Rythmik L12 Subwoofer Review

[xovaan]

Great info-- thank you again!

When you say a time alignment using the main speakers, am I following your steps correctly in the UI?

 

[CumSum]

Great info-- thank you again!

When you say a time alignment using the main speakers, am I following your steps correctly in the UI?

Not quite, follow the steps here: https://www.minidsp.com/applications/auto-eq-with-rew/speaker-time-alignment

 

[xovaan]

Reading through now, thank you-- so I:

1. Set an acoustic timing reference for my left speaker (which I will use as reference for the rest of the system) -- output set to L and reference set to L and run a sweep with settings shown in the screenshot below
2. Change output to R (right speaker) and run another sweep, keeping all other settings in this window exactly the same
3. The guide you linked says that my reference speaker (in this case, L Speaker) should be very close to or exactly 0, however in my test using I'm seeing "Delay -0.6733 ms (-231 mm, -9.09 in)" for my system delay. Not sure what I'm looking at incorrectly here?

 

[CumSum]

Reading through now, thank you-- so I:

1. Set an acoustic timing reference for my left speaker (which I will use as reference for the rest of the system) -- output set to L and reference set to L and run a sweep with settings shown in the screenshot below
2. Change output to R (right speaker) and run another sweep, keeping all other settings in this window exactly the same
3. The guide you linked says that my reference speaker (in this case, L Speaker) should be very close to or exactly 0, however in my test using I'm seeing "Delay -0.6733 ms (-231 mm, -9.09 in)" for my system delay. Not sure what I'm looking at incorrectly here?

In the guide they are testing the left channel against itself because that should definitely be (very close to) zero. I mentioned sweeping from 0-1000Hz, that is applicable for subwoofers, so for time aligning speakers, try what the guide suggests and do it from 1000-20000Hz. Play with volume and/or sweep range until you get a number close to zero.

Afterwards the Ref Output will be the left channel and Output will be the right channel only. You have to make sure you mute the subs or speaker that is not under test. So if testing right speaker, both subs muted. If testing sub 1, sub 2 and right speaker are muted, and so on.

Read this as well: https://www.roomeqwizard.com/help/help_en-GB/html/makingmeasurements.html#acousticref

 

[xovaan]

Gotcha-- how close to zero are we talking? I'm consistently getting 0.08ms out of the left speaker between 77 and 81dB in a 1000-20000hz sweep, same with the right. Same at various other sweep frequencies as well.

 

[CumSum]

Gotcha-- how close to zero are we talking? I'm consistently getting 0.08ms out of the left speaker between 77 and 81dB in a 1000-20000hz sweep, same with the right. Same at various other sweep frequencies as well.

That is essentially zero. Under 1ms is what you should be aiming for at the least. Treat under 0.1ms as zero. Move on to the subwoofers.

Here is one thing to keep in mind when running DIRAC. The very first measurement DIRAC takes will determine the volume gains and time delays for the input Left and Right channels (not to be confused with the outputting channels). So DIRAC is performing its own time alignment and volume matching. These values are NOT user adjustable.

I only mention this because on rare occasion I have seen DIRAC apply say 4ms of delay to one of my channels when in fact it should have been on the order of 0.1ms. So I had to retake the very first measurement, which fixed the issue. So just double check that those values inside the MiniDSP after running DIRAC look like they make sense.

 

[xovaan]

Ahh, I see now-- so Dirac 1 is left channel, so I use my left speaker as the reference pitch thing, which remains the only thing on the left channel, and begin testing on Dirac 2 for every speaker / sub combo (all of which will be "R" in output selection in REW) to sync to it? Here's what I get when I do that:

Here's my time alignments based on the above:

 

[CumSum]

Yeah you can mute the channels not being tested like that, though it would be easier to use the mute button in the outputs page instead. Your finalized channel matrix should be similar to mine with the Left speaker getting DIRAC 1, Right Speaker DIRAC 2 and then the subwoofers receiving DIRAC 1 and 2.

Did you adjust the sweep frequency for the subwoofer? If the sub is located next to the speaker then the delay should be small. I would mess with volume and sweep frequency a bit to make sure the readings are consistent. The reading should make logical sense in terms of distance. It won't be exact to distance as other delays can exist (like DSP that may exist inside of a sub, speaker or amplifier).

