Time alignment of three subs

[IWC Doppel]

Hi all, I have followed a number of vides and think I have time aligned my subs, these have a small amount of EQ 6db at the two room peaks of 34 and 68 Hz, but not yet EQ;
'd beyond that,

Sorry for clipping the frequency response, the curser black line is about 30 Hz, I think I have a null at 20 Hz, annoyingly. The graph below shows each sub alone then combined actual and combined calc (top two curves)

But the question I have is what when I have set the ms delay to give me this response if I add say 7ms to each sub delay why does the combined curve change ? I thought the phase was relative to each sub ? So once you have say 5,3,3 then 12,10,10 in delay should give the same curve with nothing else changed ?

Sorry for not posting the 5,3,3 LFE response with all three subs running vs the 12,10,10 unfortunately this response wasn't saved when REW crashed on me last night

The combined response below for all three subs is the top curve (green actual, pink calculated responses) and the delays are 5,3,3 this graph looks deradful if I change the delays to 12,10,10

 

[AustinJerry]

Time alignment is the adjustment of phase to account for the difference in distance of two sound reproducers as measured from the listening position. For example, say you have two subs, one 10 feet from the MLP, and one 5 feet from the MLP. The sound from the nearest sub arrives at your ears sooner than the sound from the furthest sub, so you must delay the sound from the nearest sub to align the two signals. This is done by subtracting the two differences to arrive at the distance difference, in this example 5 feet. Then, the distance difference must be converted to milliseconds using the formula (Distance difference/Speed of sound). So for a 5 foot difference, the delay would be (60/13512)*1000 = 4.4 milliseconds, where 60 is the distance difference in inches, and 13512 is the speed of sound in inches/second. Multiply by 10000 to convert to milliseconds.

So, to time align the two subs in this example, add 4.4 milliseconds to the delay for the nearest sub, and leave the delay for the furthest sub at zero. Now, make these same calculations for your three subs.

Changing sub delays affects the combined sub signal, not the individual sub measurements, which is why you are seeing the combined sub measurement change when you adjust delays.

And if you want to improve the overall combined sub response, consider using a device like the MiniDSP 2x4, calculating PEQ from your combined sub measurement using REW's EQ tool, and importing the PEQ filters into the 2x4. PEQ should be quite effective in flattening out the broad increase you are seeing above 30Hz. This 2x4 setup guide should be useful if you decide to go that route.

And finally, when posting REW measurements, consider using the small "camera" icon on the measurement page to capture the screen image. That way the horizontal scale will be shown so we don't have to guess what your measurements are showing.

 

[Praba]

Hi all, I have followed a number of vides and think I have time aligned my subs, these have a small amount of EQ 6db at the two room peaks of 34 and 68 Hz, but not yet EQ;
'd beyond that,

Sorry for clipping the frequency response, the curser black line is about 30 Hz, I think I have a null at 20 Hz, annoyingly. The graph below shows each sub alone then combined actual and combined calc (top two curves)

But the question I have is what when I have set the ms delay to give me this response if I add say 7ms to each sub delay why does the combined curve change ? I thought the phase was relative to each sub ? So once you have say 5,3,3 then 12,10,10 in delay should give the same curve with nothing else changed ?

Sorry for not posting the 5,3,3 LFE response with all three subs running vs the 12,10,10 unfortunately this response wasn't saved when REW crashed on me last night

The combined response below for all three subs is the top curve (green actual, pink calculated responses) and the delays are 5,3,3 this graph looks deradful if I change the delays to 12,10,10

View attachment 33445

Its got to do with the rounding error -- to use distance units like cm instead of ms delay may be better. 1 ms is approximately equal to a distance traveled by sound of 1 ft. If the rounding error in the differences between the 5,3,3 combination and the 12,10,10 combination will have to seen in the context of the changing distance of the reflecting surfaces that produce the interference pattern. Please note that 5 to 12 may be rounded differently than 3 to 10. The speaker that is closer to the nearest reflecting surface will be affected more by the error. Room temperature that affects the speed of sound could also play a role.

