Another multi-sub setup

[jtalden]

Starting with the impulses near 0 ms is the manual way to get in the ballpark. We can't just align the impulses as we need to assure the phase in the XO range is closely aligned as well. We could otherwise have a large dip or sag in the SPL. The new REW alignment tool is makes this job easier as it helps us look a different settings for both SPL and phase at the same time. We pick the timing we like and then create new time aligned copies of the original measurements.

In your case SPL support through the XO range is dependent on the distance settings in the pre-pro. It is often a problem to get good results as the room effects made it difficult for either an automated systems in the pre-pro or for someone setting distances manually.

Your measurements suggest the distance settings need to be adjusted. There also appears to be a significant difference between the phase in the left and right mains in the XO range so a distance setting that works for one will not work well for the other. A compromise setting will be needed minimize the size of the dip.

First, It doesn't appear the mains are set to small in the pre-pro. There is a very slow rolloff of the mains in the low frequencies. Some people do this on purpose. It is usually referred to as "double bass". If this is not a choice you intended, I suggest you first set the mains to small and chose an XO frequency. That is the standard setup recommendation. You would then need to provide the 3 measurements I need for the process using; mic at the LP, full range measurements, and REW acoustic timing.

Measure:

1. SWs (as a group with EQ on and using the left channel for the output with the SWs turned off)
2. Left (alone- no SWs)
3. Right (alone- no SWs)
4. Optionally, Center (alone- no SWs)

I will recomended distance settings that are favorable.

 

[Conrad.]

I've confirmed that the mains are set as you describe: crossed over (as opposed to full range) with a crossover value of 100Hz. Once the speaker size is set to Crossed Over the eBass option becomes disabled. That's the terminology my processor uses; eBass is the setting whereby bass is left in the main channels and also sent to the subwoofer(s).

I know that my processor (Classe CT-SSP) has a fixed setting of 12dB/Octave on the mains. This is regardless of the crossover slope setting which only affects the slope of the subs. The LFE channel also has a fixed 24dB/Octave slope, set at 80Hz. I'm not using the LFE in these measurements.

One other thing to note, distance settings in my processor are in 2" increments. I had previously time aligned each of the mains, but it looks like the LCR are aligned but not time aligned with the SBL (which is what I use as a timing reference). Maybe the centre could be delayed by another couple of inches.

As I mentioned, I don't have any auto EQ available, I only have 5 band PEQ per channel. I'm not using any for the subs (it's all in the MiniDSP) and I'm using two or three filters to achieve the proper target slope for the mains.

Given the above, I think the measurements you've asked for are included in my latest file.

1. SWs = Measurement 6 - All + EQ
2. Left = Measurement 7 - Left
3. Right = Measurement 8 - Right
4. Center = Measurement 9 - Center

Let me know if that's not the case, or if you want the measurements of the mains without the EQ.

 

[jtalden]

Okay, thanks for confirming. I will use that data then.

 

[Conrad.]

Thanks. I've been looking at the alignment tool as well and the best I could find was a -3.26ms delay on the subs. That seems to give a slight improvement in the right and center channel integration, with a significant improvement in the left channel integration.

It'll be interesting to see if you arrive at the same value.

 

[jtalden]

This requires a compromise design. There are different ways to approach it and different resulting answers. I decided to first average left and right mains to find the favorable alignments for SW + (L+R) as I labeled them in the attached. All 5 alignments found in the file are favorable alignments for SPL support. The most conventional alignment is measurement 15, but it also provided the weakest SPL support. You can pursue the results and select any you like to try out. The one with the most SPL support is measurement 7 and it also has the most leading SW impulse setting. The others are in between these two extremes. I would expect significantly different sounds between some of them. If I were choosing a few to try, I might pick those two and maybe 13. Measurement 11 and 13 have the SWs inverted and the SPL support of 13 is better. It might be interesting to try if you are into experimentation.

 

[Conrad.]

That's great, thank you.
It doesn't look like too much work to try those measurements out. I'm certainly interested to see how what appears to be the same results would have different sound signatures. If I'm reading it right you're only moving the subs as a unit?

I understand that you're looking at the combined L+R response, which makes sense. When I look at the individual responses merged with the subs, each of those positions looks better for L but worse for R?

I guess it's free to try so I'll give all the positions a listen at the weekend and report back.

Thank you very much for all your help and the time you've dedicated to my set-up.

 

[Conrad.]

Ha. Every time I look again I learn something new about REW. I see what you mean now when you say "has the most SPL". I always though the numbers next to the measurements were the values under the cursor but you've shown me that you can use them to reflect the (I guess) average level in the viewport. By zooming in on the crossover region you're looking at the totals rather than just looking at the shape. It's always easy to misinterpret when just eyeballing, my eyes are drawn to the peaks and troughs, which perhaps isn't the most effective way to analyse the responses.

 

[jtalden]

If I'm reading it right you're only moving the subs as a unit?

That's right we adjust the SWs as a group to align with the mains.

I understand that you're looking at the combined L+R response, which makes sense. When I look at the individual responses merged with the subs, each of those positions looks better for L but worse for R?

There will be big differences between left and right. I didn't look at which will be worse as I did not do that analysis. I thought it would be quicker to find favorable SPL support using the average. It would have been difficult to do them separately as there is no setting that will look good for both. That leads to continuous searching for a good compromise answer.

 

[jtalden]

Ha. Every time I look again I learn something new about REW. I see what you mean now when you say "has the most SPL". I always though the numbers next to the measurements were the values under the cursor but you've shown me that you can use them to reflect the (I guess) average level in the viewport. By zooming in on the crossover region you're looking at the totals rather than just looking at the shape. It's always easy to misinterpret when just eyeballing, my eyes are drawn to the peaks and troughs, which perhaps isn't the most effective way to analyse the responses.

??, The SPL in legend is the value at the cursor. I did not set the cursor in this case so it is probably at the minimum (17 Hz). I did just zoom in on the XO range so you could see the subtle difference there and not be distracted with other info. The differences show are trivial as all these are favorable alignments and will accept applied EQ well.

 

[Conrad.]

Ok, sorry, I misread things then. I assumed that the total db measurement without a cursor was the average of what was in the view.

I'll play with the delays tomorrow and report. Thanks.

 

[Conrad.]

I watched Scary Stories to Tell in the Dark last night and it sounded awesome. I've had to readjust some things this morning (turn down the rear sub so I re-measured and re-EQd but didn't need to touch the time alignment) and I re-eq'd with a slightly less aggressive house curve. I'll keep tweaking but it's sounding as good as ever.

Thanks @jtalden!

 

[jtalden]

 

[CAAudiophile]

When having multi Subwoofers in a room, time/phase alignment is crucial.
Need to use the Default sitting for each Subwoofer, meaning no EQs are applied at this time.
Take the frequency response to each sub at the same volume level you may need to do this many times until all are at the same volume level. Take the Impulse Response of each of the subwoofer, at this time (This will show you if there are Time delay(s) between the subs, not much can be done here since Subwoofer wavelength are long) now you would see if any of the subs are in phase with each other if not use the Phase switch for that sub out of phase and re-run the frequency response and again the Impulse Response to make sure all are in phase.
Then look at the frequency response curve for each subwoofer individually and adjust the EQ on each of the sub or the DSP if it is in use for each sub according to the sub's Freq. response. Adjusting the dip and peaks accordingly for each.