE) measurement, should I lower the AVR by 10dB?
 

·
Premium Member
Joined
·
1,562 Posts
...
Normally I do 15Hz - 20kZ sweeps. Is that Ok for subwoofer? If not, what sweep range settings should I choose?
There is no problem using 15-20k sweeps. That is what I normally do. This can be advantageous if loopback measurements are being made. In this case there is no advantage one way or the other.

- I use a laptop with HDMI connected to an AVR. If I select the subwoofer(LFE) channel, the AVR adds 10dB so that the level is higher than the redirected bass when selecting a front channel. I had done SW measurements where I lowered the AVR by 5dB in order to prevent it getting to loud, but then it cannot be compared directly with redirected bass. When doing SW(LFE) measurement, should I lower the AVR by 10dB?
The LFE level doesn't matter for just the question of a 2nd SW. Just pick a reasonable level for you.

LFE will not work though for the question of the timing between the SW and FL. The reason is that we want to see the redirected bass. That is the bass that is impacted by the XO setting in the AVR or HTPC. If you want to provide that, then it is usually easiest to just send the signal to the FL and disconnect the FL speaker. That way only the FL redirected bass will be measured and the level will be correct. This method works for both questions.

I'm assuming you are redirecting bass (your speakers are set to small)? What is the AVR and its current XO setting? Did you use an auto setup feature in the AVR or set the distances manually? What are the main speakers and SW? I am looking for a little context here so I can better understand any future data you provide regarding SW to FL timing.
 

·
Registered
Joined
·
25 Posts
Discussion Starter #15
There is no problem using 15-20k sweeps. That is what I normally do. This can be advantageous if loopback measurements are being made. In this case there is no advantage one way or the other.



The LFE level doesn't matter for just the question of a 2nd SW. Just pick a reasonable level for you.

LFE will not work though for the question of the timing between the SW and FL. The reason is that we want to see the redirected bass. That is the bass that is impacted by the XO setting in the AVR or HTPC. If you want to provide that, then it is usually easiest to just send the signal to the FL and disconnect the FL speaker. That way only the FL redirected bass will be measured and the level will be correct. This method works for both questions.

I'm assuming you are redirecting bass (your speakers are set to small)? What is the AVR and its current XO setting? Did you use an auto setup feature in the AVR or set the distances manually? What are the main speakers and SW? I am looking for a little context here so I can better understand any future data you provide regarding SW to FL timing.
hello jtalden,
Thanks for the reply. Here are the measurements. Unfortunately, for subwoofer only measurements, I only have measurements with signal sent to LFE channel, so the crossover at 110Hz is not included. But I am concerned for the dip between 50Hz and 60Hz, so in that range it is comparable to the redirected signal to a front speaker.
Signal to sub(LFE) is with AVR set at -40dB (and the AVR adds 10dB). Signal to front is at -35dB. I did not shift the curves in REW.
Setup is 5.1 system; small GoldenEar speakers with crossover set at 110Hz. Subwoofer is B&W ASW608.
Distances for front speakers were set manually, but the AVR does not have a setting for the sub distance.
The AVR does not have Audyssey.
Thanks for the help.
 

Attachments

·
Premium Member
Joined
·
1,562 Posts
A second SW will not significantly help the smoothness of your room response if the 2 are located in the positions you measured. Multiple subs can be large improvement. but the locations in the room are very important as to how much help they are in that regard.

Looking at the Data:
From the LFE IRs shown below we can see that REW aligned the largest peaks at 0 ms very accurately.

ir 3a.jpg

Normally I would expect the initial peak to be the largest and representative of the direct signal from the SW. If we realign IRs to the initial small peak we find that the time shift needed is only 1.185 ms or about 14" in distance as shown below.

ir 3b.jpg

This is not enough to be significant as we can see in the resulting calculated SPLs below. One trace is with the timing aligned as REW calculated and the other is with one manually shifted 1.185 ms to align the initial peaks.

spl 3a.jpg

The measurements you provided regarding the mains are not the info needed to manually adjust timing to see if improved SPL can be obtained with different distance settings. I had requested a measurement of the FL main without the SW. The only overall observation that can be made is that there not a major change to the responses whether the SW is placed in the corner or in the middle. There is a little more SPL when placed in the corner, but that could be offset with a minor level control adjustment.

As far as SW to main timing you can check that out experimentally yourself using the REW RTA feature. I can help with that process if needed. It is relatively easy. I will not be available today however.
 

·
Registered
Joined
·
25 Posts
Discussion Starter #17
A second SW will not significantly help the smoothness of your room response if the 2 are located in the positions you measured. Multiple subs can be large improvement. but the locations in the room are very important as to how much help they are in that regard.

Looking at the Data:
From the LFE IRs shown below we can see that REW aligned the largest peaks at 0 ms very accurately.

Normally I would expect the initial peak to be the largest and representative of the direct signal from the SW. If we realign IRs to the initial small peak we find that the time shift needed is only 1.185 ms or about 14" in distance as shown below.

This is not enough to be significant as we can see in the resulting calculated SPLs below. One trace is with the timing aligned as REW calculated and the other is with one manually shifted 1.185 ms to align the initial peaks.