 

[AustinJerry]

Its got to do with the rounding error -- to use distance units like cm instead of ms delay may be better. 1 ms is approximately equal to a distance traveled by sound of 1 ft. If the rounding error in the differences between the 5,3,3 combination and the 12,10,10 combination will have to seen in the context of the changing distance of the reflecting surfaces that produce the interference pattern. Please note that 5 to 12 may be rounded differently than 3 to 10. The speaker that is closer to the nearest reflecting surface will be affected more by the error. Room temperature that affects the speed of sound could also play a role.

Sorry, but did you read my response? No disrespect, but your advice just makes no sense at all.

 

[IWC Doppel]

I will remeasure again as I am still unsure why changing the absolute delay for the subs as a group and retaining the relative delay between them should make a difference, the delays are set in ms. I also realised that sub alignment is different in each seating position which does make sense

 

[Praba]

It appears that my explanation has not been understood clearly. Let me spell it out.
Sound recorded at the microphone is not just the direct sound from the speakers, but the result of the interference of all the waves produced at the sources, and the delayed reflections from the various surfaces. This is not unlike throwing several pebbles in the still water in a bathtub. The complex wave patterns produced are the result of interference, and that is what is recorded, and not the clean sum of individual, single, pure waves.
Now, for the delay input values given in ms, the arrival delay at the MLP will depend on the speed of sound and the distances of directed and reflected sounds. The speed changes from about 330 m/sec to about 360 m/sec from 0 deg C to about 50 deg C. So, the interference pattern could change with temperature. In most systems, you will be asked to enter the speed of sound or the temperature, if you want to enter distances instead of delay times, for this reason. Therefore a change in temperature of the room, or small change in position or the angle of placement of the speakers could also explain your findings.
But, my suggestion for time alignment is to use the spectrogram in REW. The red regions of spectrogram will show you the timing of maximum output by frequency. For each sub, try to get the optimal output to line up with the main speakers. I say optimal because it has to be found by trial and error because the red region will be smeared wide in time. You also have to recheck the phase between subs, and the phase between the subs group and the mains, whenever you change the delays.

 

[IWC Doppel]

Praba, hi

I'll admit I am really struggling with this at the moment, I have an annoying in built need to understand which sometimes is useful other times frustrating. I am starting to pick up some pointers but I am finding very real changes when I just record the LFE channel or the LFE only turned on for the center, or left or right and just change the sub delays, when only listening to the sub, if I try 2,0,0 or 5,3,3 or 12,10,10 All very different curves.

Here is my LFE at the moment (below the sim) with three subs only green is time alignment 5,3,3 blue just changing to 12,10,10 Both have exactly the same settings and they both have PEQ filters of 6db around 34 Hz and 4db at 68Hz. I am trying to understand why I am so light at 20Hz and what I might or should do about it, the panels for the subs have a manual low pass at 80 Hz (I can't go any higher without removing this within the subs, in anycae I currently crossover at 50 Hz). I know with my room simulator I should expect to see a delta of some 30db between before my filters so I believe this is my challenge getting 20Hz in this room ? Have looked at how the position in the room will improve this and I will be able to utilise the corners more when I move to 7.3.0 which is imminent but even then the best part of a 30db delta starting point.

Interested in trying Spectograph, as an example this is what my Centre shows with the subs at 180 degrees out of phase, seems the centre looks better with the subs (all back of the room) out of phase and the FL,FR look better in phase . I must admit I dont know if this is good or bad !