And when you finish time alignment, do a measurement with the speakers and subs all at once to make sure they are constructively interfering (in phase). With that measurement you can also take a look at the wavelet. It will show you if the speakers and subs play together at the same time in the time domain. You can view the wavelet in the Spectrogram window, hit controls and change the mode to wavelet.

Small explanation of that here

 

[xovaan]

Yeah, the distance calculator is pretty much spot on-- my left sub is roughly over a foot behind my left speaker, and right sub and right speaker vary a few inches from the mic position. Here's my measurements adjusted for delay:

First I add delay to the right speaker and sub to match the front left speaker, then I add .9088ms to all other channels to zero and match everything to the sub. Does my math check out?

Watching video now and checking the sweep! When I try to run all speakers and subs together, I get a really high dB-- should I lower the volume for the test? Here's my subwoofers combined after alignment:

Green is both subs, blue is both speakers, and orange is all combined

The above is with a crossover applied (80hz 24dB LR to speakers and subs)

 

[3dbinCanada]

xovann, I meant to post this much earlier for you but I got distracted with housecleaning and forgot to post this. I'm glad CumSum stepped in. Here is a rather lomg tutorial that I followed to dial in 3 subs in my great room.

This LINK deals with gain matching the subs so they are working equally.

The next link shows how to use REW and how to measure the subs. I used REW's swept frequency response limiting it from 15Hz to 200 Hz, setting all other speakers to small and jacking the crossover to 200 Hz on the AVR while doing the sub measurements and looking for the 1st good location for sub1 in the room. A good place gives you the widest and flattest frequency response BTW.

Here's my work on my greatroom..start at post 5

3db's Great Room Setup

This system has seen alot of changes from a speaker perspective as I continue to collect PSB's Century line up of speakers. Here's a history starting with PSB 400. The surrounds are in-ceiling..not as good as having them closer to the listener position but its better than not having them at...

www.audiosciencereview.com

My apologies for getting distracted.

 

[xovaan]

Ahhh I think I figured it out!! I'm an idiot and was putting my ms delays in the SHD's matrix's gain box watching 3dbin's video made me double check so thank you hahaha

Here's what they look like when combined, 60hz 24dB/octave LR for speakers and 80hz for subs:

Looks pretty good, yeah? I'd imagine now it's Dirac time?

When you say "3. I failed to mention earlier that you want the subs to receive summed input from both the L/R channels using the channel matrix in the MiniDSP. So Sub 1 and Sub 2 should receive both L/R signals." -- when I do this, REW cuts off because the combined SPL is >90dB. I'm assuming you were referring to setting REW test's Output as L+R when Measurement testing and not as the final output in the MiniDSP SHD matrix for Dirac?

 

[Chromatischism]

You should set the crossover frequency to the same value on both speakers and subs

 

[xovaan]

You should set the crossover frequency to the same value on both speakers and subs

Here're the results with 80/80 LR 24dB/octave

edit: even if I'm still doing everything a bit wonkily, I gotta say that running these results through Dirac makes everything sound absolutely exceptional, especially compared to my first unguided attempts-- I really really appreciate you three helping me limp through this haha.

With the above, Dirac sets my left channel volume slightly lower (~1dB) and the delay on the right channel to 0.22ms

 

[3dbinCanada]

View attachment 159015
Here're the results with 80/80 LR 24dB/octave

edit: even if I'm still doing everything a bit wonkily, I gotta say that running these results through Dirac makes everything sound absolutely exceptional, especially compared to my first unguided attempts-- I really really appreciate you three helping me limp through this haha.

With the above, Dirac sets my left channel volume slightly lower (~1dB) and the delay on the right channel to 0.22ms

Looking very good.

 

[xovaan]

Looking very good.

yay! Just as a sanity check before doing these tests, I wiped the SHD. The only thing in those charts are the ms delays and crossovers listed above added to the respective outputs. I am totally surprised at how incredibly flat the measurements are!

Edit: is there a general guide this chain of posts could be summed up into or repurposed for existing ones? I think there’s a lot of great info from you all’s posts that cleared up a lot of what I couldn’t find through hours of independent reading and research, most from decentralized and even at times disparate sources

 

[CumSum]

You should set the crossover frequency to the same value on both speakers and subs

I disagree, subwoofer should always be crossed over at 80Hz or above, and if you have bass capable speakers than you can certainly cross them lower. Having 4 sources of bass is better than 2 in the scheme of fighting nulls.