The measurements you provided regarding the mains are not the info needed to manually adjust timing to see if improved SPL can be obtained with different distance settings. I had requested a measurement of the FL main without the SW. The only overall observation that can be made is that there not a major change to the responses whether the SW is placed in the corner or in the middle. There is a little more SPL when placed in the corner, but that could be offset with a minor level control adjustment.

As far as SW to main timing you can check that out experimentally yourself using the REW RTA feature. I can help with that process if needed. It is relatively easy. I will not be available today however.
hello jtalden
Thanks for the support. I understand that it is not expected to have improved response with 2 sub operation with the positions I have tested so far. I would like to ask some more questions:
- as no improvement is expected, I will probably stay with 1 sub. Which sub position do you expect to be best AFTER PEQ (which I intend to do in future): my corner or middle position?
- How to do the timing with RTA? Do I need separate front measurement and sub measurement?
 

·
Premium Member
Joined
·
1,562 Posts
I do not see a significant difference between the corner and middle SW positions with regard to EQ. My expectation is that they will both respond the same. The large dip between 50 and 60 is the result of a room mode and it will probably create a bigger problem if it is addressed using EQ. It is better to leave deep narrow dips alone.

The RTA alignment method runs a main+SW at the same time. I will provide a procedure for the RTA method of alignment, but I first need the following info.

> What are what are the physical distances of all your speakers, including the SW?
> Does your AVR accept negative values for the main speaker distances?
> What are increments for distance in the AVR?
 

·
Registered
Joined
·
25 Posts
Discussion Starter #19
I do not see a significant difference between the corner and middle SW positions with regard to EQ. My expectation is that they will both respond the same. The large dip between 50 and 60 is the result of a room mode and it will probably create a bigger problem if it is addressed using EQ. It is better to leave deep narrow dips alone.

The RTA alignment method runs a main+SW at the same time. I will provide a procedure for the RTA method of alignment, but I first need the following info.

> What are what are the physical distances of all your speakers, including the SW?
> Does your AVR accept negative values for the main speaker distances?
> What are increments for distance in the AVR?
hello jtalden,
The distances are:
Front Left and Right 2.8m
Front Center 2.7m
Surround Left 2.6m
Surround Right 1.8m
Subwoofer 2.7m (subwoofer distance cannot be set in the AVR)

Negative values are not possible in the AVR setting. Increment of 0.1meter.
 

·
Premium Member
Joined
·
1,562 Posts
I don't know what your AVR manual says, but would expect it to suggest setting the mains to settings as shown in the table below (SW distance + "Add-on" - main distance). The small Add-on value is just to account for the additional delay that occurs in a SW with a digital P-Amp. An Add-on may not be included as its impact is small in practical terms. I just selected a 0.3 m value as a reasonable one for a typical SW based on comments from others in various forum posts.

Table 5a.png

If there is conflicting information in the AVR manual regarding the setup then my procedure below may be erroneous for your AVR.

This general advice for setup should work well assuming that the polarity of the SW and the midwoofer in the mains are in the same polarity. There are also influences from the room response, speaker locations, LP location and the interaction of the XO with the main speakers. These influence the SPL response in the XO range as well.

RTA alignment method for your particular case:
[See the table below.]
1. Measure distances from the LP to the SW and the mains with a tape measure (Columns A and C). Enter a 0.1 m Add-on (Column B)
2. Calculate the initial AVR distances settings (A + B - C) and enter them into Column D and into the AVR.
[The 0.1 value is just the minimum add-on needed to avoid an initial negative distance setting for the FL/FR.]
3. Place the Mic at the LP.
4. Route the REW signal to the AVR such that it will activate only the FL channel. [We want the signal to play through only the FL and the SW (FL+SW).]
5. Open and set the REW Signal Generator to "Pink PN"
6. Open and set the REW RTA mode [RTA 1/48 octave and Rectangular window]
7. Start the Signal Generator and the RTA (be sure your levels are set reasonably).
8. Watch the SPL in the XO range (maybe 50-230 Hz in your case). Increase the FL Main distance in increments to find the setting that results in the greatest SPL fill in the XO range.
9. Record the setting (column E).
10. Repeat 5. thru 9. for the FR+SW and CC+SW.

Table 5b.png

We hope to find the FL, FR and CC mains offset (column F) pretty similar, maybe within 0.5 m of each other. In the example table the FL/FR/CC was found at 0.3/0.5/0.6 m. This is a very good result. We can pick the 0.5 m offset as a good compromise for the 3 mains. The final distance settings (Column G) would then be the "initial settings + 0.5" for all main speakers as shown. We want to keep all the mains distances with the same relative relationship to each other that we started with. So we picked a compromise offset based on the 3 front speakers and move all initial distance settings up by that same offset.

Since 0.1 m increments is very small change we may find it easier to initially move in 0.4 m increments. We can then fine tune after we identify the general area. The expectation is that the initial recommended settings per the top table would be pretty good, but if its relatively poor, the SW polarity may be opposite of the Midwoofer in the mains. In that case we may find a better SPL fill about 1.6 m higher than those initial distance settings (1/2 wavelength at 110 Hz). We shouldn't need to move above 2.5 m to find a good setting.

If we do find the best SPL fill at a distance >1.6 m then we may want to change the polarity setting on the SW and start again. It is okay though to use either polarity setting of the SW so long as the SPL fill is good. You may prefer the sound more one way or the other. Let us know how this works out.
 
1 - 20 of 30 Posts
Top