 

[Praba]

Hi
What is the receiver/processor you use? Select something other than All Channel Stereo, like Dolby surround output on you receiver. Then do a measurement for each channel to see how the sub integrates with each of the channels. If you see a dip at the crossover point, try changing the phase for that channel. Remember that phase will change with delay. You absolutely have to get the phase correct for all the channels.
Your spectrogram suggests that the subs are a little late. So, you perhaps should not be using delays on the subs. You can increase the distance for the subs in your receiver to advance them. Timing and phase have to be checked every time. These are the most important.
Sometimes subs perform better if they are off the ground by a foot or so.
You may need acoustic treatment to improve clarity in the lows. There are two types velocity absorbers and pressure absorbers. Do a little research before jumping into these. Their usage and location for optimal placement should match your situation.

 

[IWC Doppel]

Thanks, the processor is an Acurus ACT 4, the processor has delays for each sub and each channel. Do I simply ignore the classic maths of how far things physically are from the MLP and just use that as the staring point ? I'd love to read up on phase vs delay, I had seen quoted that they are the same if the frequency is the same, just maths on 1/2 was length ?

Best

 

[Praba]

Change the scale on the Y-axis of the spectrogram. This would give you better idea of the spread of the high-energy content by frequency. A delay of less than 20 ms in the lower sub range is not bad. In the sub frequencies, 30 to 40 Hz is the most critical for explosions or gunfire in movies like The Accountant. The lower energy levels detected between 45 and 60 Hz is more important, and that has to be addressed. If this because of a null in the room response, you can try moving the subs a little to try and fill out that region.
The frequencies below 30 Hz will not be handled well by most subs of sealed box or ported designs. You will need something like a tapped horn to handle those frequencies well. But they too come with their own problems.
Page 23 of the user guide for ACT 4 talks about adjusting phase for all channels independently. Check the effect of changing this for each channel, one at a time.

Your crossover frequency, I will suggest at 80 Hz and not at 50, and use direct inputs to the subs, bypassing the internal crossovers built-in to the sub. The lack of energy in the 45 to 60 Hz range could be because of this choice of 50 Hz crossover.

 

[IWC Doppel]

Here are two images of ,my centre channel cover at 50 Hz and 60 Hz, I prefer the sound of the 50 Hz over when listening

The SPL look as shown below green 40, pink 50

 

[IWC Doppel]

PS sorry about the thick lines I'll look at why the pictures don't look like my screen !

 

[Praba]

What is the model of your subs? Are they all the same? The 60 Hz plot shows a wider red region. Your mains do not appear to be producing sufficient output below 70 Hz. Your perception might change when you get all parameters right at 80 Hz. My advice is not to get comfortable at a setting that measurements indicate are sub-optimal.
The delay of the low frequencies indicates that you need to introduces delays on other channels except the subs. To be more precise, align all your subs first. Then align all the other channels, except the subs. Ensure the polarity matching of all channels with the subs. Then uniformly increase the delays on all the channels in steps, checking the polarity at each stage.
Your magical spot will be when the energy output from all the speakers are in phase, at the same time, and uniform across neighboring frequencies, as heard from the MLP.

 

[IWC Doppel]

Tks the plot is from the centre channel only, not mains. The subs are Sonus Faber Cremona M's

 

[Praba]

You cannot assume that your subs are in phase without measurement, just on the basis of the position of the phase switch.
The manual says that it has a switchable low pass filter. It is imperative that you switch it off. Multiple filters interfere and create other problems as well. Your receiver is the place you need to set the frequencies to be handled by the sub.
You have phase reversal switches on the subs. There are three ways in which these can be paired. All three possible pairings should be in phase. This can be confirmed through measurements. If they are out of phase, output will be reduced, and the bass will sound anemic. When you have done this all three will be in phase, and sound louder.

