Subwoofer 2.2 vs. 2.1 REW measurements

It's been cold here in Texas so I thought I'd do some subwoofer comparison using Audiolense and measuring the final result in REW. This is a real challenge to do because Audiolense generates an FIR filter in the form of an impulse response which is contained in a file Jriver can read in its convolution engine. The problem with measuring is that REW can't measure straight through Jriver using the ASIO driver or the Jriver WDM. I've talked to Mitch Barnett and he's helped show me how to use the uber powerful Lynx Hilo to accomplish this task.

I set up the Jriver ASIO line in to 48khz, 2CH, channel offset 6. I then set the Hilo to patch channels 7/8 outputs to 7/8 inputs. I then setup REW to output to Lynx Hilo channels 7 or 8 (left and right channels) and selected Lynx hilo input 1 to capture my earthworks m23 with calibration file. Setup looks like this:


I have a 24'x14.5'x9'3" room on the second floor of my home. The room is mostly treated with mostly RPG gear carefully selected to reduce LF ringing and diffuse HF excess decay to my liking. I listen to a WIDE variety of music.

Audiolense XO is a powerful tool which can easily integrate subwoofers with main R/L speakers as well as offer target based correction. My focus here will be the subwoofer integration. IMO, subwoofer integration is all about obtaining flat bass. That means it's important that the main speaker and the sub measure as flat as possible, especially at the crossover. Of course, phase and frequency response are closely related to one another. In the case of frequencies below 200hz, it's really tough to get a flat phase response. The best most folks could hope for is plus or minus 30 degrees. Really, it's more about avoiding 180 degree phase shifts. Such shift typically signify a poor crossover integration or an SBIR (boundary interference). Either way, it will affect the frequency response in a negative way.

Audiolense can adjust in the time domain, phase response and frequency response. Time domain is important with sub integration because subs should NOT normally go right next to a loudspeaker. My system is a stereo system with two subwoofers (for now). I have tried placing the subs in a number of configurations and tried a number of different crossover slopes and crossover frequencies. Audiolense's Bernt built his own FIR crossover. To me, it looks similar to an NT crossover, except it doesn't exhibit the V shaped end. The slope is very shallow up to the crossover point and then is increasingly sharp. The crossover exhibits a perfect impulse response. I'm not worried about pre-ringing.


I've narrowed the setups down to the following based on best measured and subjective results:

1. 2.2 with the subs in a stereo o configuration and a 2 octave crossover at 60 hz.
2. 2.2 with the subs in a mono configuration and a 2 octave crossover at 60hz
3. 2.1 with the sub in a mono configuration and a 2 octave crossover at 60hz
4. 2. 1 with the sub in a mono configuration and a 2 octave crossover at 80hz.

I'll post my REW shots next. All shots are 1/12 octave smoothing from a single measurement R/L at seated position.

Setup #1:


Setup #2:


Setups 3 and 4 to follow.

Setup 3:


Setup 4:

Conclusions

Subjectively, setup #4 sounds best. The bass is fullest and and most complete sounding. The bass is very articulate and mega dynamic! Final conclusions:

1. Sometimes one sub is better than two.
2. An 80hz crossover in a mono setup is NOT too high. I thought that the 80hz crossover would mean that I would lose important R/L stereo imaging information with a mono sub. This is simply not true at all. The sub is so well integrated, it would be impossible to pick out the sub in the room and the stereo image is unaffected by the higher crossover. Why did I pick such a high crossover? You can see in all but the 80hz crossover shots, there's a dip centered just between 75hz and 80hz. That's a partial sidewall SBIR dip from my right speaker. By increasing the crossover, I could eliminate the sidewall reflection by off loading that bass onto the sub which is located on the frontwall about 10 feet behind the right speaker.

I forgot to mention how I can switch easily between each of these setups. I learned a trick from the great Jriverologist known as Mojave.

I set up zones with all settings the same. The only difference is the Audiolense correction folder to which Jriver convolution engine points. So, each setup is its own zone and I can switch between each one from my iPad. Pretty cool. Thanks Mojave!

Thanks! That's a great tip. Jriver outputs to channels USB 1,2 and 4 in setup #4 above and I use hilo USB input 1 for my mic. So, if I select channel 1+2 in the Hilo ouptut and patch that channel to Hilo USB channel 3 input, I would then select channel 3 input as time reference in REW?

3ll3d00d said:

FWIW one little publicised feature in REW is that there is no requirement for an output signal when using a loopback timing reference, you just need to get the signal to the specified input channel. This is where a mixer app & a loopback capability in your ASIO driver comes in v handy as it means you copy the output from jriver to an unused channel then loop it back to some unused input channel and feed that to REW. This then means you can make measurements with accurate time of flight data after it has passed through whatever you do in jriver for subsequent use in trace arithmetic. Another use, with a different loopback configuration, is measuring the time of flight through jriver itself.

Click to expand...

Okay. I've done more measuring with a 80hz crossover in the 2.2 mono sub setup which is setup number 2 above. The decay looks a lot better in this setup and it sounds much better too; tighter and more accurate sounding to me.

Brucemck2 said:

Great improvement. I too prefer mono subs for my two channel setup. In my case it's five subs crossed at 50hz, and there's no loss in either imaging or directionality.

Is there any smoothing in the frequency response comparisons further up in this thread? (The REW waterfalls appear to have more variability than the REW frequency response plots.)

Click to expand...

My understanding of how FFT works is that the more time domain information the graph shows, the less accurate the frequency domain will be and vice versa. Maybe that's not the reason, but it sounds right to me. I only used 1/12 smoothing. I will post more later.

I have a little more tweeking to do in Audiolense. I may also go back to Acourate and do a head to head comparison in REW between the two.

This screenshot is partially cutoff above 13.5khz but it gives a good picture of what things are like with the 2.2 mono setup with an 80hz XO. My settings in audiolense xo are default TTD I tried a more aggressive custom setting and got even flatter bass but the soundstage suffered so I wont use it. I plan on setting up Acourate again with the same setup and compare the two in REW. I'll post here later.

If you have a question, I'll try to answer it.

The thread is about me trying different sub setups and measuring the results. In my view, there are 3 or 4 measurementd which are helpful to finding the best results. SPL and phase are key to sub integration. I also think waterfall helps show decay times. I also like the step measurement which show how my system works at the seated position in the time domain.

CGabriel said:

Dallas, I know how much effort it takes to post measurements and graphs as you are doing. It is very much appreciated. However, I don't think 90% of the people here have any idea what you are talking about or what the graphs that you are posting mean. I'm not being critical of the effort - I just think it would be helpful if you explained more about what each graph means - what to look for in each graph and what it correlates to.

Even for those of us that might know what these mean it is helpful to get a bit more context.

Thank you.

Click to expand...

3ll3d00d said:

can you reduce the y axis scale to be consistent with the earlier plots? 170dB is rather wide, makes it hard to see what is going on

Click to expand...

I can't post better screenshots now since I'm not home. Overall, you can see the SPL is +or- 5db with 1/12 octave smoothing. The bass is pretty tight and excellent. . I'm just curious about how Acourate would compare.

Step response is a handy way to analyze a system's time domain response. Here's JA on the topic:
http://www.stereophile.com/content/measuring-loudspeakers-part-two-page-2

The step in 2.2 mono subs crossed at 80hz is excellent.