First I add delay to the right speaker and sub to match the front left speaker, then I add .9088ms to all other channels to zero and match everything to the sub. Does my math check out?

Watching video now and checking the sweep! When I try to run all speakers and subs together, I get a really high dB-- should I lower the volume for the test? Here's my subwoofers combined after alignment:

The math will add up when you get acoustic timing results close to zero when testing the right speaker, and both subs against the left speaker with the delays you've come up with. Your channel matrix looks good.

Your REW charts have too much span in terms of SPL (-20dB to 160dB), which makes your results look flatter than what they really are. Change the span so you only have a 50dB spread, example from 55-105dB. Even with your "flat" looking graph, I can see it isn't flat and I see suck out from 35Hz to 65Hz that looks substantial. Resize the graph to get a better idea of how many dB you are down in that region.

Another good way to tell if the time alignment is having a positive effect is to compare it to having no delays at all. The one with correctly set delays will have more output across most of the spectrum. Here is an example of such a difference I measured, green is with a time alignment, red without. It makes a huge difference in my case since the delay's go up to 10+ms. Your delays are pretty small so the difference probably won't be as dramatic, but there should be an improvement.



Comparison of the group delay



I want to create a guide, and have already to some degree, but I'll eventually get around to finishing it so I can just drop it every time someone has a question about integrating subs.

 

[xovaan]

I disagree, subwoofer should always be crossed over at 80Hz or above, and if you have bass capable speakers than you can certainly cross them lower. Having 4 sources of bass is better than 2 in the scheme of fighting nulls.

The math will add up when you get acoustic timing results close to zero when testing the right speaker, and both subs against the left speaker with the delays you've come up with. Your channel matrix looks good.

Your REW charts have too much span in terms of SPL (-20dB to 160dB), which makes your results look flatter than what they really are. Change the span so you only have a 50dB spread, example from 55-105dB. Even with your "flat" looking graph, I can see it isn't flat and I see suck out from 35Hz to 65Hz that looks substantial. Resize the graph to get a better idea of how many dB you are down in that region.

Another good way to tell if the time alignment is having a positive effect is to compare it to having no delays at all. The one with correctly set delays will have more output across most of the spectrum. Here is an example of such a difference I measured, green is with a time alignment, red without. It makes a huge difference in my case since the delay's go up to 10+ms. Your delays are pretty small so the difference probably won't be as dramatic, but there should be an improvement.

View attachment 159072

Comparison of the group delay

View attachment 159074

I want to create a guide, and have already to some degree, but I'll eventually get around to finishing it so I can just drop it every time someone has a question about integrating subs.

Ahh, I see. Like this?

I'll note that for "All Channels" I had to turn down the REW volume from -15 to -20 since it was going well into the high 90s for dB with both subs and speakers firing together

 

[CumSum]

Yeah turn down the volume when doing them all together, it will be a lot louder.

You've got a considerable suck out, do all your individual measurements show this null? If so then I would mess with sub placement, if not then you have a phase issue.

 

[xovaan]

Yeah turn down the volume when doing them all together, it will be a lot louder.

You've got a considerable suck out, do all your individual measurements show this null? If so then I would mess with sub placement, if not then you have a phase issue.

Yeah I have a really bad null, likely because of my 3 foot wide cloth ottoman I gridded the entire front of my living room with purple masking tape and ran sweeps in every position including including moving the mic forward and backward and the positions they're in right now are the only places I've found to help. The further out from the wall, the worse the 40-60hz null becomes. I tried stacking them but that didn't really help because the right sub (green) is compensating for the left sub's (red line) waterfalled null between 65 and 90hz

Here're my subs again individually and combined (with timing alignment delays applied) (-15dB in REW for all three)

The left speaker null exists even when the sub and speaker crossovers are at 50hz/80hz and goes away when the crossovers are 80/80. However that dip makes me think that moving the speaker is ideal, yeah?

 

[xovaan]

Some more graphs:

All tests, crossed over 50hz /24dB for speakers, 100hz/24dB for subwoofers (time aligned):

Subwoofers without speakers:

Speakers without subwoofers (something crazy is going on with the left speaker at 65hz):

Left and Right Channels:

All firing together:

Not entirely sure what I'm looking at here though? That dip on my left speaker is absurd though, wondering where I should move it