 

[IWC Doppel]

I have measured each sub and time aligned with reference to the SPL and using the alignment tool as well as checking the addition of each sub so 1+1+1 -= 3 and not less, I haven't made any assumptions ? Its the alignment of the crossovers to the center and mains that I am finding more difficult as the alignment tool says one thing, but doesn't measure well when trying ? Also the understanding of why I am showing differences with the aligned subs when I just add delay to them all equally and only measure them

 

[Praba]

Basically things are all interconnected. Everything affects everything else. Moving the subs by a few inches can make a big difference. Additional delay of 5 ms is same as moving by 1.7 m. The receiver has to make complex calculations for figuring out how to convert the input signals to account for the different delays between the subs and the other channels to come up with a signal to send to the subs, summed over all the channels for every sampling instance. When the delay changes, the signal to the sub changes as well, as a result of this calculation. So, the interference pattern changes completely.
You will not get better results without using the measurements to guide you about the channels that need the delay, and the extent of the delay. Once you are approximately in the optimal area, you can then make very small corrections (equivalent of a few centimeters) by the ear. Even then, you should periodically check to see if you have drifted off too much.
So, I would say that you should concentrate on your measurements, and follow the best practices, some of which I have mentioned already. It will be a shame not to get the best out of your very expensive system.

 

[IWC Doppel]

Praba, hi

None of what you say is new to me or anything I disagree with, hence the reason for my questions about the specific measurements. I'm new to REW but have spent many years setting up two channel systems by ear and very aware about how important position of the main speaker are for example. My reasons for raising questions are indeed very much related to getting the best out of my system and I do want to learn enough be able to do this and spend the time required with REW and the analysis. I have invested a fair bit in the system over the years, so couldn't agree more in making sure it performs as well as it can !

 

[IWC Doppel]

I am still floundering here to be honest, I have now read for hours and hours and just taken more readings to see if I can align the subwoofers. Very frustrating I am finding significant inconsistencies and happy to do it empirically, can people confirm that if I do LFE sweeps and simply try delays then review the combined curves I can see where they combine successfully as the phase alignment is telling me things, I think, that don't appear to work then testing the results - ie the maximum output is what I am seeking

 

[Praba]

How is your REW setup? If you use USB microphone with HDMI, you need to use a timing reference signal. The steps to follow are:
Make sure that when you do the sweep, it plays through only one channel at a time (No multi-channel sound)
Set your reference, say as the center channel. For each of the other channels, measure the delays using a sweep (for non LFE channels, I prefer 180 Hz to 2800 Hz with subs muted), with respect to the reference. You will see the results in the window on the left. Record the delay values in a paper for each channel. Positive values mean arriving late, and negative means early.
Do the same for each of your subs, one by one, using a sweep from 25 to 75 Hz. Once again, do not use the crossover in the subs, and use 80 Hz in the receiver. With a gap in frequency, you will never be able to get a good integration.
Find the last arriving response from all the channels. Increase the delays in all other channels to match that.
Recheck, and fine tune.
Start a fresh set of measurements.
Now, run the sweep from 15 Hz to 20000 for the LFE with only one sub. Beyond about 150 Hz, there will be no response. But that does not matter.
Next mute the first sub, and do a sweep on the second. Then do a sweep with first and second. Then with the polarity inverted only on the second.
Now compare the 4 recordings in the All SPL tab with 1/3 Octave smoothing. The phase position that gives the maximum output for the two subs should be selected.
Next, mute the first two subs, and run a sweep of the third. Then run a sweep with all turned on. Repeat with the polarity of the third reversed. Select the best of the two last measurements to set the phase accordingly.
Note that the output of all the subs should be nearly equal.
Now with the three subs on, run the sweep on Left channel. The sub level will come down from the sweep for LFE. But should be higher than the average for about 80 Hz and above. Repeat with reversing the polarity on the Left channel. Select the polarity that produces higher output around 80 to 100 Hz.
Repeat the above for other channels.

I too have 3 subs in my system. Even though they are all different DIY designs -- 2 18 in sealed box, and 1 15 in 5m tapped horn, and use the tapped horn for <44 Hz, and the sealed boxes for 44 to 80, I have been able to get good alignment.
I was struggling with a hit or miss approach before REW. I too believed previously that I could tune by the ear from years of experience. I was wrong!

 

[IWC Doppel]

Praba, hi

I have set up REW with a UMM-6 mic, acoustic reference and I believe the timing references are correct. I have done something similar, so will, try your method

I have done LFE sweeps for each sub using FL as the timing reference and then used the time alignment tool to 'align at cursor' around the xover frequency and also checking 20,30,40,50,60,70 and 80 Hz there was good consistency with the required delay. I was about 2ms out on one sub to another then 1ms when adding he third. I then double check by seeing how the combined sums using trace arithmetic should look like and how they measured. The provided the strongest SPL combination of all three subs in the MLP. I also checked the curves of each sub at the Left and right seating position and combined.

Looking at how they combine using the simulator and numerous readings I was able to EQ individually so that if one sub was in need fo a greater cut at on node I could 'weight' the cut on that sub so the combination would indeed come down but I was also improving cross seat consistency

I have a few challenges :

1. The subs have a frequency cut off control that can't be bypassed and has a maximum of 80 Hz
2. The subs have an inbuilt low cut of at 19Hz, I don't know what impact this has at 20Hz

I looked int he back of the subs to see what I might disconnect but goy t lost, then I have an interesting idea.... I have build another sub for my Horns but not finished it yet and use a really nice 500W with 1,000 W peaks base amplifier that have a massive 700VA transformer, so I plugged this in and used it to drive one sub to see how it sounded, it measured very similar but seemed to sound a little cleaner down lo and this had LFE bypass and additional phase as well as more power (400W internal amps) so I may buy three of these amps and drive the cinema subs with these in time !

Anyway back to my current challenge - with the subs aligned just measuring the LFE channel only the plot changes if I add delay and I just dont understand this. So LFE's on their own I add 5 or 10ms to each sub and retain their relative alignment then measure LFE only and the plot changes ? very confused about this !

Using the alignment tool and measuring full sweeps of L,C,R with timing references with no LFE and LFE only I believe I need to add something like 15ms of delay, but suing the graphs and experimenting I can get nearly as good response by 4ms for each channel so I will try this and measure, but I suspect that just adding 4 Ms to each sub will affect the subs

 

[IWC Doppel]

This is the plot of my centre measured blue and with 4 Ms additional delay from the LFE channel, I will then look to EQ the 68/70 sonf 90 peaks

 

[Praba]

Your LFE channel shows a delay of 5.3943 ms (bottom left of your image) with respect to the acoustic reference on L. You seem to have excess delay on the subs. When you change that, aligning each sub, and of the other channels, one by one to the Left channel, you need to recheck your polarity, for each sub, and the integration of LFE with each of the other channels. When everything is properly aligned, it could remove the ripple around the 80 Hz crossover frequencies without equalization.
I suspect that the total LFE output is too low in your plot of the Center channel. If you run All Channel Stereo, with all channels enabled with subs, you should get a flat response on a sweep from 20 to 20000 Hz. If there is a dip below 80 Hz, you need to increase the output of the subs

 

[IWC Doppel]

Your LFE channel shows a delay of 5.3943 ms (bottom left of your image) with respect to the acoustic reference on L. You seem to have excess delay on the subs. When you change that, aligning each sub, and of the other channels, one by one to the Left channel, you need to recheck your polarity, for each sub, and the integration of LFE with each of the other channels. When everything is properly aligned, it could remove the ripple around the 80 Hz crossover frequencies without equalization.
I suspect that the total LFE output is too low in your plot of the Center channel. If you run All Channel Stereo, with all channels enabled with subs, you should get a flat response on a sweep from 20 to 20000 Hz. If there is a dip below 80 Hz, you need to increase the output of the subs

Thanks, I am currently running 0 delay on the centre and 5ms on sub 1, 3ms on sub2 and 3ms on sub 3, very interesting it sees the delay as 5.4 ms. I did try inverting the subs and the LFE plot looks identical (thankfully !) and the centre channel looks better but the FL, FR looks awful inverted

 

[Praba]

So, invert the FL and FR so that all look good.